diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,11221 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.0631463871386596,
+  "eval_steps": 500,
+  "global_step": 8000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "grad_norm": 20.459917068481445,
+      "learning_rate": 5e-06,
+      "loss": 13.121,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 17.737674713134766,
+      "learning_rate": 1e-05,
+      "loss": 12.9108,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 18.073278427124023,
+      "learning_rate": 1.5e-05,
+      "loss": 12.361,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 14.38332748413086,
+      "learning_rate": 2e-05,
+      "loss": 11.3831,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 12.622296333312988,
+      "learning_rate": 2.5e-05,
+      "loss": 10.2597,
+      "step": 25
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 7.543446063995361,
+      "learning_rate": 3e-05,
+      "loss": 9.5043,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 5.848616600036621,
+      "learning_rate": 3.5e-05,
+      "loss": 8.9002,
+      "step": 35
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 4.801801681518555,
+      "learning_rate": 4e-05,
+      "loss": 8.4651,
+      "step": 40
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 3.4637742042541504,
+      "learning_rate": 4.5e-05,
+      "loss": 8.2318,
+      "step": 45
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.9955459833145142,
+      "learning_rate": 5e-05,
+      "loss": 8.0392,
+      "step": 50
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.532737135887146,
+      "learning_rate": 5.500000000000001e-05,
+      "loss": 7.952,
+      "step": 55
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.7763688564300537,
+      "learning_rate": 6e-05,
+      "loss": 7.9232,
+      "step": 60
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.7964303493499756,
+      "learning_rate": 6.500000000000001e-05,
+      "loss": 7.7864,
+      "step": 65
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.391261100769043,
+      "learning_rate": 7e-05,
+      "loss": 7.7277,
+      "step": 70
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3876303434371948,
+      "learning_rate": 7.500000000000001e-05,
+      "loss": 7.633,
+      "step": 75
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.2257343530654907,
+      "learning_rate": 8e-05,
+      "loss": 7.4955,
+      "step": 80
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 4.47312068939209,
+      "learning_rate": 8.5e-05,
+      "loss": 7.387,
+      "step": 85
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 3.4937548637390137,
+      "learning_rate": 9e-05,
+      "loss": 7.2341,
+      "step": 90
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 2.564572334289551,
+      "learning_rate": 9.5e-05,
+      "loss": 7.1328,
+      "step": 95
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4583632946014404,
+      "learning_rate": 0.0001,
+      "loss": 7.0787,
+      "step": 100
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 2.585265636444092,
+      "learning_rate": 9.999999961506464e-05,
+      "loss": 6.7964,
+      "step": 105
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.89126455783844,
+      "learning_rate": 9.999999846025851e-05,
+      "loss": 6.7193,
+      "step": 110
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.5131787061691284,
+      "learning_rate": 9.999999653558166e-05,
+      "loss": 6.5699,
+      "step": 115
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.5249619483947754,
+      "learning_rate": 9.999999384103412e-05,
+      "loss": 6.3972,
+      "step": 120
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 2.061691999435425,
+      "learning_rate": 9.999999037661592e-05,
+      "loss": 6.3146,
+      "step": 125
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.7977921962738037,
+      "learning_rate": 9.999998614232713e-05,
+      "loss": 6.207,
+      "step": 130
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.6410233974456787,
+      "learning_rate": 9.99999811381678e-05,
+      "loss": 6.1829,
+      "step": 135
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.398443341255188,
+      "learning_rate": 9.9999975364138e-05,
+      "loss": 6.0927,
+      "step": 140
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 3.0345592498779297,
+      "learning_rate": 9.999996882023784e-05,
+      "loss": 5.9083,
+      "step": 145
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 2.1451151371002197,
+      "learning_rate": 9.99999615064674e-05,
+      "loss": 5.8396,
+      "step": 150
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 2.4007577896118164,
+      "learning_rate": 9.99999534228268e-05,
+      "loss": 5.743,
+      "step": 155
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.9859821796417236,
+      "learning_rate": 9.999994456931619e-05,
+      "loss": 5.6725,
+      "step": 160
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.5622456073760986,
+      "learning_rate": 9.999993494593566e-05,
+      "loss": 5.5976,
+      "step": 165
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.8456835746765137,
+      "learning_rate": 9.999992455268539e-05,
+      "loss": 5.562,
+      "step": 170
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 2.257460594177246,
+      "learning_rate": 9.999991338956554e-05,
+      "loss": 5.5969,
+      "step": 175
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.6177560091018677,
+      "learning_rate": 9.999990145657626e-05,
+      "loss": 5.3943,
+      "step": 180
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.6249388456344604,
+      "learning_rate": 9.999988875371776e-05,
+      "loss": 5.3738,
+      "step": 185
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.8749276399612427,
+      "learning_rate": 9.999987528099021e-05,
+      "loss": 5.3125,
+      "step": 190
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 3.24286150932312,
+      "learning_rate": 9.999986103839384e-05,
+      "loss": 5.2791,
+      "step": 195
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 2.5453410148620605,
+      "learning_rate": 9.999984602592884e-05,
+      "loss": 5.2254,
+      "step": 200
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 2.416992664337158,
+      "learning_rate": 9.999983024359549e-05,
+      "loss": 5.1412,
+      "step": 205
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 2.038947343826294,
+      "learning_rate": 9.9999813691394e-05,
+      "loss": 5.1899,
+      "step": 210
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 2.023149013519287,
+      "learning_rate": 9.999979636932461e-05,
+      "loss": 5.1525,
+      "step": 215
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.7797149419784546,
+      "learning_rate": 9.999977827738762e-05,
+      "loss": 4.9574,
+      "step": 220
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.973610281944275,
+      "learning_rate": 9.999975941558328e-05,
+      "loss": 5.0014,
+      "step": 225
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 2.884382963180542,
+      "learning_rate": 9.999973978391191e-05,
+      "loss": 4.9444,
+      "step": 230
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 2.1445908546447754,
+      "learning_rate": 9.99997193823738e-05,
+      "loss": 4.9078,
+      "step": 235
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.9023029804229736,
+      "learning_rate": 9.999969821096924e-05,
+      "loss": 4.9537,
+      "step": 240
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.8744838237762451,
+      "learning_rate": 9.99996762696986e-05,
+      "loss": 4.6988,
+      "step": 245
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 2.330733299255371,
+      "learning_rate": 9.999965355856218e-05,
+      "loss": 4.7534,
+      "step": 250
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.7803536653518677,
+      "learning_rate": 9.999963007756035e-05,
+      "loss": 4.752,
+      "step": 255
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.9157418012619019,
+      "learning_rate": 9.999960582669346e-05,
+      "loss": 4.7243,
+      "step": 260
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 2.236593723297119,
+      "learning_rate": 9.99995808059619e-05,
+      "loss": 4.7349,
+      "step": 265
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.5074450969696045,
+      "learning_rate": 9.999955501536603e-05,
+      "loss": 4.6409,
+      "step": 270
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.7028778791427612,
+      "learning_rate": 9.999952845490626e-05,
+      "loss": 4.5337,
+      "step": 275
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.806962013244629,
+      "learning_rate": 9.999950112458302e-05,
+      "loss": 4.626,
+      "step": 280
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4318017959594727,
+      "learning_rate": 9.99994730243967e-05,
+      "loss": 4.4976,
+      "step": 285
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.7295138835906982,
+      "learning_rate": 9.999944415434774e-05,
+      "loss": 4.4854,
+      "step": 290
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.6514060497283936,
+      "learning_rate": 9.99994145144366e-05,
+      "loss": 4.409,
+      "step": 295
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.6269079446792603,
+      "learning_rate": 9.999938410466371e-05,
+      "loss": 4.4946,
+      "step": 300
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4645394086837769,
+      "learning_rate": 9.999935292502958e-05,
+      "loss": 4.4046,
+      "step": 305
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.393951416015625,
+      "learning_rate": 9.999932097553463e-05,
+      "loss": 4.4509,
+      "step": 310
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.8279653787612915,
+      "learning_rate": 9.999928825617941e-05,
+      "loss": 4.4594,
+      "step": 315
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.5632216930389404,
+      "learning_rate": 9.999925476696439e-05,
+      "loss": 4.3308,
+      "step": 320
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.6052900552749634,
+      "learning_rate": 9.99992205078901e-05,
+      "loss": 4.3206,
+      "step": 325
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.7239354848861694,
+      "learning_rate": 9.999918547895705e-05,
+      "loss": 4.3186,
+      "step": 330
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.5922297239303589,
+      "learning_rate": 9.999914968016579e-05,
+      "loss": 4.3092,
+      "step": 335
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.8730278015136719,
+      "learning_rate": 9.999911311151689e-05,
+      "loss": 4.2495,
+      "step": 340
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 2.0404999256134033,
+      "learning_rate": 9.999907577301089e-05,
+      "loss": 4.3106,
+      "step": 345
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.8017346858978271,
+      "learning_rate": 9.999903766464837e-05,
+      "loss": 4.1902,
+      "step": 350
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 2.2076609134674072,
+      "learning_rate": 9.99989987864299e-05,
+      "loss": 4.0246,
+      "step": 355
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4327454566955566,
+      "learning_rate": 9.999895913835613e-05,
+      "loss": 4.1105,
+      "step": 360
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.9013521671295166,
+      "learning_rate": 9.999891872042762e-05,
+      "loss": 4.261,
+      "step": 365
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.6055341958999634,
+      "learning_rate": 9.9998877532645e-05,
+      "loss": 4.0631,
+      "step": 370
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4047565460205078,
+      "learning_rate": 9.999883557500892e-05,
+      "loss": 4.1889,
+      "step": 375
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.7069621086120605,
+      "learning_rate": 9.999879284752003e-05,
+      "loss": 4.192,
+      "step": 380
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.7576422691345215,
+      "learning_rate": 9.999874935017897e-05,
+      "loss": 3.9992,
+      "step": 385
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.7146271467208862,
+      "learning_rate": 9.999870508298641e-05,
+      "loss": 4.0678,
+      "step": 390
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.5461151599884033,
+      "learning_rate": 9.999866004594303e-05,
+      "loss": 4.0353,
+      "step": 395
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.8388198614120483,
+      "learning_rate": 9.999861423904954e-05,
+      "loss": 4.0547,
+      "step": 400
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.7166547775268555,
+      "learning_rate": 9.999856766230665e-05,
+      "loss": 4.064,
+      "step": 405
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.6476255655288696,
+      "learning_rate": 9.999852031571506e-05,
+      "loss": 3.9976,
+      "step": 410
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.5955204963684082,
+      "learning_rate": 9.999847219927549e-05,
+      "loss": 3.9531,
+      "step": 415
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.253585934638977,
+      "learning_rate": 9.999842331298872e-05,
+      "loss": 4.0262,
+      "step": 420
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4218100309371948,
+      "learning_rate": 9.999837365685544e-05,
+      "loss": 3.9,
+      "step": 425
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3079349994659424,
+      "learning_rate": 9.999832323087649e-05,
+      "loss": 3.9858,
+      "step": 430
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3216156959533691,
+      "learning_rate": 9.999827203505259e-05,
+      "loss": 3.8721,
+      "step": 435
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.356337070465088,
+      "learning_rate": 9.999822006938455e-05,
+      "loss": 4.0161,
+      "step": 440
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.5120365619659424,
+      "learning_rate": 9.999816733387316e-05,
+      "loss": 3.9427,
+      "step": 445
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.2851651906967163,
+      "learning_rate": 9.999811382851924e-05,
+      "loss": 3.7621,
+      "step": 450
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.7548803091049194,
+      "learning_rate": 9.999805955332362e-05,
+      "loss": 3.894,
+      "step": 455
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.5187057256698608,
+      "learning_rate": 9.999800450828711e-05,
+      "loss": 3.8326,
+      "step": 460
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.6038079261779785,
+      "learning_rate": 9.99979486934106e-05,
+      "loss": 3.6777,
+      "step": 465
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.51413094997406,
+      "learning_rate": 9.99978921086949e-05,
+      "loss": 3.754,
+      "step": 470
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4592801332473755,
+      "learning_rate": 9.999783475414093e-05,
+      "loss": 3.7402,
+      "step": 475
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4571690559387207,
+      "learning_rate": 9.999777662974952e-05,
+      "loss": 3.8695,
+      "step": 480
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.2846940755844116,
+      "learning_rate": 9.999771773552162e-05,
+      "loss": 3.8384,
+      "step": 485
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4743348360061646,
+      "learning_rate": 9.999765807145811e-05,
+      "loss": 3.6958,
+      "step": 490
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4810373783111572,
+      "learning_rate": 9.99975976375599e-05,
+      "loss": 3.6948,
+      "step": 495
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3809741735458374,
+      "learning_rate": 9.999753643382793e-05,
+      "loss": 3.7539,
+      "step": 500
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.1016466617584229,
+      "learning_rate": 9.999747446026315e-05,
+      "loss": 3.6037,
+      "step": 505
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.2977675199508667,
+      "learning_rate": 9.999741171686649e-05,
+      "loss": 3.6134,
+      "step": 510
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4078669548034668,
+      "learning_rate": 9.999734820363895e-05,
+      "loss": 3.6477,
+      "step": 515
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4831340312957764,
+      "learning_rate": 9.99972839205815e-05,
+      "loss": 3.6135,
+      "step": 520
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3736069202423096,
+      "learning_rate": 9.999721886769511e-05,
+      "loss": 3.6206,
+      "step": 525
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.2076282501220703,
+      "learning_rate": 9.999715304498078e-05,
+      "loss": 3.6243,
+      "step": 530
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4046889543533325,
+      "learning_rate": 9.999708645243955e-05,
+      "loss": 3.6878,
+      "step": 535
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3431384563446045,
+      "learning_rate": 9.999701909007242e-05,
+      "loss": 3.5432,
+      "step": 540
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3973991870880127,
+      "learning_rate": 9.999695095788044e-05,
+      "loss": 3.5532,
+      "step": 545
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4100786447525024,
+      "learning_rate": 9.999688205586468e-05,
+      "loss": 3.5329,
+      "step": 550
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.2155297994613647,
+      "learning_rate": 9.999681238402615e-05,
+      "loss": 3.5439,
+      "step": 555
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.0509446859359741,
+      "learning_rate": 9.999674194236597e-05,
+      "loss": 3.4784,
+      "step": 560
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3428361415863037,
+      "learning_rate": 9.999667073088519e-05,
+      "loss": 3.5207,
+      "step": 565
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.205237865447998,
+      "learning_rate": 9.999659874958493e-05,
+      "loss": 3.5324,
+      "step": 570
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.1141941547393799,
+      "learning_rate": 9.99965259984663e-05,
+      "loss": 3.5843,
+      "step": 575
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3119527101516724,
+      "learning_rate": 9.999645247753039e-05,
+      "loss": 3.543,
+      "step": 580
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.213752031326294,
+      "learning_rate": 9.999637818677836e-05,
+      "loss": 3.4732,
+      "step": 585
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3168935775756836,
+      "learning_rate": 9.999630312621136e-05,
+      "loss": 3.5471,
+      "step": 590
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.1608147621154785,
+      "learning_rate": 9.999622729583051e-05,
+      "loss": 3.4863,
+      "step": 595
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3571707010269165,
+      "learning_rate": 9.999615069563701e-05,
+      "loss": 3.3082,
+      "step": 600
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.1521594524383545,
+      "learning_rate": 9.999607332563205e-05,
+      "loss": 3.4599,
+      "step": 605
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.2302151918411255,
+      "learning_rate": 9.999599518581676e-05,
+      "loss": 3.5109,
+      "step": 610
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.1598073244094849,
+      "learning_rate": 9.999591627619242e-05,
+      "loss": 3.4341,
+      "step": 615
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4125642776489258,
+      "learning_rate": 9.999583659676019e-05,
+      "loss": 3.4994,
+      "step": 620
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.2236433029174805,
+      "learning_rate": 9.999575614752133e-05,
+      "loss": 3.5163,
+      "step": 625
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.3545787334442139,
+      "learning_rate": 9.999567492847706e-05,
+      "loss": 3.4422,
+      "step": 630
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.294967532157898,
+      "learning_rate": 9.999559293962864e-05,
+      "loss": 3.3406,
+      "step": 635
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.1197681427001953,
+      "learning_rate": 9.99955101809773e-05,
+      "loss": 3.4242,
+      "step": 640
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.1507911682128906,
+      "learning_rate": 9.999542665252437e-05,
+      "loss": 3.455,
+      "step": 645
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.1903128623962402,
+      "learning_rate": 9.999534235427109e-05,
+      "loss": 3.387,
+      "step": 650
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.1529518365859985,
+      "learning_rate": 9.999525728621879e-05,
+      "loss": 3.4206,
+      "step": 655
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.0737675428390503,
+      "learning_rate": 9.999517144836876e-05,
+      "loss": 3.4391,
+      "step": 660
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.211099624633789,
+      "learning_rate": 9.999508484072233e-05,
+      "loss": 3.3694,
+      "step": 665
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.5973243713378906,
+      "learning_rate": 9.999499746328083e-05,
+      "loss": 3.3148,
+      "step": 670
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.1036471128463745,
+      "learning_rate": 9.99949093160456e-05,
+      "loss": 3.3176,
+      "step": 675
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.088607907295227,
+      "learning_rate": 9.9994820399018e-05,
+      "loss": 3.3824,
+      "step": 680
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.1028391122817993,
+      "learning_rate": 9.999473071219941e-05,
+      "loss": 3.3735,
+      "step": 685
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.009720802307129,
+      "learning_rate": 9.999464025559121e-05,
+      "loss": 3.3238,
+      "step": 690
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.1137475967407227,
+      "learning_rate": 9.999454902919478e-05,
+      "loss": 3.3928,
+      "step": 695
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.572962760925293,
+      "learning_rate": 9.999445703301154e-05,
+      "loss": 3.3264,
+      "step": 700
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.9704469442367554,
+      "learning_rate": 9.999436426704287e-05,
+      "loss": 3.3176,
+      "step": 705
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.010072946548462,
+      "learning_rate": 9.999427073129026e-05,
+      "loss": 3.2963,
+      "step": 710
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.112356424331665,
+      "learning_rate": 9.999417642575509e-05,
+      "loss": 3.5072,
+      "step": 715
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.9525521993637085,
+      "learning_rate": 9.999408135043886e-05,
+      "loss": 3.3461,
+      "step": 720
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.1183339357376099,
+      "learning_rate": 9.999398550534298e-05,
+      "loss": 3.2823,
+      "step": 725
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.9731379747390747,
+      "learning_rate": 9.999388889046899e-05,
+      "loss": 3.282,
+      "step": 730
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.1696686744689941,
+      "learning_rate": 9.999379150581833e-05,
+      "loss": 3.2523,
+      "step": 735
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.0432387590408325,
+      "learning_rate": 9.999369335139251e-05,
+      "loss": 3.2921,
+      "step": 740
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.0290107727050781,
+      "learning_rate": 9.999359442719306e-05,
+      "loss": 3.3492,
+      "step": 745
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.9444539546966553,
+      "learning_rate": 9.999349473322149e-05,
+      "loss": 3.3591,
+      "step": 750
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.2591805458068848,
+      "learning_rate": 9.999339426947932e-05,
+      "loss": 3.2956,
+      "step": 755
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.2058292627334595,
+      "learning_rate": 9.999329303596812e-05,
+      "loss": 3.2728,
+      "step": 760
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.13230562210083,
+      "learning_rate": 9.999319103268945e-05,
+      "loss": 3.3439,
+      "step": 765
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.064125895500183,
+      "learning_rate": 9.999308825964485e-05,
+      "loss": 3.2139,
+      "step": 770
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.025323510169983,
+      "learning_rate": 9.999298471683593e-05,
+      "loss": 3.2623,
+      "step": 775
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.9728690981864929,
+      "learning_rate": 9.999288040426429e-05,
+      "loss": 3.2871,
+      "step": 780
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.9104892611503601,
+      "learning_rate": 9.999277532193151e-05,
+      "loss": 3.2439,
+      "step": 785
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.9523934125900269,
+      "learning_rate": 9.999266946983923e-05,
+      "loss": 3.2422,
+      "step": 790
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.9808526635169983,
+      "learning_rate": 9.999256284798906e-05,
+      "loss": 3.2752,
+      "step": 795
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.9006328582763672,
+      "learning_rate": 9.999245545638267e-05,
+      "loss": 3.2036,
+      "step": 800
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.9106430411338806,
+      "learning_rate": 9.999234729502167e-05,
+      "loss": 3.1738,
+      "step": 805
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.8753567934036255,
+      "learning_rate": 9.999223836390777e-05,
+      "loss": 3.189,
+      "step": 810
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.9866809844970703,
+      "learning_rate": 9.999212866304262e-05,
+      "loss": 3.2819,
+      "step": 815
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.1108899116516113,
+      "learning_rate": 9.999201819242794e-05,
+      "loss": 3.2658,
+      "step": 820
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.8501389622688293,
+      "learning_rate": 9.99919069520654e-05,
+      "loss": 3.2668,
+      "step": 825
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.883297324180603,
+      "learning_rate": 9.99917949419567e-05,
+      "loss": 3.2688,
+      "step": 830
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.9315003156661987,
+      "learning_rate": 9.999168216210362e-05,
+      "loss": 3.367,
+      "step": 835
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.053619623184204,
+      "learning_rate": 9.999156861250784e-05,
+      "loss": 3.1704,
+      "step": 840
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.11088228225708,
+      "learning_rate": 9.999145429317114e-05,
+      "loss": 3.289,
+      "step": 845
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.0356810092926025,
+      "learning_rate": 9.999133920409527e-05,
+      "loss": 3.3043,
+      "step": 850
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.9832237958908081,
+      "learning_rate": 9.999122334528199e-05,
+      "loss": 3.2569,
+      "step": 855
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.1251569986343384,
+      "learning_rate": 9.999110671673313e-05,
+      "loss": 3.0631,
+      "step": 860
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.9449043869972229,
+      "learning_rate": 9.999098931845044e-05,
+      "loss": 3.1749,
+      "step": 865
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.9812780618667603,
+      "learning_rate": 9.999087115043573e-05,
+      "loss": 3.1419,
+      "step": 870
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.125901460647583,
+      "learning_rate": 9.999075221269083e-05,
+      "loss": 3.0845,
+      "step": 875
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.1885478496551514,
+      "learning_rate": 9.999063250521759e-05,
+      "loss": 3.1717,
+      "step": 880
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.9898957014083862,
+      "learning_rate": 9.999051202801781e-05,
+      "loss": 3.1878,
+      "step": 885
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.9520403146743774,
+      "learning_rate": 9.999039078109341e-05,
+      "loss": 3.126,
+      "step": 890
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.8973997831344604,
+      "learning_rate": 9.999026876444617e-05,
+      "loss": 3.2309,
+      "step": 895
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.9779958128929138,
+      "learning_rate": 9.999014597807805e-05,
+      "loss": 3.1418,
+      "step": 900
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.8986417651176453,
+      "learning_rate": 9.99900224219909e-05,
+      "loss": 3.1961,
+      "step": 905
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.003785252571106,
+      "learning_rate": 9.998989809618662e-05,
+      "loss": 3.1642,
+      "step": 910
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.9117045402526855,
+      "learning_rate": 9.998977300066715e-05,
+      "loss": 3.1258,
+      "step": 915
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.8941423892974854,
+      "learning_rate": 9.998964713543437e-05,
+      "loss": 3.0955,
+      "step": 920
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.8637370467185974,
+      "learning_rate": 9.998952050049026e-05,
+      "loss": 3.0858,
+      "step": 925
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.8547463417053223,
+      "learning_rate": 9.998939309583677e-05,
+      "loss": 3.1662,
+      "step": 930
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.8521625995635986,
+      "learning_rate": 9.998926492147585e-05,
+      "loss": 3.1578,
+      "step": 935
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.93804532289505,
+      "learning_rate": 9.998913597740945e-05,
+      "loss": 3.1249,
+      "step": 940
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.9055214524269104,
+      "learning_rate": 9.99890062636396e-05,
+      "loss": 3.1697,
+      "step": 945
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.9208787679672241,
+      "learning_rate": 9.998887578016827e-05,
+      "loss": 3.0546,
+      "step": 950
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.9371682405471802,
+      "learning_rate": 9.998874452699747e-05,
+      "loss": 3.153,
+      "step": 955
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.8045409917831421,
+      "learning_rate": 9.998861250412922e-05,
+      "loss": 3.1394,
+      "step": 960
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.8667465448379517,
+      "learning_rate": 9.998847971156556e-05,
+      "loss": 3.1607,
+      "step": 965
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.8654932975769043,
+      "learning_rate": 9.998834614930855e-05,
+      "loss": 3.0927,
+      "step": 970
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.8387807011604309,
+      "learning_rate": 9.99882118173602e-05,
+      "loss": 3.1345,
+      "step": 975
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.8060160279273987,
+      "learning_rate": 9.998807671572263e-05,
+      "loss": 3.0043,
+      "step": 980
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.9758902788162231,
+      "learning_rate": 9.998794084439789e-05,
+      "loss": 3.1297,
+      "step": 985
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.038002848625183,
+      "learning_rate": 9.998780420338809e-05,
+      "loss": 3.0867,
+      "step": 990
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.84876948595047,
+      "learning_rate": 9.998766679269531e-05,
+      "loss": 3.0957,
+      "step": 995
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.8563096523284912,
+      "learning_rate": 9.998752861232168e-05,
+      "loss": 2.9909,
+      "step": 1000
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.8337981104850769,
+      "learning_rate": 9.998738966226933e-05,
+      "loss": 3.1083,
+      "step": 1005
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.7832773923873901,
+      "learning_rate": 9.998724994254041e-05,
+      "loss": 3.123,
+      "step": 1010
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.8175668716430664,
+      "learning_rate": 9.998710945313705e-05,
+      "loss": 3.0328,
+      "step": 1015
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.8549673557281494,
+      "learning_rate": 9.998696819406141e-05,
+      "loss": 2.9414,
+      "step": 1020
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.8991570472717285,
+      "learning_rate": 9.99868261653157e-05,
+      "loss": 3.1504,
+      "step": 1025
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.9488316774368286,
+      "learning_rate": 9.998668336690206e-05,
+      "loss": 3.0843,
+      "step": 1030
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.8520652055740356,
+      "learning_rate": 9.998653979882271e-05,
+      "loss": 3.2325,
+      "step": 1035
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.8085082173347473,
+      "learning_rate": 9.998639546107989e-05,
+      "loss": 3.0475,
+      "step": 1040
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.013977289199829,
+      "learning_rate": 9.998625035367577e-05,
+      "loss": 3.1159,
+      "step": 1045
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.9785688519477844,
+      "learning_rate": 9.998610447661261e-05,
+      "loss": 3.0894,
+      "step": 1050
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.922626256942749,
+      "learning_rate": 9.998595782989266e-05,
+      "loss": 3.0214,
+      "step": 1055
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.874560534954071,
+      "learning_rate": 9.998581041351818e-05,
+      "loss": 3.143,
+      "step": 1060
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.9196802973747253,
+      "learning_rate": 9.998566222749142e-05,
+      "loss": 3.0572,
+      "step": 1065
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.8362348675727844,
+      "learning_rate": 9.998551327181468e-05,
+      "loss": 3.184,
+      "step": 1070
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.8632582426071167,
+      "learning_rate": 9.998536354649026e-05,
+      "loss": 3.0566,
+      "step": 1075
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.9518134593963623,
+      "learning_rate": 9.998521305152045e-05,
+      "loss": 3.0815,
+      "step": 1080
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.9680737257003784,
+      "learning_rate": 9.998506178690755e-05,
+      "loss": 2.9745,
+      "step": 1085
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.000510334968567,
+      "learning_rate": 9.998490975265393e-05,
+      "loss": 3.0234,
+      "step": 1090
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.9040667414665222,
+      "learning_rate": 9.99847569487619e-05,
+      "loss": 3.0094,
+      "step": 1095
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.848025918006897,
+      "learning_rate": 9.998460337523383e-05,
+      "loss": 2.9541,
+      "step": 1100
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.8614380359649658,
+      "learning_rate": 9.998444903207207e-05,
+      "loss": 3.0003,
+      "step": 1105
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.8908402323722839,
+      "learning_rate": 9.998429391927902e-05,
+      "loss": 3.0414,
+      "step": 1110
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.8353506326675415,
+      "learning_rate": 9.998413803685705e-05,
+      "loss": 3.0291,
+      "step": 1115
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.846549928188324,
+      "learning_rate": 9.998398138480855e-05,
+      "loss": 3.0639,
+      "step": 1120
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.8831397891044617,
+      "learning_rate": 9.998382396313597e-05,
+      "loss": 2.9562,
+      "step": 1125
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.7780085802078247,
+      "learning_rate": 9.998366577184168e-05,
+      "loss": 3.0179,
+      "step": 1130
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.7761319279670715,
+      "learning_rate": 9.998350681092817e-05,
+      "loss": 3.0096,
+      "step": 1135
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.8256552815437317,
+      "learning_rate": 9.998334708039785e-05,
+      "loss": 2.8595,
+      "step": 1140
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.8092740178108215,
+      "learning_rate": 9.998318658025318e-05,
+      "loss": 2.9698,
+      "step": 1145
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.8753529191017151,
+      "learning_rate": 9.998302531049667e-05,
+      "loss": 3.022,
+      "step": 1150
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.8494768142700195,
+      "learning_rate": 9.998286327113076e-05,
+      "loss": 3.0115,
+      "step": 1155
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.7127054929733276,
+      "learning_rate": 9.998270046215797e-05,
+      "loss": 3.0155,
+      "step": 1160
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.757195234298706,
+      "learning_rate": 9.998253688358078e-05,
+      "loss": 2.9962,
+      "step": 1165
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.7134938836097717,
+      "learning_rate": 9.998237253540174e-05,
+      "loss": 3.0609,
+      "step": 1170
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.8338983058929443,
+      "learning_rate": 9.998220741762338e-05,
+      "loss": 3.0285,
+      "step": 1175
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.731052041053772,
+      "learning_rate": 9.998204153024821e-05,
+      "loss": 2.9313,
+      "step": 1180
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.758245587348938,
+      "learning_rate": 9.998187487327882e-05,
+      "loss": 2.9785,
+      "step": 1185
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.7815524339675903,
+      "learning_rate": 9.998170744671775e-05,
+      "loss": 3.0495,
+      "step": 1190
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.8022199273109436,
+      "learning_rate": 9.998153925056758e-05,
+      "loss": 2.8873,
+      "step": 1195
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.8022090196609497,
+      "learning_rate": 9.998137028483092e-05,
+      "loss": 2.9527,
+      "step": 1200
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.8440799713134766,
+      "learning_rate": 9.998120054951035e-05,
+      "loss": 2.9493,
+      "step": 1205
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.8050652742385864,
+      "learning_rate": 9.99810300446085e-05,
+      "loss": 3.038,
+      "step": 1210
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.7151529788970947,
+      "learning_rate": 9.9980858770128e-05,
+      "loss": 3.0308,
+      "step": 1215
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.7793344259262085,
+      "learning_rate": 9.998068672607146e-05,
+      "loss": 2.939,
+      "step": 1220
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.6985414028167725,
+      "learning_rate": 9.998051391244155e-05,
+      "loss": 2.96,
+      "step": 1225
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.7378638982772827,
+      "learning_rate": 9.998034032924092e-05,
+      "loss": 2.9482,
+      "step": 1230
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.8037704229354858,
+      "learning_rate": 9.998016597647225e-05,
+      "loss": 3.0277,
+      "step": 1235
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.742202877998352,
+      "learning_rate": 9.997999085413823e-05,
+      "loss": 2.9306,
+      "step": 1240
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.7312039136886597,
+      "learning_rate": 9.997981496224153e-05,
+      "loss": 2.9917,
+      "step": 1245
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.6443310976028442,
+      "learning_rate": 9.99796383007849e-05,
+      "loss": 2.9997,
+      "step": 1250
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.6594488620758057,
+      "learning_rate": 9.997946086977103e-05,
+      "loss": 3.0237,
+      "step": 1255
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.7630845904350281,
+      "learning_rate": 9.997928266920266e-05,
+      "loss": 2.9778,
+      "step": 1260
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.6756492257118225,
+      "learning_rate": 9.997910369908254e-05,
+      "loss": 2.9365,
+      "step": 1265
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.7698631286621094,
+      "learning_rate": 9.99789239594134e-05,
+      "loss": 2.9429,
+      "step": 1270
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.7417536377906799,
+      "learning_rate": 9.997874345019805e-05,
+      "loss": 3.0321,
+      "step": 1275
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.7356902956962585,
+      "learning_rate": 9.997856217143921e-05,
+      "loss": 2.9217,
+      "step": 1280
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.7902883291244507,
+      "learning_rate": 9.997838012313974e-05,
+      "loss": 2.9362,
+      "step": 1285
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.7465853691101074,
+      "learning_rate": 9.99781973053024e-05,
+      "loss": 2.9676,
+      "step": 1290
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.7494348883628845,
+      "learning_rate": 9.997801371793001e-05,
+      "loss": 2.9944,
+      "step": 1295
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.8696385025978088,
+      "learning_rate": 9.997782936102542e-05,
+      "loss": 2.9679,
+      "step": 1300
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.8700966835021973,
+      "learning_rate": 9.997764423459144e-05,
+      "loss": 2.931,
+      "step": 1305
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.7460966110229492,
+      "learning_rate": 9.997745833863092e-05,
+      "loss": 3.0215,
+      "step": 1310
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.7626380920410156,
+      "learning_rate": 9.997727167314676e-05,
+      "loss": 2.9712,
+      "step": 1315
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.793410062789917,
+      "learning_rate": 9.997708423814178e-05,
+      "loss": 2.9337,
+      "step": 1320
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.778752863407135,
+      "learning_rate": 9.99768960336189e-05,
+      "loss": 2.9291,
+      "step": 1325
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.6851357221603394,
+      "learning_rate": 9.997670705958102e-05,
+      "loss": 2.8944,
+      "step": 1330
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.7874002456665039,
+      "learning_rate": 9.997651731603104e-05,
+      "loss": 2.9601,
+      "step": 1335
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.6728459000587463,
+      "learning_rate": 9.997632680297188e-05,
+      "loss": 2.928,
+      "step": 1340
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.7089241147041321,
+      "learning_rate": 9.997613552040648e-05,
+      "loss": 2.8185,
+      "step": 1345
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.7464758157730103,
+      "learning_rate": 9.997594346833779e-05,
+      "loss": 2.8769,
+      "step": 1350
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.7073016166687012,
+      "learning_rate": 9.997575064676875e-05,
+      "loss": 2.8502,
+      "step": 1355
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.6740908622741699,
+      "learning_rate": 9.997555705570232e-05,
+      "loss": 2.9709,
+      "step": 1360
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.7342933416366577,
+      "learning_rate": 9.997536269514152e-05,
+      "loss": 2.8916,
+      "step": 1365
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.7282151579856873,
+      "learning_rate": 9.99751675650893e-05,
+      "loss": 2.927,
+      "step": 1370
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.7149432301521301,
+      "learning_rate": 9.99749716655487e-05,
+      "loss": 3.022,
+      "step": 1375
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.6434655785560608,
+      "learning_rate": 9.997477499652273e-05,
+      "loss": 2.812,
+      "step": 1380
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.6700690388679504,
+      "learning_rate": 9.99745775580144e-05,
+      "loss": 2.8894,
+      "step": 1385
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.712579607963562,
+      "learning_rate": 9.997437935002676e-05,
+      "loss": 2.9381,
+      "step": 1390
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.6926534175872803,
+      "learning_rate": 9.997418037256285e-05,
+      "loss": 2.9599,
+      "step": 1395
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.7290432453155518,
+      "learning_rate": 9.997398062562575e-05,
+      "loss": 2.981,
+      "step": 1400
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.660201370716095,
+      "learning_rate": 9.997378010921856e-05,
+      "loss": 2.9423,
+      "step": 1405
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.6473440527915955,
+      "learning_rate": 9.99735788233443e-05,
+      "loss": 2.9415,
+      "step": 1410
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.776526153087616,
+      "learning_rate": 9.997337676800613e-05,
+      "loss": 2.9527,
+      "step": 1415
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.7468796968460083,
+      "learning_rate": 9.997317394320713e-05,
+      "loss": 2.8768,
+      "step": 1420
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.6884674429893494,
+      "learning_rate": 9.997297034895044e-05,
+      "loss": 2.8269,
+      "step": 1425
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.6779197454452515,
+      "learning_rate": 9.99727659852392e-05,
+      "loss": 2.9236,
+      "step": 1430
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.7137473821640015,
+      "learning_rate": 9.997256085207652e-05,
+      "loss": 2.8115,
+      "step": 1435
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.6741832494735718,
+      "learning_rate": 9.997235494946557e-05,
+      "loss": 2.8325,
+      "step": 1440
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.6962840557098389,
+      "learning_rate": 9.997214827740955e-05,
+      "loss": 2.9176,
+      "step": 1445
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.6266164183616638,
+      "learning_rate": 9.997194083591164e-05,
+      "loss": 2.8571,
+      "step": 1450
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.7190635204315186,
+      "learning_rate": 9.9971732624975e-05,
+      "loss": 2.8002,
+      "step": 1455
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.7160282731056213,
+      "learning_rate": 9.997152364460287e-05,
+      "loss": 2.8548,
+      "step": 1460
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.7277812957763672,
+      "learning_rate": 9.997131389479844e-05,
+      "loss": 2.8884,
+      "step": 1465
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.7035467624664307,
+      "learning_rate": 9.997110337556494e-05,
+      "loss": 2.9061,
+      "step": 1470
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.6659943461418152,
+      "learning_rate": 9.997089208690564e-05,
+      "loss": 2.8888,
+      "step": 1475
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.6794031262397766,
+      "learning_rate": 9.997068002882377e-05,
+      "loss": 2.97,
+      "step": 1480
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.6852645874023438,
+      "learning_rate": 9.997046720132261e-05,
+      "loss": 2.9024,
+      "step": 1485
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.6788705587387085,
+      "learning_rate": 9.997025360440541e-05,
+      "loss": 2.9507,
+      "step": 1490
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.7534631490707397,
+      "learning_rate": 9.997003923807551e-05,
+      "loss": 2.8754,
+      "step": 1495
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.7379429936408997,
+      "learning_rate": 9.996982410233615e-05,
+      "loss": 2.8811,
+      "step": 1500
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.723779022693634,
+      "learning_rate": 9.996960819719068e-05,
+      "loss": 2.8735,
+      "step": 1505
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.6529448628425598,
+      "learning_rate": 9.996939152264242e-05,
+      "loss": 3.0002,
+      "step": 1510
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.6638311147689819,
+      "learning_rate": 9.99691740786947e-05,
+      "loss": 2.935,
+      "step": 1515
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.7000625133514404,
+      "learning_rate": 9.996895586535086e-05,
+      "loss": 2.8496,
+      "step": 1520
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.7199581861495972,
+      "learning_rate": 9.996873688261428e-05,
+      "loss": 2.7621,
+      "step": 1525
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.7979674339294434,
+      "learning_rate": 9.996851713048831e-05,
+      "loss": 2.9608,
+      "step": 1530
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.7413047552108765,
+      "learning_rate": 9.996829660897637e-05,
+      "loss": 2.8065,
+      "step": 1535
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.6603520512580872,
+      "learning_rate": 9.996807531808181e-05,
+      "loss": 2.9678,
+      "step": 1540
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.755696177482605,
+      "learning_rate": 9.996785325780806e-05,
+      "loss": 2.8734,
+      "step": 1545
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.6926345825195312,
+      "learning_rate": 9.996763042815853e-05,
+      "loss": 2.9031,
+      "step": 1550
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.6777913570404053,
+      "learning_rate": 9.996740682913668e-05,
+      "loss": 2.8665,
+      "step": 1555
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.67631596326828,
+      "learning_rate": 9.996718246074591e-05,
+      "loss": 2.873,
+      "step": 1560
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.6446473002433777,
+      "learning_rate": 9.99669573229897e-05,
+      "loss": 2.8566,
+      "step": 1565
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.6347461342811584,
+      "learning_rate": 9.996673141587151e-05,
+      "loss": 2.9322,
+      "step": 1570
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.7319153547286987,
+      "learning_rate": 9.996650473939482e-05,
+      "loss": 2.796,
+      "step": 1575
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.6857085824012756,
+      "learning_rate": 9.996627729356314e-05,
+      "loss": 2.7776,
+      "step": 1580
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.6385776400566101,
+      "learning_rate": 9.996604907837993e-05,
+      "loss": 2.7546,
+      "step": 1585
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.6503087282180786,
+      "learning_rate": 9.996582009384874e-05,
+      "loss": 2.8128,
+      "step": 1590
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.6655900478363037,
+      "learning_rate": 9.996559033997307e-05,
+      "loss": 2.878,
+      "step": 1595
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.7392303347587585,
+      "learning_rate": 9.996535981675648e-05,
+      "loss": 2.8876,
+      "step": 1600
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.6283818483352661,
+      "learning_rate": 9.996512852420251e-05,
+      "loss": 2.8986,
+      "step": 1605
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.652241051197052,
+      "learning_rate": 9.996489646231471e-05,
+      "loss": 2.7872,
+      "step": 1610
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.6998395919799805,
+      "learning_rate": 9.996466363109667e-05,
+      "loss": 2.9283,
+      "step": 1615
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.7832385301589966,
+      "learning_rate": 9.996443003055196e-05,
+      "loss": 2.8345,
+      "step": 1620
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.7121884822845459,
+      "learning_rate": 9.99641956606842e-05,
+      "loss": 2.8613,
+      "step": 1625
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.6953356266021729,
+      "learning_rate": 9.996396052149698e-05,
+      "loss": 2.8545,
+      "step": 1630
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.6852560639381409,
+      "learning_rate": 9.996372461299392e-05,
+      "loss": 2.9042,
+      "step": 1635
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.632819414138794,
+      "learning_rate": 9.996348793517866e-05,
+      "loss": 2.8664,
+      "step": 1640
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.6674362421035767,
+      "learning_rate": 9.996325048805484e-05,
+      "loss": 2.7924,
+      "step": 1645
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.6544278264045715,
+      "learning_rate": 9.996301227162612e-05,
+      "loss": 2.7972,
+      "step": 1650
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.6664507985115051,
+      "learning_rate": 9.996277328589618e-05,
+      "loss": 2.8151,
+      "step": 1655
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.6448256969451904,
+      "learning_rate": 9.996253353086866e-05,
+      "loss": 2.8761,
+      "step": 1660
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.6505598425865173,
+      "learning_rate": 9.996229300654729e-05,
+      "loss": 2.8807,
+      "step": 1665
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.6235655546188354,
+      "learning_rate": 9.996205171293576e-05,
+      "loss": 2.8681,
+      "step": 1670
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.6468040347099304,
+      "learning_rate": 9.996180965003778e-05,
+      "loss": 2.8269,
+      "step": 1675
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.6571494936943054,
+      "learning_rate": 9.996156681785708e-05,
+      "loss": 2.8208,
+      "step": 1680
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.7034950256347656,
+      "learning_rate": 9.996132321639741e-05,
+      "loss": 2.935,
+      "step": 1685
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.6909993886947632,
+      "learning_rate": 9.996107884566252e-05,
+      "loss": 2.8555,
+      "step": 1690
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.6134126782417297,
+      "learning_rate": 9.996083370565616e-05,
+      "loss": 2.7765,
+      "step": 1695
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.6870012879371643,
+      "learning_rate": 9.99605877963821e-05,
+      "loss": 2.7713,
+      "step": 1700
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.7347362637519836,
+      "learning_rate": 9.996034111784413e-05,
+      "loss": 2.9402,
+      "step": 1705
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.6576223373413086,
+      "learning_rate": 9.996009367004607e-05,
+      "loss": 2.7167,
+      "step": 1710
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.673149585723877,
+      "learning_rate": 9.995984545299171e-05,
+      "loss": 2.7983,
+      "step": 1715
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.642084538936615,
+      "learning_rate": 9.995959646668488e-05,
+      "loss": 2.7118,
+      "step": 1720
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.636038601398468,
+      "learning_rate": 9.995934671112941e-05,
+      "loss": 2.8,
+      "step": 1725
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.6760165691375732,
+      "learning_rate": 9.995909618632915e-05,
+      "loss": 2.8576,
+      "step": 1730
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.6632430553436279,
+      "learning_rate": 9.995884489228794e-05,
+      "loss": 2.8031,
+      "step": 1735
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.6511415243148804,
+      "learning_rate": 9.995859282900967e-05,
+      "loss": 2.8342,
+      "step": 1740
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.6549462676048279,
+      "learning_rate": 9.995833999649821e-05,
+      "loss": 2.7322,
+      "step": 1745
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.6769255995750427,
+      "learning_rate": 9.995808639475746e-05,
+      "loss": 2.8057,
+      "step": 1750
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.6149609088897705,
+      "learning_rate": 9.995783202379131e-05,
+      "loss": 2.8258,
+      "step": 1755
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.6772927641868591,
+      "learning_rate": 9.99575768836037e-05,
+      "loss": 2.8079,
+      "step": 1760
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.6345034241676331,
+      "learning_rate": 9.995732097419855e-05,
+      "loss": 2.8002,
+      "step": 1765
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.6717026829719543,
+      "learning_rate": 9.995706429557978e-05,
+      "loss": 2.7477,
+      "step": 1770
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.6907855272293091,
+      "learning_rate": 9.995680684775137e-05,
+      "loss": 2.8431,
+      "step": 1775
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.6253895163536072,
+      "learning_rate": 9.995654863071729e-05,
+      "loss": 2.8863,
+      "step": 1780
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.6945659518241882,
+      "learning_rate": 9.995628964448147e-05,
+      "loss": 2.7987,
+      "step": 1785
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.8099337220191956,
+      "learning_rate": 9.995602988904793e-05,
+      "loss": 2.7022,
+      "step": 1790
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.6706368923187256,
+      "learning_rate": 9.995576936442068e-05,
+      "loss": 2.8437,
+      "step": 1795
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.7580943703651428,
+      "learning_rate": 9.995550807060371e-05,
+      "loss": 2.8322,
+      "step": 1800
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.7035062909126282,
+      "learning_rate": 9.995524600760105e-05,
+      "loss": 2.7218,
+      "step": 1805
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.7245374917984009,
+      "learning_rate": 9.995498317541674e-05,
+      "loss": 2.8132,
+      "step": 1810
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.6628081798553467,
+      "learning_rate": 9.995471957405483e-05,
+      "loss": 2.8436,
+      "step": 1815
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.6123233437538147,
+      "learning_rate": 9.995445520351935e-05,
+      "loss": 2.7969,
+      "step": 1820
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.6512640714645386,
+      "learning_rate": 9.995419006381442e-05,
+      "loss": 2.8564,
+      "step": 1825
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.6438198685646057,
+      "learning_rate": 9.995392415494407e-05,
+      "loss": 2.7486,
+      "step": 1830
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.601305365562439,
+      "learning_rate": 9.995365747691243e-05,
+      "loss": 2.8082,
+      "step": 1835
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.5955356359481812,
+      "learning_rate": 9.995339002972359e-05,
+      "loss": 2.8414,
+      "step": 1840
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.6768097281455994,
+      "learning_rate": 9.995312181338168e-05,
+      "loss": 2.7051,
+      "step": 1845
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.6040539145469666,
+      "learning_rate": 9.995285282789082e-05,
+      "loss": 2.7353,
+      "step": 1850
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.6346877813339233,
+      "learning_rate": 9.995258307325515e-05,
+      "loss": 2.7684,
+      "step": 1855
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.5837364196777344,
+      "learning_rate": 9.995231254947883e-05,
+      "loss": 2.6003,
+      "step": 1860
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.6347076296806335,
+      "learning_rate": 9.995204125656603e-05,
+      "loss": 2.7752,
+      "step": 1865
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.6250349283218384,
+      "learning_rate": 9.995176919452092e-05,
+      "loss": 2.7095,
+      "step": 1870
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.6535242795944214,
+      "learning_rate": 9.995149636334769e-05,
+      "loss": 2.7406,
+      "step": 1875
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.5767995715141296,
+      "learning_rate": 9.995122276305054e-05,
+      "loss": 2.7389,
+      "step": 1880
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.6453811526298523,
+      "learning_rate": 9.995094839363368e-05,
+      "loss": 2.7643,
+      "step": 1885
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.6315706372261047,
+      "learning_rate": 9.995067325510132e-05,
+      "loss": 2.7943,
+      "step": 1890
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.6660724878311157,
+      "learning_rate": 9.995039734745774e-05,
+      "loss": 2.8042,
+      "step": 1895
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.7025864720344543,
+      "learning_rate": 9.995012067070715e-05,
+      "loss": 2.7302,
+      "step": 1900
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.6025550365447998,
+      "learning_rate": 9.994984322485383e-05,
+      "loss": 2.7969,
+      "step": 1905
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.730593204498291,
+      "learning_rate": 9.994956500990205e-05,
+      "loss": 2.7473,
+      "step": 1910
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.6390718221664429,
+      "learning_rate": 9.994928602585608e-05,
+      "loss": 2.6996,
+      "step": 1915
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.6751176118850708,
+      "learning_rate": 9.994900627272023e-05,
+      "loss": 2.7572,
+      "step": 1920
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.6733884215354919,
+      "learning_rate": 9.99487257504988e-05,
+      "loss": 2.7639,
+      "step": 1925
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.6398480534553528,
+      "learning_rate": 9.994844445919612e-05,
+      "loss": 2.735,
+      "step": 1930
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5983346104621887,
+      "learning_rate": 9.994816239881648e-05,
+      "loss": 2.7216,
+      "step": 1935
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.6451579332351685,
+      "learning_rate": 9.99478795693643e-05,
+      "loss": 2.7273,
+      "step": 1940
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.6150055527687073,
+      "learning_rate": 9.994759597084384e-05,
+      "loss": 2.6432,
+      "step": 1945
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.6546292901039124,
+      "learning_rate": 9.994731160325955e-05,
+      "loss": 2.8854,
+      "step": 1950
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.6489946246147156,
+      "learning_rate": 9.994702646661577e-05,
+      "loss": 2.741,
+      "step": 1955
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.665304958820343,
+      "learning_rate": 9.994674056091689e-05,
+      "loss": 2.7018,
+      "step": 1960
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.6176490187644958,
+      "learning_rate": 9.994645388616732e-05,
+      "loss": 2.7203,
+      "step": 1965
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.6725919246673584,
+      "learning_rate": 9.994616644237148e-05,
+      "loss": 2.7487,
+      "step": 1970
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.6747321486473083,
+      "learning_rate": 9.994587822953378e-05,
+      "loss": 2.764,
+      "step": 1975
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.6290194988250732,
+      "learning_rate": 9.994558924765866e-05,
+      "loss": 2.7465,
+      "step": 1980
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.6608841419219971,
+      "learning_rate": 9.99452994967506e-05,
+      "loss": 2.8012,
+      "step": 1985
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.7030755281448364,
+      "learning_rate": 9.994500897681401e-05,
+      "loss": 2.7318,
+      "step": 1990
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.6196607947349548,
+      "learning_rate": 9.994471768785339e-05,
+      "loss": 2.7237,
+      "step": 1995
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.6086550354957581,
+      "learning_rate": 9.994442562987324e-05,
+      "loss": 2.6639,
+      "step": 2000
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.617693305015564,
+      "learning_rate": 9.994413280287803e-05,
+      "loss": 2.8139,
+      "step": 2005
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.6554850339889526,
+      "learning_rate": 9.994383920687229e-05,
+      "loss": 2.6673,
+      "step": 2010
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.6311039924621582,
+      "learning_rate": 9.994354484186053e-05,
+      "loss": 2.7801,
+      "step": 2015
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.6508556008338928,
+      "learning_rate": 9.994324970784727e-05,
+      "loss": 2.7609,
+      "step": 2020
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.588809072971344,
+      "learning_rate": 9.994295380483709e-05,
+      "loss": 2.7569,
+      "step": 2025
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.6931658387184143,
+      "learning_rate": 9.99426571328345e-05,
+      "loss": 2.7992,
+      "step": 2030
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.6994143724441528,
+      "learning_rate": 9.994235969184411e-05,
+      "loss": 2.784,
+      "step": 2035
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.6947830319404602,
+      "learning_rate": 9.994206148187046e-05,
+      "loss": 2.7603,
+      "step": 2040
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.6069614887237549,
+      "learning_rate": 9.994176250291818e-05,
+      "loss": 2.7683,
+      "step": 2045
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5963587164878845,
+      "learning_rate": 9.994146275499186e-05,
+      "loss": 2.7955,
+      "step": 2050
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.6070330739021301,
+      "learning_rate": 9.994116223809611e-05,
+      "loss": 2.6986,
+      "step": 2055
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.630490779876709,
+      "learning_rate": 9.994086095223557e-05,
+      "loss": 2.7193,
+      "step": 2060
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.6512714624404907,
+      "learning_rate": 9.994055889741485e-05,
+      "loss": 2.6726,
+      "step": 2065
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.6699522137641907,
+      "learning_rate": 9.994025607363862e-05,
+      "loss": 2.7154,
+      "step": 2070
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.6233505606651306,
+      "learning_rate": 9.993995248091155e-05,
+      "loss": 2.7484,
+      "step": 2075
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5717555284500122,
+      "learning_rate": 9.993964811923831e-05,
+      "loss": 2.7284,
+      "step": 2080
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5785918235778809,
+      "learning_rate": 9.993934298862358e-05,
+      "loss": 2.6549,
+      "step": 2085
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.6514252424240112,
+      "learning_rate": 9.993903708907205e-05,
+      "loss": 2.7312,
+      "step": 2090
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.6372067332267761,
+      "learning_rate": 9.993873042058845e-05,
+      "loss": 2.7262,
+      "step": 2095
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.6199125647544861,
+      "learning_rate": 9.993842298317749e-05,
+      "loss": 2.7637,
+      "step": 2100
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5739587545394897,
+      "learning_rate": 9.993811477684392e-05,
+      "loss": 2.7659,
+      "step": 2105
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5999652743339539,
+      "learning_rate": 9.993780580159246e-05,
+      "loss": 2.7041,
+      "step": 2110
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.56877601146698,
+      "learning_rate": 9.993749605742788e-05,
+      "loss": 2.7147,
+      "step": 2115
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.6056075096130371,
+      "learning_rate": 9.993718554435495e-05,
+      "loss": 2.7424,
+      "step": 2120
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.6411418914794922,
+      "learning_rate": 9.993687426237846e-05,
+      "loss": 2.7668,
+      "step": 2125
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5816123485565186,
+      "learning_rate": 9.993656221150318e-05,
+      "loss": 2.7567,
+      "step": 2130
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.6219111084938049,
+      "learning_rate": 9.993624939173394e-05,
+      "loss": 2.8552,
+      "step": 2135
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.6099264025688171,
+      "learning_rate": 9.993593580307553e-05,
+      "loss": 2.7158,
+      "step": 2140
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.6999974250793457,
+      "learning_rate": 9.99356214455328e-05,
+      "loss": 2.7551,
+      "step": 2145
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.7445834279060364,
+      "learning_rate": 9.993530631911059e-05,
+      "loss": 2.7607,
+      "step": 2150
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.613828182220459,
+      "learning_rate": 9.993499042381373e-05,
+      "loss": 2.7311,
+      "step": 2155
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.633206307888031,
+      "learning_rate": 9.993467375964711e-05,
+      "loss": 2.7215,
+      "step": 2160
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.6213359832763672,
+      "learning_rate": 9.99343563266156e-05,
+      "loss": 2.7204,
+      "step": 2165
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.687370777130127,
+      "learning_rate": 9.993403812472408e-05,
+      "loss": 2.7624,
+      "step": 2170
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5912699103355408,
+      "learning_rate": 9.993371915397745e-05,
+      "loss": 2.7066,
+      "step": 2175
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5537623763084412,
+      "learning_rate": 9.993339941438062e-05,
+      "loss": 2.7614,
+      "step": 2180
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.6721332669258118,
+      "learning_rate": 9.993307890593853e-05,
+      "loss": 2.7324,
+      "step": 2185
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.6714296936988831,
+      "learning_rate": 9.993275762865609e-05,
+      "loss": 2.6645,
+      "step": 2190
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5782533288002014,
+      "learning_rate": 9.993243558253826e-05,
+      "loss": 2.7666,
+      "step": 2195
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.6402076482772827,
+      "learning_rate": 9.993211276759e-05,
+      "loss": 2.6749,
+      "step": 2200
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5687276721000671,
+      "learning_rate": 9.993178918381628e-05,
+      "loss": 2.5847,
+      "step": 2205
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5715349316596985,
+      "learning_rate": 9.993146483122207e-05,
+      "loss": 2.7355,
+      "step": 2210
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5898202657699585,
+      "learning_rate": 9.993113970981238e-05,
+      "loss": 2.662,
+      "step": 2215
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5721566677093506,
+      "learning_rate": 9.993081381959221e-05,
+      "loss": 2.6909,
+      "step": 2220
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5766157507896423,
+      "learning_rate": 9.99304871605666e-05,
+      "loss": 2.7232,
+      "step": 2225
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5969125628471375,
+      "learning_rate": 9.993015973274053e-05,
+      "loss": 2.7423,
+      "step": 2230
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5490349531173706,
+      "learning_rate": 9.992983153611908e-05,
+      "loss": 2.8285,
+      "step": 2235
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5740174055099487,
+      "learning_rate": 9.99295025707073e-05,
+      "loss": 2.7195,
+      "step": 2240
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5884872674942017,
+      "learning_rate": 9.992917283651024e-05,
+      "loss": 2.7451,
+      "step": 2245
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5237897634506226,
+      "learning_rate": 9.992884233353298e-05,
+      "loss": 2.7179,
+      "step": 2250
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5672905445098877,
+      "learning_rate": 9.992851106178063e-05,
+      "loss": 2.7654,
+      "step": 2255
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.6006327271461487,
+      "learning_rate": 9.992817902125826e-05,
+      "loss": 2.6982,
+      "step": 2260
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5753222107887268,
+      "learning_rate": 9.9927846211971e-05,
+      "loss": 2.7364,
+      "step": 2265
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5688061714172363,
+      "learning_rate": 9.992751263392397e-05,
+      "loss": 2.6469,
+      "step": 2270
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.6104544401168823,
+      "learning_rate": 9.992717828712232e-05,
+      "loss": 2.8129,
+      "step": 2275
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.602768063545227,
+      "learning_rate": 9.992684317157117e-05,
+      "loss": 2.6673,
+      "step": 2280
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.6040850877761841,
+      "learning_rate": 9.992650728727571e-05,
+      "loss": 2.6902,
+      "step": 2285
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.6339841485023499,
+      "learning_rate": 9.99261706342411e-05,
+      "loss": 2.7764,
+      "step": 2290
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.6991773843765259,
+      "learning_rate": 9.99258332124725e-05,
+      "loss": 2.6985,
+      "step": 2295
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.6763634085655212,
+      "learning_rate": 9.992549502197514e-05,
+      "loss": 2.6391,
+      "step": 2300
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.6468992233276367,
+      "learning_rate": 9.992515606275423e-05,
+      "loss": 2.6651,
+      "step": 2305
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.6157127022743225,
+      "learning_rate": 9.992481633481495e-05,
+      "loss": 2.8039,
+      "step": 2310
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.641611635684967,
+      "learning_rate": 9.992447583816257e-05,
+      "loss": 2.7191,
+      "step": 2315
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.6178282499313354,
+      "learning_rate": 9.992413457280231e-05,
+      "loss": 2.5942,
+      "step": 2320
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5593740344047546,
+      "learning_rate": 9.992379253873943e-05,
+      "loss": 2.6672,
+      "step": 2325
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.6506845951080322,
+      "learning_rate": 9.99234497359792e-05,
+      "loss": 2.6851,
+      "step": 2330
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5828286409378052,
+      "learning_rate": 9.992310616452689e-05,
+      "loss": 2.7015,
+      "step": 2335
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.6091148853302002,
+      "learning_rate": 9.99227618243878e-05,
+      "loss": 2.6862,
+      "step": 2340
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.609283447265625,
+      "learning_rate": 9.992241671556724e-05,
+      "loss": 2.6063,
+      "step": 2345
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.6670018434524536,
+      "learning_rate": 9.99220708380705e-05,
+      "loss": 2.7403,
+      "step": 2350
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.6938735842704773,
+      "learning_rate": 9.99217241919029e-05,
+      "loss": 2.7498,
+      "step": 2355
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5821777582168579,
+      "learning_rate": 9.992137677706982e-05,
+      "loss": 2.6726,
+      "step": 2360
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.6034961342811584,
+      "learning_rate": 9.99210285935766e-05,
+      "loss": 2.6038,
+      "step": 2365
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5970882773399353,
+      "learning_rate": 9.992067964142856e-05,
+      "loss": 2.7467,
+      "step": 2370
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5309962630271912,
+      "learning_rate": 9.992032992063109e-05,
+      "loss": 2.6774,
+      "step": 2375
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5913887023925781,
+      "learning_rate": 9.99199794311896e-05,
+      "loss": 2.687,
+      "step": 2380
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5764299631118774,
+      "learning_rate": 9.991962817310947e-05,
+      "loss": 2.7537,
+      "step": 2385
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5750162601470947,
+      "learning_rate": 9.991927614639612e-05,
+      "loss": 2.6923,
+      "step": 2390
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5983758568763733,
+      "learning_rate": 9.991892335105493e-05,
+      "loss": 2.7077,
+      "step": 2395
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5748116970062256,
+      "learning_rate": 9.991856978709137e-05,
+      "loss": 2.7748,
+      "step": 2400
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5736396908760071,
+      "learning_rate": 9.99182154545109e-05,
+      "loss": 2.7763,
+      "step": 2405
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5424933433532715,
+      "learning_rate": 9.991786035331893e-05,
+      "loss": 2.7038,
+      "step": 2410
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.6289833784103394,
+      "learning_rate": 9.991750448352096e-05,
+      "loss": 2.6713,
+      "step": 2415
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.6490828990936279,
+      "learning_rate": 9.991714784512243e-05,
+      "loss": 2.7822,
+      "step": 2420
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.6322473883628845,
+      "learning_rate": 9.991679043812887e-05,
+      "loss": 2.7188,
+      "step": 2425
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.6169449687004089,
+      "learning_rate": 9.99164322625458e-05,
+      "loss": 2.7029,
+      "step": 2430
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.6692633628845215,
+      "learning_rate": 9.991607331837867e-05,
+      "loss": 2.7554,
+      "step": 2435
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.6933794021606445,
+      "learning_rate": 9.991571360563306e-05,
+      "loss": 2.7734,
+      "step": 2440
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5623266696929932,
+      "learning_rate": 9.99153531243145e-05,
+      "loss": 2.7046,
+      "step": 2445
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.586891233921051,
+      "learning_rate": 9.991499187442852e-05,
+      "loss": 2.6785,
+      "step": 2450
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5435975790023804,
+      "learning_rate": 9.99146298559807e-05,
+      "loss": 2.6237,
+      "step": 2455
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5415518283843994,
+      "learning_rate": 9.991426706897661e-05,
+      "loss": 2.6983,
+      "step": 2460
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5765369534492493,
+      "learning_rate": 9.991390351342185e-05,
+      "loss": 2.7098,
+      "step": 2465
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5773332118988037,
+      "learning_rate": 9.991353918932198e-05,
+      "loss": 2.6634,
+      "step": 2470
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5211060643196106,
+      "learning_rate": 9.991317409668264e-05,
+      "loss": 2.6,
+      "step": 2475
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5817126631736755,
+      "learning_rate": 9.991280823550945e-05,
+      "loss": 2.7171,
+      "step": 2480
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5403845906257629,
+      "learning_rate": 9.991244160580803e-05,
+      "loss": 2.6576,
+      "step": 2485
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.57779860496521,
+      "learning_rate": 9.991207420758404e-05,
+      "loss": 2.7119,
+      "step": 2490
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5543129444122314,
+      "learning_rate": 9.991170604084314e-05,
+      "loss": 2.6238,
+      "step": 2495
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5874466896057129,
+      "learning_rate": 9.991133710559097e-05,
+      "loss": 2.6252,
+      "step": 2500
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.6209591031074524,
+      "learning_rate": 9.991096740183324e-05,
+      "loss": 2.6769,
+      "step": 2505
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.6356781125068665,
+      "learning_rate": 9.991059692957563e-05,
+      "loss": 2.7388,
+      "step": 2510
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5884143114089966,
+      "learning_rate": 9.991022568882384e-05,
+      "loss": 2.7081,
+      "step": 2515
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5594564080238342,
+      "learning_rate": 9.99098536795836e-05,
+      "loss": 2.6749,
+      "step": 2520
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5984323024749756,
+      "learning_rate": 9.990948090186063e-05,
+      "loss": 2.6325,
+      "step": 2525
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.6070553660392761,
+      "learning_rate": 9.990910735566065e-05,
+      "loss": 2.7255,
+      "step": 2530
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.55217444896698,
+      "learning_rate": 9.990873304098946e-05,
+      "loss": 2.6048,
+      "step": 2535
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5925323367118835,
+      "learning_rate": 9.990835795785278e-05,
+      "loss": 2.6717,
+      "step": 2540
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5467322468757629,
+      "learning_rate": 9.990798210625642e-05,
+      "loss": 2.681,
+      "step": 2545
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.4988623559474945,
+      "learning_rate": 9.990760548620613e-05,
+      "loss": 2.6212,
+      "step": 2550
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5875710844993591,
+      "learning_rate": 9.990722809770773e-05,
+      "loss": 2.6838,
+      "step": 2555
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5813723206520081,
+      "learning_rate": 9.990684994076702e-05,
+      "loss": 2.6651,
+      "step": 2560
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.6005766987800598,
+      "learning_rate": 9.990647101538983e-05,
+      "loss": 2.6771,
+      "step": 2565
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5362436771392822,
+      "learning_rate": 9.990609132158201e-05,
+      "loss": 2.7204,
+      "step": 2570
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5813928246498108,
+      "learning_rate": 9.990571085934938e-05,
+      "loss": 2.6838,
+      "step": 2575
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5389395356178284,
+      "learning_rate": 9.990532962869783e-05,
+      "loss": 2.7062,
+      "step": 2580
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5982468724250793,
+      "learning_rate": 9.990494762963318e-05,
+      "loss": 2.6969,
+      "step": 2585
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5913338661193848,
+      "learning_rate": 9.990456486216136e-05,
+      "loss": 2.6712,
+      "step": 2590
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5049490928649902,
+      "learning_rate": 9.990418132628824e-05,
+      "loss": 2.5927,
+      "step": 2595
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.52427738904953,
+      "learning_rate": 9.990379702201974e-05,
+      "loss": 2.6131,
+      "step": 2600
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5589879155158997,
+      "learning_rate": 9.990341194936173e-05,
+      "loss": 2.5454,
+      "step": 2605
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5354021787643433,
+      "learning_rate": 9.990302610832021e-05,
+      "loss": 2.6685,
+      "step": 2610
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5741543173789978,
+      "learning_rate": 9.990263949890108e-05,
+      "loss": 2.7274,
+      "step": 2615
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5555664896965027,
+      "learning_rate": 9.990225212111029e-05,
+      "loss": 2.6916,
+      "step": 2620
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5612133145332336,
+      "learning_rate": 9.990186397495383e-05,
+      "loss": 2.669,
+      "step": 2625
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5987478494644165,
+      "learning_rate": 9.990147506043766e-05,
+      "loss": 2.6221,
+      "step": 2630
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5772538185119629,
+      "learning_rate": 9.990108537756775e-05,
+      "loss": 2.7216,
+      "step": 2635
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5765659809112549,
+      "learning_rate": 9.990069492635015e-05,
+      "loss": 2.6295,
+      "step": 2640
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5559474229812622,
+      "learning_rate": 9.990030370679082e-05,
+      "loss": 2.6187,
+      "step": 2645
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5544140934944153,
+      "learning_rate": 9.989991171889581e-05,
+      "loss": 2.6433,
+      "step": 2650
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5299431681632996,
+      "learning_rate": 9.989951896267115e-05,
+      "loss": 2.6643,
+      "step": 2655
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.6052196621894836,
+      "learning_rate": 9.989912543812289e-05,
+      "loss": 2.6199,
+      "step": 2660
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5104473829269409,
+      "learning_rate": 9.989873114525709e-05,
+      "loss": 2.654,
+      "step": 2665
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.552469789981842,
+      "learning_rate": 9.989833608407981e-05,
+      "loss": 2.5741,
+      "step": 2670
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5314912796020508,
+      "learning_rate": 9.989794025459715e-05,
+      "loss": 2.6442,
+      "step": 2675
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.597647488117218,
+      "learning_rate": 9.989754365681519e-05,
+      "loss": 2.6623,
+      "step": 2680
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5405697822570801,
+      "learning_rate": 9.989714629074006e-05,
+      "loss": 2.7265,
+      "step": 2685
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5137559771537781,
+      "learning_rate": 9.989674815637784e-05,
+      "loss": 2.6695,
+      "step": 2690
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.6408671140670776,
+      "learning_rate": 9.989634925373468e-05,
+      "loss": 2.6163,
+      "step": 2695
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5638792514801025,
+      "learning_rate": 9.989594958281673e-05,
+      "loss": 2.7539,
+      "step": 2700
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5039884448051453,
+      "learning_rate": 9.989554914363015e-05,
+      "loss": 2.4833,
+      "step": 2705
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5392247438430786,
+      "learning_rate": 9.989514793618109e-05,
+      "loss": 2.7495,
+      "step": 2710
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5745425820350647,
+      "learning_rate": 9.989474596047572e-05,
+      "loss": 2.7901,
+      "step": 2715
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.571323037147522,
+      "learning_rate": 9.989434321652024e-05,
+      "loss": 2.6045,
+      "step": 2720
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5741243362426758,
+      "learning_rate": 9.989393970432085e-05,
+      "loss": 2.5054,
+      "step": 2725
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5181983113288879,
+      "learning_rate": 9.989353542388378e-05,
+      "loss": 2.5841,
+      "step": 2730
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.604209303855896,
+      "learning_rate": 9.989313037521524e-05,
+      "loss": 2.7155,
+      "step": 2735
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5166040658950806,
+      "learning_rate": 9.989272455832145e-05,
+      "loss": 2.7394,
+      "step": 2740
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5321922898292542,
+      "learning_rate": 9.989231797320869e-05,
+      "loss": 2.6602,
+      "step": 2745
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5643978714942932,
+      "learning_rate": 9.989191061988319e-05,
+      "loss": 2.6051,
+      "step": 2750
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5863328576087952,
+      "learning_rate": 9.989150249835125e-05,
+      "loss": 2.7124,
+      "step": 2755
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5373572111129761,
+      "learning_rate": 9.989109360861913e-05,
+      "loss": 2.65,
+      "step": 2760
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.6019943356513977,
+      "learning_rate": 9.989068395069313e-05,
+      "loss": 2.7213,
+      "step": 2765
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.6263430118560791,
+      "learning_rate": 9.989027352457958e-05,
+      "loss": 2.5849,
+      "step": 2770
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5649701952934265,
+      "learning_rate": 9.988986233028478e-05,
+      "loss": 2.608,
+      "step": 2775
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5405789613723755,
+      "learning_rate": 9.988945036781507e-05,
+      "loss": 2.6434,
+      "step": 2780
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5272497534751892,
+      "learning_rate": 9.988903763717679e-05,
+      "loss": 2.6461,
+      "step": 2785
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.6319659352302551,
+      "learning_rate": 9.988862413837628e-05,
+      "loss": 2.5883,
+      "step": 2790
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5725058317184448,
+      "learning_rate": 9.988820987141993e-05,
+      "loss": 2.6331,
+      "step": 2795
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5227794051170349,
+      "learning_rate": 9.988779483631409e-05,
+      "loss": 2.6443,
+      "step": 2800
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5595724582672119,
+      "learning_rate": 9.98873790330652e-05,
+      "loss": 2.6604,
+      "step": 2805
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5499975085258484,
+      "learning_rate": 9.988696246167961e-05,
+      "loss": 2.7407,
+      "step": 2810
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5480114221572876,
+      "learning_rate": 9.988654512216378e-05,
+      "loss": 2.6302,
+      "step": 2815
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.6080053448677063,
+      "learning_rate": 9.988612701452409e-05,
+      "loss": 2.5702,
+      "step": 2820
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.505370557308197,
+      "learning_rate": 9.988570813876701e-05,
+      "loss": 2.6062,
+      "step": 2825
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5510420203208923,
+      "learning_rate": 9.988528849489897e-05,
+      "loss": 2.693,
+      "step": 2830
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5333909392356873,
+      "learning_rate": 9.988486808292646e-05,
+      "loss": 2.6553,
+      "step": 2835
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.555931806564331,
+      "learning_rate": 9.988444690285592e-05,
+      "loss": 2.6331,
+      "step": 2840
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5535305142402649,
+      "learning_rate": 9.988402495469385e-05,
+      "loss": 2.6618,
+      "step": 2845
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5386797189712524,
+      "learning_rate": 9.988360223844673e-05,
+      "loss": 2.5676,
+      "step": 2850
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5144925713539124,
+      "learning_rate": 9.98831787541211e-05,
+      "loss": 2.5554,
+      "step": 2855
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5252721905708313,
+      "learning_rate": 9.988275450172346e-05,
+      "loss": 2.638,
+      "step": 2860
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5819863677024841,
+      "learning_rate": 9.988232948126037e-05,
+      "loss": 2.6083,
+      "step": 2865
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5913493633270264,
+      "learning_rate": 9.988190369273833e-05,
+      "loss": 2.5842,
+      "step": 2870
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.544028103351593,
+      "learning_rate": 9.988147713616391e-05,
+      "loss": 2.6351,
+      "step": 2875
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5454347133636475,
+      "learning_rate": 9.988104981154371e-05,
+      "loss": 2.5372,
+      "step": 2880
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.550319492816925,
+      "learning_rate": 9.988062171888429e-05,
+      "loss": 2.7124,
+      "step": 2885
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5082794427871704,
+      "learning_rate": 9.988019285819221e-05,
+      "loss": 2.6929,
+      "step": 2890
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.6078589558601379,
+      "learning_rate": 9.987976322947411e-05,
+      "loss": 2.7297,
+      "step": 2895
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5638788342475891,
+      "learning_rate": 9.98793328327366e-05,
+      "loss": 2.6205,
+      "step": 2900
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5580174326896667,
+      "learning_rate": 9.98789016679863e-05,
+      "loss": 2.6068,
+      "step": 2905
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5718059539794922,
+      "learning_rate": 9.987846973522985e-05,
+      "loss": 2.6441,
+      "step": 2910
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5545582175254822,
+      "learning_rate": 9.98780370344739e-05,
+      "loss": 2.7292,
+      "step": 2915
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5168636441230774,
+      "learning_rate": 9.987760356572512e-05,
+      "loss": 2.6924,
+      "step": 2920
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5569320917129517,
+      "learning_rate": 9.987716932899018e-05,
+      "loss": 2.6483,
+      "step": 2925
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5048226714134216,
+      "learning_rate": 9.987673432427576e-05,
+      "loss": 2.6053,
+      "step": 2930
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5176882147789001,
+      "learning_rate": 9.987629855158855e-05,
+      "loss": 2.7085,
+      "step": 2935
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5857419967651367,
+      "learning_rate": 9.98758620109353e-05,
+      "loss": 2.5296,
+      "step": 2940
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5672727823257446,
+      "learning_rate": 9.98754247023227e-05,
+      "loss": 2.6384,
+      "step": 2945
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.4911515414714813,
+      "learning_rate": 9.987498662575747e-05,
+      "loss": 2.5969,
+      "step": 2950
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5802367329597473,
+      "learning_rate": 9.987454778124639e-05,
+      "loss": 2.6371,
+      "step": 2955
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5470579266548157,
+      "learning_rate": 9.987410816879619e-05,
+      "loss": 2.4736,
+      "step": 2960
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5062541365623474,
+      "learning_rate": 9.987366778841365e-05,
+      "loss": 2.6416,
+      "step": 2965
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5446923971176147,
+      "learning_rate": 9.987322664010554e-05,
+      "loss": 2.6122,
+      "step": 2970
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5244519114494324,
+      "learning_rate": 9.987278472387867e-05,
+      "loss": 2.5687,
+      "step": 2975
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5444034337997437,
+      "learning_rate": 9.987234203973986e-05,
+      "loss": 2.7432,
+      "step": 2980
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5287409424781799,
+      "learning_rate": 9.987189858769586e-05,
+      "loss": 2.7428,
+      "step": 2985
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5107440948486328,
+      "learning_rate": 9.987145436775357e-05,
+      "loss": 2.557,
+      "step": 2990
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.510742723941803,
+      "learning_rate": 9.98710093799198e-05,
+      "loss": 2.6342,
+      "step": 2995
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5040067434310913,
+      "learning_rate": 9.987056362420138e-05,
+      "loss": 2.6143,
+      "step": 3000
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5180338025093079,
+      "learning_rate": 9.98701171006052e-05,
+      "loss": 2.6114,
+      "step": 3005
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5323936939239502,
+      "learning_rate": 9.986966980913817e-05,
+      "loss": 2.6513,
+      "step": 3010
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5110459923744202,
+      "learning_rate": 9.986922174980709e-05,
+      "loss": 2.547,
+      "step": 3015
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5365689396858215,
+      "learning_rate": 9.986877292261893e-05,
+      "loss": 2.6234,
+      "step": 3020
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5105913281440735,
+      "learning_rate": 9.986832332758058e-05,
+      "loss": 2.5931,
+      "step": 3025
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5679191946983337,
+      "learning_rate": 9.986787296469895e-05,
+      "loss": 2.655,
+      "step": 3030
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5128864645957947,
+      "learning_rate": 9.986742183398099e-05,
+      "loss": 2.6172,
+      "step": 3035
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.518369197845459,
+      "learning_rate": 9.986696993543366e-05,
+      "loss": 2.5874,
+      "step": 3040
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.4891831576824188,
+      "learning_rate": 9.986651726906387e-05,
+      "loss": 2.6832,
+      "step": 3045
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.485911101102829,
+      "learning_rate": 9.986606383487862e-05,
+      "loss": 2.5132,
+      "step": 3050
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5017186403274536,
+      "learning_rate": 9.98656096328849e-05,
+      "loss": 2.6087,
+      "step": 3055
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5399907827377319,
+      "learning_rate": 9.98651546630897e-05,
+      "loss": 2.5782,
+      "step": 3060
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5191899538040161,
+      "learning_rate": 9.986469892550001e-05,
+      "loss": 2.5931,
+      "step": 3065
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5090551376342773,
+      "learning_rate": 9.986424242012287e-05,
+      "loss": 2.5557,
+      "step": 3070
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5340197086334229,
+      "learning_rate": 9.986378514696528e-05,
+      "loss": 2.6562,
+      "step": 3075
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5349440574645996,
+      "learning_rate": 9.98633271060343e-05,
+      "loss": 2.6376,
+      "step": 3080
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5303201079368591,
+      "learning_rate": 9.9862868297337e-05,
+      "loss": 2.6228,
+      "step": 3085
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5016071796417236,
+      "learning_rate": 9.98624087208804e-05,
+      "loss": 2.5989,
+      "step": 3090
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5264772176742554,
+      "learning_rate": 9.98619483766716e-05,
+      "loss": 2.5559,
+      "step": 3095
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5290347337722778,
+      "learning_rate": 9.98614872647177e-05,
+      "loss": 2.5726,
+      "step": 3100
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5177962183952332,
+      "learning_rate": 9.986102538502579e-05,
+      "loss": 2.6848,
+      "step": 3105
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5287919640541077,
+      "learning_rate": 9.986056273760297e-05,
+      "loss": 2.6159,
+      "step": 3110
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5595407485961914,
+      "learning_rate": 9.986009932245637e-05,
+      "loss": 2.5244,
+      "step": 3115
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5530325174331665,
+      "learning_rate": 9.985963513959313e-05,
+      "loss": 2.6169,
+      "step": 3120
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.6002697944641113,
+      "learning_rate": 9.985917018902041e-05,
+      "loss": 2.4956,
+      "step": 3125
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.52015221118927,
+      "learning_rate": 9.985870447074535e-05,
+      "loss": 2.5692,
+      "step": 3130
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5241773128509521,
+      "learning_rate": 9.985823798477512e-05,
+      "loss": 2.6298,
+      "step": 3135
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.4885421395301819,
+      "learning_rate": 9.985777073111691e-05,
+      "loss": 2.4719,
+      "step": 3140
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.49956104159355164,
+      "learning_rate": 9.985730270977791e-05,
+      "loss": 2.6468,
+      "step": 3145
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.46617591381073,
+      "learning_rate": 9.985683392076534e-05,
+      "loss": 2.5804,
+      "step": 3150
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5351355075836182,
+      "learning_rate": 9.98563643640864e-05,
+      "loss": 2.5375,
+      "step": 3155
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5175086855888367,
+      "learning_rate": 9.985589403974833e-05,
+      "loss": 2.5813,
+      "step": 3160
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5062843561172485,
+      "learning_rate": 9.985542294775838e-05,
+      "loss": 2.6006,
+      "step": 3165
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.47827327251434326,
+      "learning_rate": 9.985495108812377e-05,
+      "loss": 2.5545,
+      "step": 3170
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.5502398014068604,
+      "learning_rate": 9.985447846085181e-05,
+      "loss": 2.6709,
+      "step": 3175
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.543904721736908,
+      "learning_rate": 9.985400506594974e-05,
+      "loss": 2.6048,
+      "step": 3180
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.481058269739151,
+      "learning_rate": 9.985353090342487e-05,
+      "loss": 2.6135,
+      "step": 3185
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.5367178916931152,
+      "learning_rate": 9.985305597328449e-05,
+      "loss": 2.5536,
+      "step": 3190
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.5199053287506104,
+      "learning_rate": 9.985258027553593e-05,
+      "loss": 2.6187,
+      "step": 3195
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.493454247713089,
+      "learning_rate": 9.98521038101865e-05,
+      "loss": 2.5367,
+      "step": 3200
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.5170598030090332,
+      "learning_rate": 9.985162657724354e-05,
+      "loss": 2.6103,
+      "step": 3205
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.5908184051513672,
+      "learning_rate": 9.98511485767144e-05,
+      "loss": 2.5348,
+      "step": 3210
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.5765467286109924,
+      "learning_rate": 9.985066980860642e-05,
+      "loss": 2.5941,
+      "step": 3215
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.49380967020988464,
+      "learning_rate": 9.9850190272927e-05,
+      "loss": 2.6713,
+      "step": 3220
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.48647281527519226,
+      "learning_rate": 9.984970996968352e-05,
+      "loss": 2.5877,
+      "step": 3225
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.5079899430274963,
+      "learning_rate": 9.984922889888335e-05,
+      "loss": 2.6233,
+      "step": 3230
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.5467026233673096,
+      "learning_rate": 9.984874706053393e-05,
+      "loss": 2.5929,
+      "step": 3235
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.4852086007595062,
+      "learning_rate": 9.984826445464266e-05,
+      "loss": 2.5811,
+      "step": 3240
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.5032371282577515,
+      "learning_rate": 9.984778108121698e-05,
+      "loss": 2.6486,
+      "step": 3245
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.4881698787212372,
+      "learning_rate": 9.984729694026431e-05,
+      "loss": 2.5671,
+      "step": 3250
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.5246455073356628,
+      "learning_rate": 9.984681203179214e-05,
+      "loss": 2.5498,
+      "step": 3255
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.5395731925964355,
+      "learning_rate": 9.984632635580792e-05,
+      "loss": 2.6499,
+      "step": 3260
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.49171948432922363,
+      "learning_rate": 9.984583991231913e-05,
+      "loss": 2.6128,
+      "step": 3265
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.500321090221405,
+      "learning_rate": 9.984535270133324e-05,
+      "loss": 2.542,
+      "step": 3270
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.5410853028297424,
+      "learning_rate": 9.984486472285777e-05,
+      "loss": 2.5524,
+      "step": 3275
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.531204879283905,
+      "learning_rate": 9.984437597690024e-05,
+      "loss": 2.4671,
+      "step": 3280
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.5418881177902222,
+      "learning_rate": 9.984388646346817e-05,
+      "loss": 2.5162,
+      "step": 3285
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.5275403261184692,
+      "learning_rate": 9.984339618256909e-05,
+      "loss": 2.5734,
+      "step": 3290
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.4769732356071472,
+      "learning_rate": 9.984290513421057e-05,
+      "loss": 2.6024,
+      "step": 3295
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.5629391670227051,
+      "learning_rate": 9.984241331840015e-05,
+      "loss": 2.5811,
+      "step": 3300
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.4940381348133087,
+      "learning_rate": 9.984192073514538e-05,
+      "loss": 2.6319,
+      "step": 3305
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.5072768330574036,
+      "learning_rate": 9.98414273844539e-05,
+      "loss": 2.551,
+      "step": 3310
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.49973148107528687,
+      "learning_rate": 9.984093326633325e-05,
+      "loss": 2.5009,
+      "step": 3315
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.4888647794723511,
+      "learning_rate": 9.98404383807911e-05,
+      "loss": 2.5894,
+      "step": 3320
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.5176356434822083,
+      "learning_rate": 9.983994272783502e-05,
+      "loss": 2.5825,
+      "step": 3325
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.4572547674179077,
+      "learning_rate": 9.983944630747266e-05,
+      "loss": 2.5558,
+      "step": 3330
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.47556301951408386,
+      "learning_rate": 9.983894911971167e-05,
+      "loss": 2.6361,
+      "step": 3335
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.48908770084381104,
+      "learning_rate": 9.983845116455969e-05,
+      "loss": 2.6257,
+      "step": 3340
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.4645969867706299,
+      "learning_rate": 9.983795244202439e-05,
+      "loss": 2.6128,
+      "step": 3345
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.48000428080558777,
+      "learning_rate": 9.983745295211347e-05,
+      "loss": 2.4907,
+      "step": 3350
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.4813070595264435,
+      "learning_rate": 9.983695269483459e-05,
+      "loss": 2.6143,
+      "step": 3355
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.4964388906955719,
+      "learning_rate": 9.983645167019546e-05,
+      "loss": 2.5773,
+      "step": 3360
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.510116696357727,
+      "learning_rate": 9.983594987820381e-05,
+      "loss": 2.6534,
+      "step": 3365
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.47809067368507385,
+      "learning_rate": 9.983544731886735e-05,
+      "loss": 2.5677,
+      "step": 3370
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.5593171119689941,
+      "learning_rate": 9.983494399219386e-05,
+      "loss": 2.7021,
+      "step": 3375
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.4847419559955597,
+      "learning_rate": 9.983443989819103e-05,
+      "loss": 2.6028,
+      "step": 3380
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.4975123703479767,
+      "learning_rate": 9.983393503686665e-05,
+      "loss": 2.6004,
+      "step": 3385
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.5049152970314026,
+      "learning_rate": 9.983342940822849e-05,
+      "loss": 2.5931,
+      "step": 3390
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.4946036636829376,
+      "learning_rate": 9.983292301228432e-05,
+      "loss": 2.6388,
+      "step": 3395
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.4994827210903168,
+      "learning_rate": 9.983241584904197e-05,
+      "loss": 2.6019,
+      "step": 3400
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.5198677778244019,
+      "learning_rate": 9.983190791850925e-05,
+      "loss": 2.5192,
+      "step": 3405
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.5216033458709717,
+      "learning_rate": 9.983139922069394e-05,
+      "loss": 2.5777,
+      "step": 3410
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.47109347581863403,
+      "learning_rate": 9.983088975560391e-05,
+      "loss": 2.6076,
+      "step": 3415
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.1555533409118652,
+      "learning_rate": 9.983037952324699e-05,
+      "loss": 2.5838,
+      "step": 3420
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.5344683527946472,
+      "learning_rate": 9.982986852363102e-05,
+      "loss": 2.566,
+      "step": 3425
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.498261034488678,
+      "learning_rate": 9.98293567567639e-05,
+      "loss": 2.5613,
+      "step": 3430
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.5390560030937195,
+      "learning_rate": 9.982884422265349e-05,
+      "loss": 2.4527,
+      "step": 3435
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.49176347255706787,
+      "learning_rate": 9.982833092130768e-05,
+      "loss": 2.6637,
+      "step": 3440
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.5294800996780396,
+      "learning_rate": 9.982781685273438e-05,
+      "loss": 2.57,
+      "step": 3445
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.5037844777107239,
+      "learning_rate": 9.982730201694153e-05,
+      "loss": 2.6343,
+      "step": 3450
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.503943145275116,
+      "learning_rate": 9.982678641393701e-05,
+      "loss": 2.6364,
+      "step": 3455
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.4968451261520386,
+      "learning_rate": 9.98262700437288e-05,
+      "loss": 2.6214,
+      "step": 3460
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.525917112827301,
+      "learning_rate": 9.982575290632481e-05,
+      "loss": 2.5556,
+      "step": 3465
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.5483548045158386,
+      "learning_rate": 9.982523500173304e-05,
+      "loss": 2.6046,
+      "step": 3470
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.5123624205589294,
+      "learning_rate": 9.982471632996144e-05,
+      "loss": 2.5533,
+      "step": 3475
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.46716803312301636,
+      "learning_rate": 9.9824196891018e-05,
+      "loss": 2.5841,
+      "step": 3480
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.4695746600627899,
+      "learning_rate": 9.982367668491075e-05,
+      "loss": 2.5975,
+      "step": 3485
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.485144704580307,
+      "learning_rate": 9.982315571164765e-05,
+      "loss": 2.5906,
+      "step": 3490
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.5364880561828613,
+      "learning_rate": 9.982263397123675e-05,
+      "loss": 2.5063,
+      "step": 3495
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.5460398197174072,
+      "learning_rate": 9.982211146368608e-05,
+      "loss": 2.6254,
+      "step": 3500
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.48980286717414856,
+      "learning_rate": 9.982158818900369e-05,
+      "loss": 2.5919,
+      "step": 3505
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.5242944359779358,
+      "learning_rate": 9.982106414719762e-05,
+      "loss": 2.6082,
+      "step": 3510
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.5110018849372864,
+      "learning_rate": 9.982053933827595e-05,
+      "loss": 2.67,
+      "step": 3515
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.4983656704425812,
+      "learning_rate": 9.982001376224677e-05,
+      "loss": 2.677,
+      "step": 3520
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.501055896282196,
+      "learning_rate": 9.981948741911816e-05,
+      "loss": 2.5834,
+      "step": 3525
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.5072791576385498,
+      "learning_rate": 9.981896030889821e-05,
+      "loss": 2.5365,
+      "step": 3530
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.47576603293418884,
+      "learning_rate": 9.981843243159507e-05,
+      "loss": 2.6288,
+      "step": 3535
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.4397449493408203,
+      "learning_rate": 9.981790378721684e-05,
+      "loss": 2.5283,
+      "step": 3540
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.46026861667633057,
+      "learning_rate": 9.98173743757717e-05,
+      "loss": 2.4948,
+      "step": 3545
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.49059316515922546,
+      "learning_rate": 9.981684419726775e-05,
+      "loss": 2.6161,
+      "step": 3550
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.5209315419197083,
+      "learning_rate": 9.981631325171318e-05,
+      "loss": 2.5697,
+      "step": 3555
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.4736451506614685,
+      "learning_rate": 9.981578153911616e-05,
+      "loss": 2.5746,
+      "step": 3560
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.4567309319972992,
+      "learning_rate": 9.981524905948487e-05,
+      "loss": 2.5661,
+      "step": 3565
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.4922211766242981,
+      "learning_rate": 9.981471581282753e-05,
+      "loss": 2.6295,
+      "step": 3570
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.5350914001464844,
+      "learning_rate": 9.981418179915234e-05,
+      "loss": 2.6727,
+      "step": 3575
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.5292612314224243,
+      "learning_rate": 9.981364701846752e-05,
+      "loss": 2.5876,
+      "step": 3580
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.5646490454673767,
+      "learning_rate": 9.981311147078128e-05,
+      "loss": 2.6677,
+      "step": 3585
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.5302039384841919,
+      "learning_rate": 9.981257515610192e-05,
+      "loss": 2.4713,
+      "step": 3590
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.5877442359924316,
+      "learning_rate": 9.981203807443764e-05,
+      "loss": 2.533,
+      "step": 3595
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.48557978868484497,
+      "learning_rate": 9.981150022579676e-05,
+      "loss": 2.5381,
+      "step": 3600
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.5066747069358826,
+      "learning_rate": 9.981096161018754e-05,
+      "loss": 2.5167,
+      "step": 3605
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.5177069306373596,
+      "learning_rate": 9.981042222761826e-05,
+      "loss": 2.5649,
+      "step": 3610
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.48275741934776306,
+      "learning_rate": 9.980988207809726e-05,
+      "loss": 2.6455,
+      "step": 3615
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.48381468653678894,
+      "learning_rate": 9.98093411616328e-05,
+      "loss": 2.4213,
+      "step": 3620
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.5026090145111084,
+      "learning_rate": 9.980879947823327e-05,
+      "loss": 2.6385,
+      "step": 3625
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.47761353850364685,
+      "learning_rate": 9.980825702790696e-05,
+      "loss": 2.5652,
+      "step": 3630
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.44203248620033264,
+      "learning_rate": 9.980771381066227e-05,
+      "loss": 2.494,
+      "step": 3635
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.4508872330188751,
+      "learning_rate": 9.980716982650753e-05,
+      "loss": 2.609,
+      "step": 3640
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.5295490026473999,
+      "learning_rate": 9.980662507545113e-05,
+      "loss": 2.5228,
+      "step": 3645
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.48674529790878296,
+      "learning_rate": 9.980607955750146e-05,
+      "loss": 2.5476,
+      "step": 3650
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.4615531265735626,
+      "learning_rate": 9.98055332726669e-05,
+      "loss": 2.5276,
+      "step": 3655
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.48559024930000305,
+      "learning_rate": 9.980498622095588e-05,
+      "loss": 2.6413,
+      "step": 3660
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.5614455938339233,
+      "learning_rate": 9.980443840237681e-05,
+      "loss": 2.462,
+      "step": 3665
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.5105487704277039,
+      "learning_rate": 9.980388981693816e-05,
+      "loss": 2.6195,
+      "step": 3670
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.5444942712783813,
+      "learning_rate": 9.980334046464832e-05,
+      "loss": 2.701,
+      "step": 3675
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.4763965904712677,
+      "learning_rate": 9.980279034551579e-05,
+      "loss": 2.5484,
+      "step": 3680
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.5001147389411926,
+      "learning_rate": 9.980223945954902e-05,
+      "loss": 2.6035,
+      "step": 3685
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.5603012442588806,
+      "learning_rate": 9.980168780675651e-05,
+      "loss": 2.5374,
+      "step": 3690
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.4977169632911682,
+      "learning_rate": 9.980113538714675e-05,
+      "loss": 2.5156,
+      "step": 3695
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.5164822340011597,
+      "learning_rate": 9.980058220072825e-05,
+      "loss": 2.6557,
+      "step": 3700
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.5307486057281494,
+      "learning_rate": 9.98000282475095e-05,
+      "loss": 2.5576,
+      "step": 3705
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.46989819407463074,
+      "learning_rate": 9.979947352749906e-05,
+      "loss": 2.4733,
+      "step": 3710
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.49327266216278076,
+      "learning_rate": 9.979891804070545e-05,
+      "loss": 2.5178,
+      "step": 3715
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.47982510924339294,
+      "learning_rate": 9.979836178713724e-05,
+      "loss": 2.5621,
+      "step": 3720
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.4735487699508667,
+      "learning_rate": 9.979780476680298e-05,
+      "loss": 2.6258,
+      "step": 3725
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.4842422604560852,
+      "learning_rate": 9.979724697971125e-05,
+      "loss": 2.5124,
+      "step": 3730
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.49693992733955383,
+      "learning_rate": 9.979668842587065e-05,
+      "loss": 2.5807,
+      "step": 3735
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.49742311239242554,
+      "learning_rate": 9.979612910528976e-05,
+      "loss": 2.6919,
+      "step": 3740
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.4781762659549713,
+      "learning_rate": 9.979556901797723e-05,
+      "loss": 2.5306,
+      "step": 3745
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.4646557867527008,
+      "learning_rate": 9.979500816394165e-05,
+      "loss": 2.6078,
+      "step": 3750
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.4689471125602722,
+      "learning_rate": 9.979444654319165e-05,
+      "loss": 2.4977,
+      "step": 3755
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.46156519651412964,
+      "learning_rate": 9.979388415573591e-05,
+      "loss": 2.6404,
+      "step": 3760
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.4947497546672821,
+      "learning_rate": 9.979332100158307e-05,
+      "loss": 2.4952,
+      "step": 3765
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.5202804803848267,
+      "learning_rate": 9.979275708074179e-05,
+      "loss": 2.4424,
+      "step": 3770
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.4524839222431183,
+      "learning_rate": 9.979219239322078e-05,
+      "loss": 2.5387,
+      "step": 3775
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.45489701628685,
+      "learning_rate": 9.979162693902871e-05,
+      "loss": 2.4988,
+      "step": 3780
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.47594428062438965,
+      "learning_rate": 9.97910607181743e-05,
+      "loss": 2.5872,
+      "step": 3785
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.5086585879325867,
+      "learning_rate": 9.979049373066626e-05,
+      "loss": 2.5137,
+      "step": 3790
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.47109028697013855,
+      "learning_rate": 9.978992597651334e-05,
+      "loss": 2.533,
+      "step": 3795
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.49588626623153687,
+      "learning_rate": 9.978935745572424e-05,
+      "loss": 2.5799,
+      "step": 3800
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.4853939414024353,
+      "learning_rate": 9.978878816830776e-05,
+      "loss": 2.5375,
+      "step": 3805
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.5055867433547974,
+      "learning_rate": 9.978821811427263e-05,
+      "loss": 2.571,
+      "step": 3810
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.49028271436691284,
+      "learning_rate": 9.978764729362766e-05,
+      "loss": 2.6085,
+      "step": 3815
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.4496750831604004,
+      "learning_rate": 9.978707570638162e-05,
+      "loss": 2.5715,
+      "step": 3820
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.4813396632671356,
+      "learning_rate": 9.978650335254332e-05,
+      "loss": 2.6062,
+      "step": 3825
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.4723066985607147,
+      "learning_rate": 9.978593023212155e-05,
+      "loss": 2.5725,
+      "step": 3830
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.4774017333984375,
+      "learning_rate": 9.978535634512515e-05,
+      "loss": 2.5474,
+      "step": 3835
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.525672972202301,
+      "learning_rate": 9.978478169156298e-05,
+      "loss": 2.567,
+      "step": 3840
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.47154009342193604,
+      "learning_rate": 9.978420627144385e-05,
+      "loss": 2.6013,
+      "step": 3845
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.48119503259658813,
+      "learning_rate": 9.978363008477665e-05,
+      "loss": 2.5469,
+      "step": 3850
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.4608328342437744,
+      "learning_rate": 9.978305313157022e-05,
+      "loss": 2.4958,
+      "step": 3855
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.5009128451347351,
+      "learning_rate": 9.978247541183348e-05,
+      "loss": 2.5151,
+      "step": 3860
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.48547592759132385,
+      "learning_rate": 9.978189692557529e-05,
+      "loss": 2.5977,
+      "step": 3865
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.4770568907260895,
+      "learning_rate": 9.978131767280456e-05,
+      "loss": 2.7146,
+      "step": 3870
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.4893927276134491,
+      "learning_rate": 9.978073765353026e-05,
+      "loss": 2.3987,
+      "step": 3875
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.4994083344936371,
+      "learning_rate": 9.978015686776125e-05,
+      "loss": 2.5058,
+      "step": 3880
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.4648747444152832,
+      "learning_rate": 9.977957531550652e-05,
+      "loss": 2.5309,
+      "step": 3885
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.4957601726055145,
+      "learning_rate": 9.977899299677502e-05,
+      "loss": 2.5532,
+      "step": 3890
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.47477811574935913,
+      "learning_rate": 9.97784099115757e-05,
+      "loss": 2.4543,
+      "step": 3895
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.4598829448223114,
+      "learning_rate": 9.977782605991754e-05,
+      "loss": 2.5991,
+      "step": 3900
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.43407076597213745,
+      "learning_rate": 9.977724144180953e-05,
+      "loss": 2.5851,
+      "step": 3905
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.4776305854320526,
+      "learning_rate": 9.977665605726067e-05,
+      "loss": 2.5858,
+      "step": 3910
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.4813562035560608,
+      "learning_rate": 9.977606990628001e-05,
+      "loss": 2.4953,
+      "step": 3915
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.501685380935669,
+      "learning_rate": 9.977548298887651e-05,
+      "loss": 2.5379,
+      "step": 3920
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.5317994952201843,
+      "learning_rate": 9.977489530505927e-05,
+      "loss": 2.6061,
+      "step": 3925
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.464073121547699,
+      "learning_rate": 9.97743068548373e-05,
+      "loss": 2.4915,
+      "step": 3930
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.45964315533638,
+      "learning_rate": 9.977371763821966e-05,
+      "loss": 2.5693,
+      "step": 3935
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.4554794132709503,
+      "learning_rate": 9.977312765521545e-05,
+      "loss": 2.5694,
+      "step": 3940
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.4799802005290985,
+      "learning_rate": 9.977253690583372e-05,
+      "loss": 2.6912,
+      "step": 3945
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.4834944009780884,
+      "learning_rate": 9.97719453900836e-05,
+      "loss": 2.4888,
+      "step": 3950
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.49164193868637085,
+      "learning_rate": 9.977135310797418e-05,
+      "loss": 2.5657,
+      "step": 3955
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.5271838307380676,
+      "learning_rate": 9.977076005951458e-05,
+      "loss": 2.5442,
+      "step": 3960
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.46106627583503723,
+      "learning_rate": 9.977016624471392e-05,
+      "loss": 2.4729,
+      "step": 3965
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.47807806730270386,
+      "learning_rate": 9.976957166358137e-05,
+      "loss": 2.5319,
+      "step": 3970
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.519291877746582,
+      "learning_rate": 9.976897631612607e-05,
+      "loss": 2.5555,
+      "step": 3975
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.5070173740386963,
+      "learning_rate": 9.97683802023572e-05,
+      "loss": 2.5282,
+      "step": 3980
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.4580894410610199,
+      "learning_rate": 9.976778332228389e-05,
+      "loss": 2.5945,
+      "step": 3985
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.4650843143463135,
+      "learning_rate": 9.97671856759154e-05,
+      "loss": 2.4913,
+      "step": 3990
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.487366646528244,
+      "learning_rate": 9.976658726326088e-05,
+      "loss": 2.6009,
+      "step": 3995
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.4760417938232422,
+      "learning_rate": 9.976598808432957e-05,
+      "loss": 2.4696,
+      "step": 4000
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.48588499426841736,
+      "learning_rate": 9.976538813913068e-05,
+      "loss": 2.6268,
+      "step": 4005
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.4786835312843323,
+      "learning_rate": 9.976478742767347e-05,
+      "loss": 2.6407,
+      "step": 4010
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.4595915675163269,
+      "learning_rate": 9.976418594996716e-05,
+      "loss": 2.6107,
+      "step": 4015
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.47205695509910583,
+      "learning_rate": 9.976358370602103e-05,
+      "loss": 2.4108,
+      "step": 4020
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.48149943351745605,
+      "learning_rate": 9.976298069584438e-05,
+      "loss": 2.5454,
+      "step": 4025
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.5053077936172485,
+      "learning_rate": 9.976237691944643e-05,
+      "loss": 2.5325,
+      "step": 4030
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.5188475251197815,
+      "learning_rate": 9.976177237683653e-05,
+      "loss": 2.4412,
+      "step": 4035
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.4757572114467621,
+      "learning_rate": 9.976116706802395e-05,
+      "loss": 2.5245,
+      "step": 4040
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.5049698352813721,
+      "learning_rate": 9.976056099301805e-05,
+      "loss": 2.5162,
+      "step": 4045
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.47997766733169556,
+      "learning_rate": 9.975995415182812e-05,
+      "loss": 2.5139,
+      "step": 4050
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.5043210983276367,
+      "learning_rate": 9.975934654446355e-05,
+      "loss": 2.5359,
+      "step": 4055
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.44927978515625,
+      "learning_rate": 9.975873817093367e-05,
+      "loss": 2.5561,
+      "step": 4060
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.4671197533607483,
+      "learning_rate": 9.975812903124782e-05,
+      "loss": 2.5303,
+      "step": 4065
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.454410195350647,
+      "learning_rate": 9.975751912541544e-05,
+      "loss": 2.4615,
+      "step": 4070
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.46183016896247864,
+      "learning_rate": 9.975690845344588e-05,
+      "loss": 2.512,
+      "step": 4075
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.462402880191803,
+      "learning_rate": 9.975629701534854e-05,
+      "loss": 2.5319,
+      "step": 4080
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.45608630776405334,
+      "learning_rate": 9.975568481113285e-05,
+      "loss": 2.4627,
+      "step": 4085
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.4614071249961853,
+      "learning_rate": 9.975507184080823e-05,
+      "loss": 2.5057,
+      "step": 4090
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.45164063572883606,
+      "learning_rate": 9.975445810438413e-05,
+      "loss": 2.5663,
+      "step": 4095
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.517968475818634,
+      "learning_rate": 9.975384360186999e-05,
+      "loss": 2.5748,
+      "step": 4100
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.5130034685134888,
+      "learning_rate": 9.975322833327527e-05,
+      "loss": 2.478,
+      "step": 4105
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.48245057463645935,
+      "learning_rate": 9.975261229860944e-05,
+      "loss": 2.5002,
+      "step": 4110
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.47144201397895813,
+      "learning_rate": 9.975199549788199e-05,
+      "loss": 2.5435,
+      "step": 4115
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.4824945330619812,
+      "learning_rate": 9.975137793110243e-05,
+      "loss": 2.4997,
+      "step": 4120
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.649274468421936,
+      "learning_rate": 9.975075959828026e-05,
+      "loss": 2.5655,
+      "step": 4125
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.4770585596561432,
+      "learning_rate": 9.975014049942499e-05,
+      "loss": 2.5203,
+      "step": 4130
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.43077823519706726,
+      "learning_rate": 9.974952063454615e-05,
+      "loss": 2.5107,
+      "step": 4135
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.4886411130428314,
+      "learning_rate": 9.974890000365332e-05,
+      "loss": 2.5412,
+      "step": 4140
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.47059714794158936,
+      "learning_rate": 9.9748278606756e-05,
+      "loss": 2.4851,
+      "step": 4145
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.49272027611732483,
+      "learning_rate": 9.974765644386382e-05,
+      "loss": 2.5329,
+      "step": 4150
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.4853112995624542,
+      "learning_rate": 9.974703351498631e-05,
+      "loss": 2.454,
+      "step": 4155
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.47347402572631836,
+      "learning_rate": 9.974640982013307e-05,
+      "loss": 2.5374,
+      "step": 4160
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.49749258160591125,
+      "learning_rate": 9.974578535931374e-05,
+      "loss": 2.5386,
+      "step": 4165
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.4391794800758362,
+      "learning_rate": 9.974516013253789e-05,
+      "loss": 2.5384,
+      "step": 4170
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.43478500843048096,
+      "learning_rate": 9.974453413981517e-05,
+      "loss": 2.5376,
+      "step": 4175
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.4634556472301483,
+      "learning_rate": 9.974390738115521e-05,
+      "loss": 2.4806,
+      "step": 4180
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.46140480041503906,
+      "learning_rate": 9.974327985656767e-05,
+      "loss": 2.5929,
+      "step": 4185
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.47639527916908264,
+      "learning_rate": 9.974265156606221e-05,
+      "loss": 2.4915,
+      "step": 4190
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.45772939920425415,
+      "learning_rate": 9.974202250964849e-05,
+      "loss": 2.5077,
+      "step": 4195
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.44640785455703735,
+      "learning_rate": 9.974139268733623e-05,
+      "loss": 2.4984,
+      "step": 4200
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.48924025893211365,
+      "learning_rate": 9.974076209913508e-05,
+      "loss": 2.5701,
+      "step": 4205
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.4869600534439087,
+      "learning_rate": 9.974013074505477e-05,
+      "loss": 2.5267,
+      "step": 4210
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.5068817734718323,
+      "learning_rate": 9.973949862510503e-05,
+      "loss": 2.5441,
+      "step": 4215
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.4380756914615631,
+      "learning_rate": 9.97388657392956e-05,
+      "loss": 2.5928,
+      "step": 4220
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.4853883385658264,
+      "learning_rate": 9.97382320876362e-05,
+      "loss": 2.5766,
+      "step": 4225
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.604583203792572,
+      "learning_rate": 9.973759767013662e-05,
+      "loss": 2.5782,
+      "step": 4230
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.5039944648742676,
+      "learning_rate": 9.973696248680659e-05,
+      "loss": 2.5827,
+      "step": 4235
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.475847065448761,
+      "learning_rate": 9.97363265376559e-05,
+      "loss": 2.4654,
+      "step": 4240
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.4324930012226105,
+      "learning_rate": 9.973568982269436e-05,
+      "loss": 2.571,
+      "step": 4245
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.5124661922454834,
+      "learning_rate": 9.973505234193177e-05,
+      "loss": 2.5559,
+      "step": 4250
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.45139962434768677,
+      "learning_rate": 9.973441409537793e-05,
+      "loss": 2.571,
+      "step": 4255
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.4386439621448517,
+      "learning_rate": 9.973377508304269e-05,
+      "loss": 2.5064,
+      "step": 4260
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.45720911026000977,
+      "learning_rate": 9.973313530493586e-05,
+      "loss": 2.5379,
+      "step": 4265
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.44324880838394165,
+      "learning_rate": 9.973249476106732e-05,
+      "loss": 2.5414,
+      "step": 4270
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.44313210248947144,
+      "learning_rate": 9.97318534514469e-05,
+      "loss": 2.4897,
+      "step": 4275
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.45094582438468933,
+      "learning_rate": 9.97312113760845e-05,
+      "loss": 2.5996,
+      "step": 4280
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.43577611446380615,
+      "learning_rate": 9.973056853499001e-05,
+      "loss": 2.5345,
+      "step": 4285
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.44543084502220154,
+      "learning_rate": 9.972992492817331e-05,
+      "loss": 2.5523,
+      "step": 4290
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.4770336449146271,
+      "learning_rate": 9.972928055564432e-05,
+      "loss": 2.5464,
+      "step": 4295
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.47398802638053894,
+      "learning_rate": 9.972863541741297e-05,
+      "loss": 2.5602,
+      "step": 4300
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.43326032161712646,
+      "learning_rate": 9.972798951348918e-05,
+      "loss": 2.5212,
+      "step": 4305
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.442023903131485,
+      "learning_rate": 9.972734284388288e-05,
+      "loss": 2.5522,
+      "step": 4310
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.4617267847061157,
+      "learning_rate": 9.972669540860406e-05,
+      "loss": 2.4947,
+      "step": 4315
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.47267743945121765,
+      "learning_rate": 9.972604720766268e-05,
+      "loss": 2.472,
+      "step": 4320
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.43336886167526245,
+      "learning_rate": 9.97253982410687e-05,
+      "loss": 2.5701,
+      "step": 4325
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.43035662174224854,
+      "learning_rate": 9.972474850883213e-05,
+      "loss": 2.5535,
+      "step": 4330
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.44428300857543945,
+      "learning_rate": 9.972409801096299e-05,
+      "loss": 2.5708,
+      "step": 4335
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.4433332681655884,
+      "learning_rate": 9.972344674747125e-05,
+      "loss": 2.565,
+      "step": 4340
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.45809710025787354,
+      "learning_rate": 9.972279471836698e-05,
+      "loss": 2.5269,
+      "step": 4345
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.46873006224632263,
+      "learning_rate": 9.972214192366021e-05,
+      "loss": 2.5212,
+      "step": 4350
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.49696722626686096,
+      "learning_rate": 9.972148836336098e-05,
+      "loss": 2.4963,
+      "step": 4355
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.4733389914035797,
+      "learning_rate": 9.972083403747937e-05,
+      "loss": 2.4253,
+      "step": 4360
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.4621347486972809,
+      "learning_rate": 9.972017894602542e-05,
+      "loss": 2.538,
+      "step": 4365
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.4704989492893219,
+      "learning_rate": 9.971952308900925e-05,
+      "loss": 2.5106,
+      "step": 4370
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.4812776446342468,
+      "learning_rate": 9.971886646644096e-05,
+      "loss": 2.4696,
+      "step": 4375
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.48963868618011475,
+      "learning_rate": 9.971820907833065e-05,
+      "loss": 2.5989,
+      "step": 4380
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.47064048051834106,
+      "learning_rate": 9.971755092468843e-05,
+      "loss": 2.5932,
+      "step": 4385
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.4493527412414551,
+      "learning_rate": 9.971689200552445e-05,
+      "loss": 2.5636,
+      "step": 4390
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.47781190276145935,
+      "learning_rate": 9.971623232084886e-05,
+      "loss": 2.4937,
+      "step": 4395
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.49653780460357666,
+      "learning_rate": 9.971557187067179e-05,
+      "loss": 2.531,
+      "step": 4400
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.45908838510513306,
+      "learning_rate": 9.971491065500345e-05,
+      "loss": 2.5154,
+      "step": 4405
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.5001177787780762,
+      "learning_rate": 9.9714248673854e-05,
+      "loss": 2.4121,
+      "step": 4410
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.4498959183692932,
+      "learning_rate": 9.971358592723361e-05,
+      "loss": 2.5122,
+      "step": 4415
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.40706178545951843,
+      "learning_rate": 9.97129224151525e-05,
+      "loss": 2.5478,
+      "step": 4420
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.49605751037597656,
+      "learning_rate": 9.971225813762094e-05,
+      "loss": 2.542,
+      "step": 4425
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.44895920157432556,
+      "learning_rate": 9.971159309464906e-05,
+      "loss": 2.5112,
+      "step": 4430
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4526841342449188,
+      "learning_rate": 9.971092728624718e-05,
+      "loss": 2.4521,
+      "step": 4435
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4661194980144501,
+      "learning_rate": 9.97102607124255e-05,
+      "loss": 2.504,
+      "step": 4440
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4578354060649872,
+      "learning_rate": 9.970959337319433e-05,
+      "loss": 2.3597,
+      "step": 4445
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.44570937752723694,
+      "learning_rate": 9.970892526856392e-05,
+      "loss": 2.531,
+      "step": 4450
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4569240212440491,
+      "learning_rate": 9.970825639854454e-05,
+      "loss": 2.5338,
+      "step": 4455
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.46618130803108215,
+      "learning_rate": 9.970758676314652e-05,
+      "loss": 2.4765,
+      "step": 4460
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.45674726366996765,
+      "learning_rate": 9.970691636238017e-05,
+      "loss": 2.4379,
+      "step": 4465
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4863596558570862,
+      "learning_rate": 9.970624519625578e-05,
+      "loss": 2.5136,
+      "step": 4470
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4823973476886749,
+      "learning_rate": 9.970557326478371e-05,
+      "loss": 2.5168,
+      "step": 4475
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4536812901496887,
+      "learning_rate": 9.97049005679743e-05,
+      "loss": 2.5336,
+      "step": 4480
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.46327781677246094,
+      "learning_rate": 9.970422710583791e-05,
+      "loss": 2.64,
+      "step": 4485
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4647650718688965,
+      "learning_rate": 9.970355287838491e-05,
+      "loss": 2.5051,
+      "step": 4490
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4536067843437195,
+      "learning_rate": 9.970287788562567e-05,
+      "loss": 2.4354,
+      "step": 4495
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4543205499649048,
+      "learning_rate": 9.97022021275706e-05,
+      "loss": 2.4899,
+      "step": 4500
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4649229645729065,
+      "learning_rate": 9.97015256042301e-05,
+      "loss": 2.5793,
+      "step": 4505
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.504259467124939,
+      "learning_rate": 9.97008483156146e-05,
+      "loss": 2.5097,
+      "step": 4510
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.453439861536026,
+      "learning_rate": 9.970017026173448e-05,
+      "loss": 2.4802,
+      "step": 4515
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.43970581889152527,
+      "learning_rate": 9.969949144260022e-05,
+      "loss": 2.5211,
+      "step": 4520
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.5128332376480103,
+      "learning_rate": 9.969881185822227e-05,
+      "loss": 2.5781,
+      "step": 4525
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4392908215522766,
+      "learning_rate": 9.969813150861111e-05,
+      "loss": 2.5462,
+      "step": 4530
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4601394832134247,
+      "learning_rate": 9.969745039377717e-05,
+      "loss": 2.4359,
+      "step": 4535
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4381711184978485,
+      "learning_rate": 9.969676851373098e-05,
+      "loss": 2.4926,
+      "step": 4540
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4565121531486511,
+      "learning_rate": 9.969608586848301e-05,
+      "loss": 2.5201,
+      "step": 4545
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4355619549751282,
+      "learning_rate": 9.969540245804377e-05,
+      "loss": 2.5117,
+      "step": 4550
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4669399857521057,
+      "learning_rate": 9.96947182824238e-05,
+      "loss": 2.5375,
+      "step": 4555
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.43494799733161926,
+      "learning_rate": 9.969403334163364e-05,
+      "loss": 2.5919,
+      "step": 4560
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4281655550003052,
+      "learning_rate": 9.969334763568383e-05,
+      "loss": 2.4447,
+      "step": 4565
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4413806200027466,
+      "learning_rate": 9.969266116458491e-05,
+      "loss": 2.4616,
+      "step": 4570
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4451961815357208,
+      "learning_rate": 9.969197392834747e-05,
+      "loss": 2.5491,
+      "step": 4575
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4472174346446991,
+      "learning_rate": 9.969128592698207e-05,
+      "loss": 2.5199,
+      "step": 4580
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4744277596473694,
+      "learning_rate": 9.969059716049934e-05,
+      "loss": 2.5816,
+      "step": 4585
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4952501952648163,
+      "learning_rate": 9.968990762890986e-05,
+      "loss": 2.5096,
+      "step": 4590
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.46614351868629456,
+      "learning_rate": 9.968921733222425e-05,
+      "loss": 2.5472,
+      "step": 4595
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.45616060495376587,
+      "learning_rate": 9.968852627045313e-05,
+      "loss": 2.4843,
+      "step": 4600
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.43455639481544495,
+      "learning_rate": 9.968783444360716e-05,
+      "loss": 2.5979,
+      "step": 4605
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4234558343887329,
+      "learning_rate": 9.968714185169698e-05,
+      "loss": 2.4895,
+      "step": 4610
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.45572859048843384,
+      "learning_rate": 9.968644849473326e-05,
+      "loss": 2.5415,
+      "step": 4615
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.43400368094444275,
+      "learning_rate": 9.968575437272667e-05,
+      "loss": 2.5779,
+      "step": 4620
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4425300657749176,
+      "learning_rate": 9.968505948568792e-05,
+      "loss": 2.4951,
+      "step": 4625
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4527803063392639,
+      "learning_rate": 9.968436383362766e-05,
+      "loss": 2.4714,
+      "step": 4630
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.419965922832489,
+      "learning_rate": 9.968366741655663e-05,
+      "loss": 2.4867,
+      "step": 4635
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.42279112339019775,
+      "learning_rate": 9.968297023448558e-05,
+      "loss": 2.4475,
+      "step": 4640
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4358624517917633,
+      "learning_rate": 9.96822722874252e-05,
+      "loss": 2.5331,
+      "step": 4645
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4550338089466095,
+      "learning_rate": 9.968157357538625e-05,
+      "loss": 2.5853,
+      "step": 4650
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.46228790283203125,
+      "learning_rate": 9.968087409837951e-05,
+      "loss": 2.4637,
+      "step": 4655
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.43807750940322876,
+      "learning_rate": 9.968017385641574e-05,
+      "loss": 2.5316,
+      "step": 4660
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.5151635408401489,
+      "learning_rate": 9.967947284950569e-05,
+      "loss": 2.5505,
+      "step": 4665
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4928719997406006,
+      "learning_rate": 9.967877107766021e-05,
+      "loss": 2.5272,
+      "step": 4670
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.46146634221076965,
+      "learning_rate": 9.967806854089006e-05,
+      "loss": 2.5287,
+      "step": 4675
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.44228842854499817,
+      "learning_rate": 9.967736523920607e-05,
+      "loss": 2.465,
+      "step": 4680
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.44120651483535767,
+      "learning_rate": 9.967666117261907e-05,
+      "loss": 2.4611,
+      "step": 4685
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4105578064918518,
+      "learning_rate": 9.96759563411399e-05,
+      "loss": 2.4622,
+      "step": 4690
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.43966078758239746,
+      "learning_rate": 9.967525074477943e-05,
+      "loss": 2.5068,
+      "step": 4695
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4567209482192993,
+      "learning_rate": 9.96745443835485e-05,
+      "loss": 2.4781,
+      "step": 4700
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.44286397099494934,
+      "learning_rate": 9.967383725745799e-05,
+      "loss": 2.4874,
+      "step": 4705
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4572209119796753,
+      "learning_rate": 9.96731293665188e-05,
+      "loss": 2.5237,
+      "step": 4710
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4515369236469269,
+      "learning_rate": 9.967242071074181e-05,
+      "loss": 2.4713,
+      "step": 4715
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4354775846004486,
+      "learning_rate": 9.967171129013797e-05,
+      "loss": 2.4889,
+      "step": 4720
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.47701355814933777,
+      "learning_rate": 9.967100110471817e-05,
+      "loss": 2.5465,
+      "step": 4725
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4447549283504486,
+      "learning_rate": 9.967029015449334e-05,
+      "loss": 2.4459,
+      "step": 4730
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.47182697057724,
+      "learning_rate": 9.966957843947444e-05,
+      "loss": 2.5468,
+      "step": 4735
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4246272146701813,
+      "learning_rate": 9.966886595967243e-05,
+      "loss": 2.5012,
+      "step": 4740
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4161452054977417,
+      "learning_rate": 9.966815271509828e-05,
+      "loss": 2.4519,
+      "step": 4745
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.44114211201667786,
+      "learning_rate": 9.966743870576297e-05,
+      "loss": 2.415,
+      "step": 4750
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.45395398139953613,
+      "learning_rate": 9.96667239316775e-05,
+      "loss": 2.4512,
+      "step": 4755
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.46382543444633484,
+      "learning_rate": 9.966600839285286e-05,
+      "loss": 2.4976,
+      "step": 4760
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4465494453907013,
+      "learning_rate": 9.966529208930008e-05,
+      "loss": 2.5748,
+      "step": 4765
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.5029381513595581,
+      "learning_rate": 9.96645750210302e-05,
+      "loss": 2.5156,
+      "step": 4770
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.48751044273376465,
+      "learning_rate": 9.966385718805423e-05,
+      "loss": 2.4854,
+      "step": 4775
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4538024365901947,
+      "learning_rate": 9.966313859038325e-05,
+      "loss": 2.4869,
+      "step": 4780
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4781196713447571,
+      "learning_rate": 9.966241922802832e-05,
+      "loss": 2.5513,
+      "step": 4785
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.45704883337020874,
+      "learning_rate": 9.96616991010005e-05,
+      "loss": 2.5964,
+      "step": 4790
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.43300169706344604,
+      "learning_rate": 9.96609782093109e-05,
+      "loss": 2.3514,
+      "step": 4795
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.45379823446273804,
+      "learning_rate": 9.96602565529706e-05,
+      "loss": 2.504,
+      "step": 4800
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4410962760448456,
+      "learning_rate": 9.965953413199071e-05,
+      "loss": 2.4429,
+      "step": 4805
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4375450909137726,
+      "learning_rate": 9.965881094638238e-05,
+      "loss": 2.4481,
+      "step": 4810
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.43936091661453247,
+      "learning_rate": 9.965808699615674e-05,
+      "loss": 2.5078,
+      "step": 4815
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.44300732016563416,
+      "learning_rate": 9.965736228132492e-05,
+      "loss": 2.4817,
+      "step": 4820
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.48639142513275146,
+      "learning_rate": 9.965663680189807e-05,
+      "loss": 2.4583,
+      "step": 4825
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4252323508262634,
+      "learning_rate": 9.965591055788738e-05,
+      "loss": 2.6091,
+      "step": 4830
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.49349576234817505,
+      "learning_rate": 9.965518354930405e-05,
+      "loss": 2.5061,
+      "step": 4835
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4374244213104248,
+      "learning_rate": 9.965445577615922e-05,
+      "loss": 2.4908,
+      "step": 4840
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.43704456090927124,
+      "learning_rate": 9.965372723846415e-05,
+      "loss": 2.4821,
+      "step": 4845
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4262220859527588,
+      "learning_rate": 9.965299793623003e-05,
+      "loss": 2.5163,
+      "step": 4850
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.437102347612381,
+      "learning_rate": 9.96522678694681e-05,
+      "loss": 2.4932,
+      "step": 4855
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4956406056880951,
+      "learning_rate": 9.965153703818959e-05,
+      "loss": 2.5438,
+      "step": 4860
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.44515368342399597,
+      "learning_rate": 9.965080544240575e-05,
+      "loss": 2.5271,
+      "step": 4865
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.44978493452072144,
+      "learning_rate": 9.965007308212786e-05,
+      "loss": 2.473,
+      "step": 4870
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4695723354816437,
+      "learning_rate": 9.964933995736721e-05,
+      "loss": 2.4263,
+      "step": 4875
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4618473947048187,
+      "learning_rate": 9.964860606813504e-05,
+      "loss": 2.5402,
+      "step": 4880
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4827546179294586,
+      "learning_rate": 9.96478714144427e-05,
+      "loss": 2.4072,
+      "step": 4885
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.49534738063812256,
+      "learning_rate": 9.964713599630146e-05,
+      "loss": 2.5767,
+      "step": 4890
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4428688585758209,
+      "learning_rate": 9.964639981372267e-05,
+      "loss": 2.4725,
+      "step": 4895
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4277462065219879,
+      "learning_rate": 9.964566286671766e-05,
+      "loss": 2.4316,
+      "step": 4900
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4413623511791229,
+      "learning_rate": 9.964492515529777e-05,
+      "loss": 2.5272,
+      "step": 4905
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.43058761954307556,
+      "learning_rate": 9.964418667947438e-05,
+      "loss": 2.5358,
+      "step": 4910
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.42738527059555054,
+      "learning_rate": 9.964344743925884e-05,
+      "loss": 2.4246,
+      "step": 4915
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.41982728242874146,
+      "learning_rate": 9.964270743466253e-05,
+      "loss": 2.4348,
+      "step": 4920
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4605098366737366,
+      "learning_rate": 9.964196666569686e-05,
+      "loss": 2.4733,
+      "step": 4925
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.44064024090766907,
+      "learning_rate": 9.964122513237322e-05,
+      "loss": 2.4409,
+      "step": 4930
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.42144641280174255,
+      "learning_rate": 9.964048283470305e-05,
+      "loss": 2.4465,
+      "step": 4935
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.44797492027282715,
+      "learning_rate": 9.963973977269776e-05,
+      "loss": 2.5566,
+      "step": 4940
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4254441261291504,
+      "learning_rate": 9.963899594636879e-05,
+      "loss": 2.4399,
+      "step": 4945
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4311037063598633,
+      "learning_rate": 9.96382513557276e-05,
+      "loss": 2.4599,
+      "step": 4950
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.40651383996009827,
+      "learning_rate": 9.963750600078565e-05,
+      "loss": 2.4102,
+      "step": 4955
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4412480890750885,
+      "learning_rate": 9.963675988155443e-05,
+      "loss": 2.4262,
+      "step": 4960
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4319073259830475,
+      "learning_rate": 9.963601299804541e-05,
+      "loss": 2.553,
+      "step": 4965
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4335537254810333,
+      "learning_rate": 9.963526535027012e-05,
+      "loss": 2.4931,
+      "step": 4970
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4274340271949768,
+      "learning_rate": 9.963451693824003e-05,
+      "loss": 2.5763,
+      "step": 4975
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4414899945259094,
+      "learning_rate": 9.963376776196669e-05,
+      "loss": 2.5037,
+      "step": 4980
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.45396876335144043,
+      "learning_rate": 9.963301782146163e-05,
+      "loss": 2.4819,
+      "step": 4985
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.44958677887916565,
+      "learning_rate": 9.96322671167364e-05,
+      "loss": 2.4371,
+      "step": 4990
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4888280928134918,
+      "learning_rate": 9.963151564780256e-05,
+      "loss": 2.524,
+      "step": 4995
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.45802339911460876,
+      "learning_rate": 9.963076341467165e-05,
+      "loss": 2.414,
+      "step": 5000
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.44857603311538696,
+      "learning_rate": 9.963001041735531e-05,
+      "loss": 2.5415,
+      "step": 5005
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4604758620262146,
+      "learning_rate": 9.962925665586509e-05,
+      "loss": 2.5088,
+      "step": 5010
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4790046513080597,
+      "learning_rate": 9.962850213021262e-05,
+      "loss": 2.4515,
+      "step": 5015
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.41994744539260864,
+      "learning_rate": 9.96277468404095e-05,
+      "loss": 2.4301,
+      "step": 5020
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.44513747096061707,
+      "learning_rate": 9.962699078646735e-05,
+      "loss": 2.4348,
+      "step": 5025
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4326728880405426,
+      "learning_rate": 9.962623396839785e-05,
+      "loss": 2.5035,
+      "step": 5030
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4415475130081177,
+      "learning_rate": 9.962547638621262e-05,
+      "loss": 2.4962,
+      "step": 5035
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4331226646900177,
+      "learning_rate": 9.962471803992334e-05,
+      "loss": 2.4809,
+      "step": 5040
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4588994085788727,
+      "learning_rate": 9.962395892954168e-05,
+      "loss": 2.4818,
+      "step": 5045
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.41550108790397644,
+      "learning_rate": 9.962319905507934e-05,
+      "loss": 2.4006,
+      "step": 5050
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4214065968990326,
+      "learning_rate": 9.9622438416548e-05,
+      "loss": 2.3816,
+      "step": 5055
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4540846645832062,
+      "learning_rate": 9.962167701395937e-05,
+      "loss": 2.4999,
+      "step": 5060
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.43333888053894043,
+      "learning_rate": 9.962091484732521e-05,
+      "loss": 2.5064,
+      "step": 5065
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.43932488560676575,
+      "learning_rate": 9.962015191665722e-05,
+      "loss": 2.4509,
+      "step": 5070
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.42549943923950195,
+      "learning_rate": 9.961938822196716e-05,
+      "loss": 2.3786,
+      "step": 5075
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.45315173268318176,
+      "learning_rate": 9.961862376326678e-05,
+      "loss": 2.5381,
+      "step": 5080
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4542273283004761,
+      "learning_rate": 9.961785854056788e-05,
+      "loss": 2.5125,
+      "step": 5085
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.42474988102912903,
+      "learning_rate": 9.961709255388221e-05,
+      "loss": 2.5013,
+      "step": 5090
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.44795501232147217,
+      "learning_rate": 9.961632580322156e-05,
+      "loss": 2.405,
+      "step": 5095
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.5000123381614685,
+      "learning_rate": 9.961555828859777e-05,
+      "loss": 2.411,
+      "step": 5100
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.45349782705307007,
+      "learning_rate": 9.961479001002264e-05,
+      "loss": 2.4932,
+      "step": 5105
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4172985255718231,
+      "learning_rate": 9.961402096750799e-05,
+      "loss": 2.528,
+      "step": 5110
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4375399053096771,
+      "learning_rate": 9.961325116106566e-05,
+      "loss": 2.4639,
+      "step": 5115
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.42545285820961,
+      "learning_rate": 9.961248059070751e-05,
+      "loss": 2.4927,
+      "step": 5120
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.44824379682540894,
+      "learning_rate": 9.961170925644542e-05,
+      "loss": 2.429,
+      "step": 5125
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.45941293239593506,
+      "learning_rate": 9.961093715829127e-05,
+      "loss": 2.4438,
+      "step": 5130
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.42795634269714355,
+      "learning_rate": 9.961016429625691e-05,
+      "loss": 2.4855,
+      "step": 5135
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4392828941345215,
+      "learning_rate": 9.960939067035426e-05,
+      "loss": 2.4267,
+      "step": 5140
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4812280833721161,
+      "learning_rate": 9.960861628059524e-05,
+      "loss": 2.4549,
+      "step": 5145
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4294860064983368,
+      "learning_rate": 9.960784112699177e-05,
+      "loss": 2.4214,
+      "step": 5150
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4709104299545288,
+      "learning_rate": 9.960706520955578e-05,
+      "loss": 2.4051,
+      "step": 5155
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4749593734741211,
+      "learning_rate": 9.960628852829923e-05,
+      "loss": 2.4838,
+      "step": 5160
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.44298142194747925,
+      "learning_rate": 9.960551108323406e-05,
+      "loss": 2.4432,
+      "step": 5165
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4405077397823334,
+      "learning_rate": 9.960473287437224e-05,
+      "loss": 2.5234,
+      "step": 5170
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.414421945810318,
+      "learning_rate": 9.960395390172579e-05,
+      "loss": 2.4858,
+      "step": 5175
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.40656471252441406,
+      "learning_rate": 9.960317416530665e-05,
+      "loss": 2.4218,
+      "step": 5180
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.40942251682281494,
+      "learning_rate": 9.960239366512686e-05,
+      "loss": 2.3549,
+      "step": 5185
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.43964359164237976,
+      "learning_rate": 9.960161240119843e-05,
+      "loss": 2.5877,
+      "step": 5190
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4252696633338928,
+      "learning_rate": 9.96008303735334e-05,
+      "loss": 2.4253,
+      "step": 5195
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4307929277420044,
+      "learning_rate": 9.960004758214379e-05,
+      "loss": 2.485,
+      "step": 5200
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.44725725054740906,
+      "learning_rate": 9.959926402704165e-05,
+      "loss": 2.4591,
+      "step": 5205
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.42361149191856384,
+      "learning_rate": 9.959847970823908e-05,
+      "loss": 2.4643,
+      "step": 5210
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.41677817702293396,
+      "learning_rate": 9.959769462574813e-05,
+      "loss": 2.3506,
+      "step": 5215
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4044424295425415,
+      "learning_rate": 9.959690877958088e-05,
+      "loss": 2.4891,
+      "step": 5220
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.45966875553131104,
+      "learning_rate": 9.959612216974946e-05,
+      "loss": 2.422,
+      "step": 5225
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4503769874572754,
+      "learning_rate": 9.959533479626596e-05,
+      "loss": 2.4083,
+      "step": 5230
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7349979877471924,
+      "learning_rate": 9.95945466591425e-05,
+      "loss": 2.4249,
+      "step": 5235
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.47178733348846436,
+      "learning_rate": 9.959375775839124e-05,
+      "loss": 2.4435,
+      "step": 5240
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4602421522140503,
+      "learning_rate": 9.95929680940243e-05,
+      "loss": 2.4793,
+      "step": 5245
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.44540679454803467,
+      "learning_rate": 9.959217766605384e-05,
+      "loss": 2.4893,
+      "step": 5250
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.45927610993385315,
+      "learning_rate": 9.959138647449205e-05,
+      "loss": 2.5357,
+      "step": 5255
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4544280767440796,
+      "learning_rate": 9.95905945193511e-05,
+      "loss": 2.4293,
+      "step": 5260
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4383193254470825,
+      "learning_rate": 9.958980180064318e-05,
+      "loss": 2.4032,
+      "step": 5265
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.42129093408584595,
+      "learning_rate": 9.958900831838051e-05,
+      "loss": 2.5102,
+      "step": 5270
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4119527339935303,
+      "learning_rate": 9.958821407257531e-05,
+      "loss": 2.5696,
+      "step": 5275
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.41305088996887207,
+      "learning_rate": 9.958741906323978e-05,
+      "loss": 2.5313,
+      "step": 5280
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4099360406398773,
+      "learning_rate": 9.95866232903862e-05,
+      "loss": 2.4233,
+      "step": 5285
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4101997911930084,
+      "learning_rate": 9.958582675402677e-05,
+      "loss": 2.4358,
+      "step": 5290
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.41214224696159363,
+      "learning_rate": 9.958502945417382e-05,
+      "loss": 2.4911,
+      "step": 5295
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.46936702728271484,
+      "learning_rate": 9.958423139083958e-05,
+      "loss": 2.4056,
+      "step": 5300
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.451890230178833,
+      "learning_rate": 9.958343256403635e-05,
+      "loss": 2.3883,
+      "step": 5305
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.44083547592163086,
+      "learning_rate": 9.958263297377643e-05,
+      "loss": 2.5052,
+      "step": 5310
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4078104794025421,
+      "learning_rate": 9.958183262007215e-05,
+      "loss": 2.4494,
+      "step": 5315
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4414960443973541,
+      "learning_rate": 9.95810315029358e-05,
+      "loss": 2.521,
+      "step": 5320
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.46115073561668396,
+      "learning_rate": 9.958022962237974e-05,
+      "loss": 2.3937,
+      "step": 5325
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4203258156776428,
+      "learning_rate": 9.957942697841632e-05,
+      "loss": 2.4833,
+      "step": 5330
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.43400266766548157,
+      "learning_rate": 9.957862357105785e-05,
+      "loss": 2.4322,
+      "step": 5335
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.46329954266548157,
+      "learning_rate": 9.957781940031676e-05,
+      "loss": 2.4742,
+      "step": 5340
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.41721540689468384,
+      "learning_rate": 9.957701446620543e-05,
+      "loss": 2.5291,
+      "step": 5345
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.49907243251800537,
+      "learning_rate": 9.957620876873621e-05,
+      "loss": 2.5473,
+      "step": 5350
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.42879197001457214,
+      "learning_rate": 9.957540230792152e-05,
+      "loss": 2.4576,
+      "step": 5355
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.42629191279411316,
+      "learning_rate": 9.957459508377381e-05,
+      "loss": 2.4772,
+      "step": 5360
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4608117640018463,
+      "learning_rate": 9.957378709630548e-05,
+      "loss": 2.5036,
+      "step": 5365
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4303702712059021,
+      "learning_rate": 9.957297834552898e-05,
+      "loss": 2.5248,
+      "step": 5370
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4434306025505066,
+      "learning_rate": 9.957216883145675e-05,
+      "loss": 2.4277,
+      "step": 5375
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.41439974308013916,
+      "learning_rate": 9.957135855410127e-05,
+      "loss": 2.4714,
+      "step": 5380
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.6444301009178162,
+      "learning_rate": 9.957054751347501e-05,
+      "loss": 2.3253,
+      "step": 5385
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.41642576456069946,
+      "learning_rate": 9.956973570959045e-05,
+      "loss": 2.4295,
+      "step": 5390
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.44771572947502136,
+      "learning_rate": 9.956892314246012e-05,
+      "loss": 2.4824,
+      "step": 5395
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.42052626609802246,
+      "learning_rate": 9.956810981209647e-05,
+      "loss": 2.4392,
+      "step": 5400
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4282350242137909,
+      "learning_rate": 9.956729571851211e-05,
+      "loss": 2.437,
+      "step": 5405
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.42310643196105957,
+      "learning_rate": 9.956648086171949e-05,
+      "loss": 2.4176,
+      "step": 5410
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4298139810562134,
+      "learning_rate": 9.956566524173121e-05,
+      "loss": 2.4975,
+      "step": 5415
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4726055860519409,
+      "learning_rate": 9.956484885855982e-05,
+      "loss": 2.4147,
+      "step": 5420
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.42097780108451843,
+      "learning_rate": 9.956403171221786e-05,
+      "loss": 2.3778,
+      "step": 5425
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4524746239185333,
+      "learning_rate": 9.956321380271796e-05,
+      "loss": 2.4631,
+      "step": 5430
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.45847946405410767,
+      "learning_rate": 9.956239513007268e-05,
+      "loss": 2.3981,
+      "step": 5435
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.41949009895324707,
+      "learning_rate": 9.956157569429464e-05,
+      "loss": 2.4404,
+      "step": 5440
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.446104496717453,
+      "learning_rate": 9.956075549539643e-05,
+      "loss": 2.4617,
+      "step": 5445
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.42756345868110657,
+      "learning_rate": 9.955993453339072e-05,
+      "loss": 2.4116,
+      "step": 5450
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.43159589171409607,
+      "learning_rate": 9.955911280829011e-05,
+      "loss": 2.4664,
+      "step": 5455
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4001917541027069,
+      "learning_rate": 9.95582903201073e-05,
+      "loss": 2.4739,
+      "step": 5460
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4200209081172943,
+      "learning_rate": 9.955746706885492e-05,
+      "loss": 2.5512,
+      "step": 5465
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4217223823070526,
+      "learning_rate": 9.955664305454564e-05,
+      "loss": 2.4634,
+      "step": 5470
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.43911516666412354,
+      "learning_rate": 9.955581827719216e-05,
+      "loss": 2.4224,
+      "step": 5475
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.43323951959609985,
+      "learning_rate": 9.955499273680719e-05,
+      "loss": 2.5861,
+      "step": 5480
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4636263847351074,
+      "learning_rate": 9.955416643340343e-05,
+      "loss": 2.5279,
+      "step": 5485
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4158070981502533,
+      "learning_rate": 9.95533393669936e-05,
+      "loss": 2.4634,
+      "step": 5490
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4155903458595276,
+      "learning_rate": 9.955251153759044e-05,
+      "loss": 2.4806,
+      "step": 5495
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.40825024247169495,
+      "learning_rate": 9.955168294520672e-05,
+      "loss": 2.4199,
+      "step": 5500
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4426273703575134,
+      "learning_rate": 9.955085358985514e-05,
+      "loss": 2.4628,
+      "step": 5505
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.3998028337955475,
+      "learning_rate": 9.955002347154851e-05,
+      "loss": 2.4424,
+      "step": 5510
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.40560078620910645,
+      "learning_rate": 9.954919259029962e-05,
+      "loss": 2.5405,
+      "step": 5515
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.45454463362693787,
+      "learning_rate": 9.954836094612123e-05,
+      "loss": 2.3914,
+      "step": 5520
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.44155600666999817,
+      "learning_rate": 9.954752853902619e-05,
+      "loss": 2.528,
+      "step": 5525
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4311385154724121,
+      "learning_rate": 9.954669536902728e-05,
+      "loss": 2.4586,
+      "step": 5530
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4565611481666565,
+      "learning_rate": 9.954586143613733e-05,
+      "loss": 2.5658,
+      "step": 5535
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.41128668189048767,
+      "learning_rate": 9.954502674036919e-05,
+      "loss": 2.4129,
+      "step": 5540
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4020937979221344,
+      "learning_rate": 9.954419128173572e-05,
+      "loss": 2.445,
+      "step": 5545
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4011147916316986,
+      "learning_rate": 9.954335506024978e-05,
+      "loss": 2.4607,
+      "step": 5550
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.41271454095840454,
+      "learning_rate": 9.954251807592423e-05,
+      "loss": 2.4064,
+      "step": 5555
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4463873505592346,
+      "learning_rate": 9.954168032877198e-05,
+      "loss": 2.4489,
+      "step": 5560
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.440402626991272,
+      "learning_rate": 9.95408418188059e-05,
+      "loss": 2.2969,
+      "step": 5565
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.41085284948349,
+      "learning_rate": 9.954000254603894e-05,
+      "loss": 2.4374,
+      "step": 5570
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4290235638618469,
+      "learning_rate": 9.9539162510484e-05,
+      "loss": 2.5156,
+      "step": 5575
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.42404285073280334,
+      "learning_rate": 9.953832171215401e-05,
+      "loss": 2.4732,
+      "step": 5580
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4279269576072693,
+      "learning_rate": 9.953748015106193e-05,
+      "loss": 2.5326,
+      "step": 5585
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4480297565460205,
+      "learning_rate": 9.953663782722069e-05,
+      "loss": 2.4711,
+      "step": 5590
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4213131070137024,
+      "learning_rate": 9.953579474064329e-05,
+      "loss": 2.3843,
+      "step": 5595
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.41238516569137573,
+      "learning_rate": 9.95349508913427e-05,
+      "loss": 2.5374,
+      "step": 5600
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.43080344796180725,
+      "learning_rate": 9.953410627933193e-05,
+      "loss": 2.4781,
+      "step": 5605
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4137932062149048,
+      "learning_rate": 9.953326090462397e-05,
+      "loss": 2.4211,
+      "step": 5610
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4768308997154236,
+      "learning_rate": 9.953241476723181e-05,
+      "loss": 2.485,
+      "step": 5615
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.44125398993492126,
+      "learning_rate": 9.953156786716851e-05,
+      "loss": 2.4905,
+      "step": 5620
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.42400476336479187,
+      "learning_rate": 9.953072020444711e-05,
+      "loss": 2.4702,
+      "step": 5625
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4208869934082031,
+      "learning_rate": 9.952987177908066e-05,
+      "loss": 2.5311,
+      "step": 5630
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4473826587200165,
+      "learning_rate": 9.95290225910822e-05,
+      "loss": 2.4464,
+      "step": 5635
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.3956182301044464,
+      "learning_rate": 9.952817264046486e-05,
+      "loss": 2.4659,
+      "step": 5640
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4229715168476105,
+      "learning_rate": 9.952732192724167e-05,
+      "loss": 2.4123,
+      "step": 5645
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4170120060443878,
+      "learning_rate": 9.952647045142576e-05,
+      "loss": 2.4577,
+      "step": 5650
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4092100262641907,
+      "learning_rate": 9.952561821303021e-05,
+      "loss": 2.4528,
+      "step": 5655
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.42861345410346985,
+      "learning_rate": 9.952476521206818e-05,
+      "loss": 2.4232,
+      "step": 5660
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.45241764187812805,
+      "learning_rate": 9.952391144855278e-05,
+      "loss": 2.484,
+      "step": 5665
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.3957231044769287,
+      "learning_rate": 9.952305692249717e-05,
+      "loss": 2.4413,
+      "step": 5670
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.43648090958595276,
+      "learning_rate": 9.952220163391451e-05,
+      "loss": 2.3679,
+      "step": 5675
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.43041202425956726,
+      "learning_rate": 9.952134558281793e-05,
+      "loss": 2.4488,
+      "step": 5680
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.40966150164604187,
+      "learning_rate": 9.952048876922066e-05,
+      "loss": 2.4432,
+      "step": 5685
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4050423800945282,
+      "learning_rate": 9.951963119313588e-05,
+      "loss": 2.4705,
+      "step": 5690
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.40170592069625854,
+      "learning_rate": 9.951877285457678e-05,
+      "loss": 2.3953,
+      "step": 5695
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.41050970554351807,
+      "learning_rate": 9.951791375355659e-05,
+      "loss": 2.408,
+      "step": 5700
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.40709617733955383,
+      "learning_rate": 9.951705389008852e-05,
+      "loss": 2.4033,
+      "step": 5705
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.4668719470500946,
+      "learning_rate": 9.951619326418582e-05,
+      "loss": 2.4294,
+      "step": 5710
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.4156099855899811,
+      "learning_rate": 9.951533187586178e-05,
+      "loss": 2.5367,
+      "step": 5715
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.41261160373687744,
+      "learning_rate": 9.951446972512958e-05,
+      "loss": 2.492,
+      "step": 5720
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.41526809334754944,
+      "learning_rate": 9.951360681200256e-05,
+      "loss": 2.4377,
+      "step": 5725
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.39695778489112854,
+      "learning_rate": 9.9512743136494e-05,
+      "loss": 2.3798,
+      "step": 5730
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.41677364706993103,
+      "learning_rate": 9.951187869861715e-05,
+      "loss": 2.4111,
+      "step": 5735
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.4094245433807373,
+      "learning_rate": 9.951101349838538e-05,
+      "loss": 2.4923,
+      "step": 5740
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.4074321687221527,
+      "learning_rate": 9.9510147535812e-05,
+      "loss": 2.4485,
+      "step": 5745
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.4164392948150635,
+      "learning_rate": 9.950928081091032e-05,
+      "loss": 2.4506,
+      "step": 5750
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.39663249254226685,
+      "learning_rate": 9.95084133236937e-05,
+      "loss": 2.4978,
+      "step": 5755
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.4294421076774597,
+      "learning_rate": 9.950754507417547e-05,
+      "loss": 2.4429,
+      "step": 5760
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.40427926182746887,
+      "learning_rate": 9.950667606236906e-05,
+      "loss": 2.3578,
+      "step": 5765
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.41513651609420776,
+      "learning_rate": 9.95058062882878e-05,
+      "loss": 2.4101,
+      "step": 5770
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.46088147163391113,
+      "learning_rate": 9.950493575194509e-05,
+      "loss": 2.423,
+      "step": 5775
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.4310627281665802,
+      "learning_rate": 9.950406445335434e-05,
+      "loss": 2.4729,
+      "step": 5780
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.44144850969314575,
+      "learning_rate": 9.950319239252896e-05,
+      "loss": 2.4755,
+      "step": 5785
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.4300086796283722,
+      "learning_rate": 9.95023195694824e-05,
+      "loss": 2.4664,
+      "step": 5790
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.4178891181945801,
+      "learning_rate": 9.950144598422809e-05,
+      "loss": 2.5256,
+      "step": 5795
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.4271286427974701,
+      "learning_rate": 9.950057163677944e-05,
+      "loss": 2.4857,
+      "step": 5800
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.4294353127479553,
+      "learning_rate": 9.949969652714997e-05,
+      "loss": 2.4746,
+      "step": 5805
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.45267602801322937,
+      "learning_rate": 9.949882065535312e-05,
+      "loss": 2.4892,
+      "step": 5810
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.43153294920921326,
+      "learning_rate": 9.949794402140239e-05,
+      "loss": 2.4615,
+      "step": 5815
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.4416215121746063,
+      "learning_rate": 9.949706662531127e-05,
+      "loss": 2.516,
+      "step": 5820
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.4342820942401886,
+      "learning_rate": 9.949618846709327e-05,
+      "loss": 2.4463,
+      "step": 5825
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.43174976110458374,
+      "learning_rate": 9.949530954676194e-05,
+      "loss": 2.5306,
+      "step": 5830
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.4359104335308075,
+      "learning_rate": 9.949442986433076e-05,
+      "loss": 2.4005,
+      "step": 5835
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.40752866864204407,
+      "learning_rate": 9.949354941981332e-05,
+      "loss": 2.423,
+      "step": 5840
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.44547757506370544,
+      "learning_rate": 9.949266821322315e-05,
+      "loss": 2.4161,
+      "step": 5845
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.40854179859161377,
+      "learning_rate": 9.949178624457384e-05,
+      "loss": 2.4603,
+      "step": 5850
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.4314653277397156,
+      "learning_rate": 9.949090351387895e-05,
+      "loss": 2.5106,
+      "step": 5855
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.44299545884132385,
+      "learning_rate": 9.949002002115208e-05,
+      "loss": 2.4326,
+      "step": 5860
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.4329400956630707,
+      "learning_rate": 9.948913576640683e-05,
+      "loss": 2.4115,
+      "step": 5865
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.42144954204559326,
+      "learning_rate": 9.948825074965682e-05,
+      "loss": 2.5452,
+      "step": 5870
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.40834516286849976,
+      "learning_rate": 9.948736497091569e-05,
+      "loss": 2.4532,
+      "step": 5875
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.3987160623073578,
+      "learning_rate": 9.948647843019705e-05,
+      "loss": 2.3375,
+      "step": 5880
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.40298992395401,
+      "learning_rate": 9.948559112751456e-05,
+      "loss": 2.4759,
+      "step": 5885
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.42205291986465454,
+      "learning_rate": 9.948470306288189e-05,
+      "loss": 2.4313,
+      "step": 5890
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.4339933693408966,
+      "learning_rate": 9.948381423631273e-05,
+      "loss": 2.4027,
+      "step": 5895
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.39590734243392944,
+      "learning_rate": 9.948292464782072e-05,
+      "loss": 2.3749,
+      "step": 5900
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.402484655380249,
+      "learning_rate": 9.948203429741961e-05,
+      "loss": 2.4262,
+      "step": 5905
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.40484076738357544,
+      "learning_rate": 9.948114318512307e-05,
+      "loss": 2.4386,
+      "step": 5910
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.38226252794265747,
+      "learning_rate": 9.948025131094482e-05,
+      "loss": 2.3632,
+      "step": 5915
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.40481722354888916,
+      "learning_rate": 9.947935867489862e-05,
+      "loss": 2.3914,
+      "step": 5920
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.4081464409828186,
+      "learning_rate": 9.94784652769982e-05,
+      "loss": 2.4526,
+      "step": 5925
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.4023546278476715,
+      "learning_rate": 9.947757111725732e-05,
+      "loss": 2.4056,
+      "step": 5930
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.4116899371147156,
+      "learning_rate": 9.947667619568974e-05,
+      "loss": 2.3711,
+      "step": 5935
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.47598007321357727,
+      "learning_rate": 9.947578051230925e-05,
+      "loss": 2.5389,
+      "step": 5940
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.3861636519432068,
+      "learning_rate": 9.947488406712962e-05,
+      "loss": 2.4908,
+      "step": 5945
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.41175225377082825,
+      "learning_rate": 9.947398686016469e-05,
+      "loss": 2.4094,
+      "step": 5950
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.4215227961540222,
+      "learning_rate": 9.947308889142824e-05,
+      "loss": 2.4673,
+      "step": 5955
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.5126275420188904,
+      "learning_rate": 9.94721901609341e-05,
+      "loss": 2.4138,
+      "step": 5960
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.46234771609306335,
+      "learning_rate": 9.947129066869614e-05,
+      "loss": 2.4644,
+      "step": 5965
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.4367867112159729,
+      "learning_rate": 9.947039041472817e-05,
+      "loss": 2.4974,
+      "step": 5970
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.4136449098587036,
+      "learning_rate": 9.946948939904407e-05,
+      "loss": 2.4679,
+      "step": 5975
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.39888525009155273,
+      "learning_rate": 9.946858762165772e-05,
+      "loss": 2.321,
+      "step": 5980
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.42329490184783936,
+      "learning_rate": 9.946768508258299e-05,
+      "loss": 2.4021,
+      "step": 5985
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.3924858868122101,
+      "learning_rate": 9.946678178183377e-05,
+      "loss": 2.4909,
+      "step": 5990
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.4089567959308624,
+      "learning_rate": 9.946587771942401e-05,
+      "loss": 2.4486,
+      "step": 5995
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.38580322265625,
+      "learning_rate": 9.946497289536758e-05,
+      "loss": 2.3772,
+      "step": 6000
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.4188164174556732,
+      "learning_rate": 9.946406730967846e-05,
+      "loss": 2.4564,
+      "step": 6005
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.397490918636322,
+      "learning_rate": 9.946316096237054e-05,
+      "loss": 2.4803,
+      "step": 6010
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.3886753022670746,
+      "learning_rate": 9.946225385345782e-05,
+      "loss": 2.4044,
+      "step": 6015
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.40456682443618774,
+      "learning_rate": 9.946134598295424e-05,
+      "loss": 2.4265,
+      "step": 6020
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.39964136481285095,
+      "learning_rate": 9.94604373508738e-05,
+      "loss": 2.5091,
+      "step": 6025
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.4189249575138092,
+      "learning_rate": 9.945952795723048e-05,
+      "loss": 2.4243,
+      "step": 6030
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.4289173185825348,
+      "learning_rate": 9.945861780203828e-05,
+      "loss": 2.5251,
+      "step": 6035
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.42690107226371765,
+      "learning_rate": 9.945770688531122e-05,
+      "loss": 2.4795,
+      "step": 6040
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.43397095799446106,
+      "learning_rate": 9.945679520706332e-05,
+      "loss": 2.4254,
+      "step": 6045
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.42214876413345337,
+      "learning_rate": 9.945588276730862e-05,
+      "loss": 2.5656,
+      "step": 6050
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.4140831530094147,
+      "learning_rate": 9.945496956606115e-05,
+      "loss": 2.4129,
+      "step": 6055
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.3972746729850769,
+      "learning_rate": 9.945405560333502e-05,
+      "loss": 2.4903,
+      "step": 6060
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.3931993246078491,
+      "learning_rate": 9.945314087914426e-05,
+      "loss": 2.3469,
+      "step": 6065
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.3971326947212219,
+      "learning_rate": 9.945222539350296e-05,
+      "loss": 2.463,
+      "step": 6070
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.44532573223114014,
+      "learning_rate": 9.945130914642524e-05,
+      "loss": 2.3935,
+      "step": 6075
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.4126253128051758,
+      "learning_rate": 9.945039213792518e-05,
+      "loss": 2.3572,
+      "step": 6080
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.40232282876968384,
+      "learning_rate": 9.94494743680169e-05,
+      "loss": 2.3874,
+      "step": 6085
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.44787362217903137,
+      "learning_rate": 9.944855583671455e-05,
+      "loss": 2.4646,
+      "step": 6090
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.42943114042282104,
+      "learning_rate": 9.944763654403226e-05,
+      "loss": 2.509,
+      "step": 6095
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.4373067617416382,
+      "learning_rate": 9.94467164899842e-05,
+      "loss": 2.3887,
+      "step": 6100
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.42681899666786194,
+      "learning_rate": 9.944579567458452e-05,
+      "loss": 2.3457,
+      "step": 6105
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.4255214333534241,
+      "learning_rate": 9.944487409784741e-05,
+      "loss": 2.5133,
+      "step": 6110
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.40245240926742554,
+      "learning_rate": 9.944395175978705e-05,
+      "loss": 2.3583,
+      "step": 6115
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.42830514907836914,
+      "learning_rate": 9.944302866041764e-05,
+      "loss": 2.471,
+      "step": 6120
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.4104147255420685,
+      "learning_rate": 9.944210479975342e-05,
+      "loss": 2.4161,
+      "step": 6125
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.41222354769706726,
+      "learning_rate": 9.944118017780857e-05,
+      "loss": 2.4712,
+      "step": 6130
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.4060019850730896,
+      "learning_rate": 9.944025479459735e-05,
+      "loss": 2.4009,
+      "step": 6135
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.4111667573451996,
+      "learning_rate": 9.943932865013404e-05,
+      "loss": 2.3866,
+      "step": 6140
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.4420322775840759,
+      "learning_rate": 9.943840174443284e-05,
+      "loss": 2.473,
+      "step": 6145
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.428916335105896,
+      "learning_rate": 9.943747407750805e-05,
+      "loss": 2.428,
+      "step": 6150
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.4222653806209564,
+      "learning_rate": 9.943654564937397e-05,
+      "loss": 2.3575,
+      "step": 6155
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.4110807180404663,
+      "learning_rate": 9.943561646004488e-05,
+      "loss": 2.47,
+      "step": 6160
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.4369598627090454,
+      "learning_rate": 9.943468650953508e-05,
+      "loss": 2.4665,
+      "step": 6165
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.4043520390987396,
+      "learning_rate": 9.943375579785889e-05,
+      "loss": 2.4444,
+      "step": 6170
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.4321403205394745,
+      "learning_rate": 9.943282432503067e-05,
+      "loss": 2.3951,
+      "step": 6175
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.4139578938484192,
+      "learning_rate": 9.943189209106472e-05,
+      "loss": 2.4799,
+      "step": 6180
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.41444599628448486,
+      "learning_rate": 9.943095909597541e-05,
+      "loss": 2.4353,
+      "step": 6185
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.4138924479484558,
+      "learning_rate": 9.943002533977713e-05,
+      "loss": 2.383,
+      "step": 6190
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.38398846983909607,
+      "learning_rate": 9.942909082248423e-05,
+      "loss": 2.43,
+      "step": 6195
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.40310612320899963,
+      "learning_rate": 9.942815554411109e-05,
+      "loss": 2.3798,
+      "step": 6200
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.3978731036186218,
+      "learning_rate": 9.942721950467214e-05,
+      "loss": 2.3701,
+      "step": 6205
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.4164629876613617,
+      "learning_rate": 9.942628270418177e-05,
+      "loss": 2.462,
+      "step": 6210
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.3953796923160553,
+      "learning_rate": 9.942534514265442e-05,
+      "loss": 2.4428,
+      "step": 6215
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.42330101132392883,
+      "learning_rate": 9.942440682010452e-05,
+      "loss": 2.4275,
+      "step": 6220
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.40899547934532166,
+      "learning_rate": 9.942346773654652e-05,
+      "loss": 2.4518,
+      "step": 6225
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.41820767521858215,
+      "learning_rate": 9.942252789199487e-05,
+      "loss": 2.3641,
+      "step": 6230
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.41114214062690735,
+      "learning_rate": 9.942158728646406e-05,
+      "loss": 2.435,
+      "step": 6235
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.39884719252586365,
+      "learning_rate": 9.942064591996855e-05,
+      "loss": 2.3143,
+      "step": 6240
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.4147103428840637,
+      "learning_rate": 9.941970379252284e-05,
+      "loss": 2.5068,
+      "step": 6245
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.40968331694602966,
+      "learning_rate": 9.941876090414146e-05,
+      "loss": 2.2803,
+      "step": 6250
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.40018534660339355,
+      "learning_rate": 9.941781725483889e-05,
+      "loss": 2.4087,
+      "step": 6255
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.4169664680957794,
+      "learning_rate": 9.941687284462969e-05,
+      "loss": 2.4283,
+      "step": 6260
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.38369739055633545,
+      "learning_rate": 9.941592767352839e-05,
+      "loss": 2.4204,
+      "step": 6265
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.3927948772907257,
+      "learning_rate": 9.941498174154954e-05,
+      "loss": 2.5234,
+      "step": 6270
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.42017558217048645,
+      "learning_rate": 9.94140350487077e-05,
+      "loss": 2.4449,
+      "step": 6275
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.40146195888519287,
+      "learning_rate": 9.941308759501747e-05,
+      "loss": 2.3816,
+      "step": 6280
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.39564546942710876,
+      "learning_rate": 9.941213938049342e-05,
+      "loss": 2.3664,
+      "step": 6285
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.45059606432914734,
+      "learning_rate": 9.941119040515017e-05,
+      "loss": 2.4335,
+      "step": 6290
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.3944545388221741,
+      "learning_rate": 9.941024066900229e-05,
+      "loss": 2.4871,
+      "step": 6295
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.3884962201118469,
+      "learning_rate": 9.940929017206445e-05,
+      "loss": 2.4824,
+      "step": 6300
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.3917818069458008,
+      "learning_rate": 9.940833891435124e-05,
+      "loss": 2.4177,
+      "step": 6305
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.38778427243232727,
+      "learning_rate": 9.940738689587735e-05,
+      "loss": 2.3866,
+      "step": 6310
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.4215979278087616,
+      "learning_rate": 9.940643411665742e-05,
+      "loss": 2.4437,
+      "step": 6315
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.4549429714679718,
+      "learning_rate": 9.940548057670612e-05,
+      "loss": 2.4716,
+      "step": 6320
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.39730244874954224,
+      "learning_rate": 9.940452627603813e-05,
+      "loss": 2.4802,
+      "step": 6325
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.36239299178123474,
+      "learning_rate": 9.940357121466813e-05,
+      "loss": 2.3577,
+      "step": 6330
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.42388901114463806,
+      "learning_rate": 9.940261539261086e-05,
+      "loss": 2.3466,
+      "step": 6335
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.4186631739139557,
+      "learning_rate": 9.940165880988102e-05,
+      "loss": 2.559,
+      "step": 6340
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.42322802543640137,
+      "learning_rate": 9.940070146649333e-05,
+      "loss": 2.4898,
+      "step": 6345
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.39554116129875183,
+      "learning_rate": 9.939974336246254e-05,
+      "loss": 2.3327,
+      "step": 6350
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.41189417243003845,
+      "learning_rate": 9.93987844978034e-05,
+      "loss": 2.4492,
+      "step": 6355
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.39223867654800415,
+      "learning_rate": 9.939782487253067e-05,
+      "loss": 2.4394,
+      "step": 6360
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.3880133032798767,
+      "learning_rate": 9.939686448665915e-05,
+      "loss": 2.3666,
+      "step": 6365
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.387031614780426,
+      "learning_rate": 9.939590334020359e-05,
+      "loss": 2.4437,
+      "step": 6370
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.38388341665267944,
+      "learning_rate": 9.939494143317881e-05,
+      "loss": 2.4403,
+      "step": 6375
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.4234341084957123,
+      "learning_rate": 9.939397876559962e-05,
+      "loss": 2.4266,
+      "step": 6380
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.44927510619163513,
+      "learning_rate": 9.939301533748084e-05,
+      "loss": 2.4011,
+      "step": 6385
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.40520554780960083,
+      "learning_rate": 9.939205114883731e-05,
+      "loss": 2.3901,
+      "step": 6390
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.36455434560775757,
+      "learning_rate": 9.939108619968386e-05,
+      "loss": 2.4106,
+      "step": 6395
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.3895112872123718,
+      "learning_rate": 9.939012049003537e-05,
+      "loss": 2.2999,
+      "step": 6400
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.4376757740974426,
+      "learning_rate": 9.93891540199067e-05,
+      "loss": 2.3916,
+      "step": 6405
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.41333508491516113,
+      "learning_rate": 9.938818678931271e-05,
+      "loss": 2.4541,
+      "step": 6410
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.41107043623924255,
+      "learning_rate": 9.938721879826832e-05,
+      "loss": 2.4292,
+      "step": 6415
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.39329200983047485,
+      "learning_rate": 9.938625004678844e-05,
+      "loss": 2.4252,
+      "step": 6420
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.39294275641441345,
+      "learning_rate": 9.938528053488796e-05,
+      "loss": 2.4957,
+      "step": 6425
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.4122803807258606,
+      "learning_rate": 9.938431026258182e-05,
+      "loss": 2.4639,
+      "step": 6430
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.3925172686576843,
+      "learning_rate": 9.938333922988497e-05,
+      "loss": 2.4044,
+      "step": 6435
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.3708503246307373,
+      "learning_rate": 9.938236743681235e-05,
+      "loss": 2.445,
+      "step": 6440
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.41597408056259155,
+      "learning_rate": 9.938139488337891e-05,
+      "loss": 2.4325,
+      "step": 6445
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.3819645941257477,
+      "learning_rate": 9.938042156959966e-05,
+      "loss": 2.35,
+      "step": 6450
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.4647391438484192,
+      "learning_rate": 9.937944749548956e-05,
+      "loss": 2.3662,
+      "step": 6455
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.40384405851364136,
+      "learning_rate": 9.937847266106359e-05,
+      "loss": 2.4162,
+      "step": 6460
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.40291324257850647,
+      "learning_rate": 9.937749706633682e-05,
+      "loss": 2.4599,
+      "step": 6465
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.4000307023525238,
+      "learning_rate": 9.937652071132422e-05,
+      "loss": 2.4854,
+      "step": 6470
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.3900514543056488,
+      "learning_rate": 9.937554359604084e-05,
+      "loss": 2.4574,
+      "step": 6475
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.4310036301612854,
+      "learning_rate": 9.937456572050172e-05,
+      "loss": 2.4943,
+      "step": 6480
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.3879684805870056,
+      "learning_rate": 9.937358708472193e-05,
+      "loss": 2.4635,
+      "step": 6485
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.40135887265205383,
+      "learning_rate": 9.937260768871654e-05,
+      "loss": 2.3961,
+      "step": 6490
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.4197976291179657,
+      "learning_rate": 9.93716275325006e-05,
+      "loss": 2.3935,
+      "step": 6495
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.4535994529724121,
+      "learning_rate": 9.937064661608924e-05,
+      "loss": 2.5043,
+      "step": 6500
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.39125746488571167,
+      "learning_rate": 9.936966493949752e-05,
+      "loss": 2.3486,
+      "step": 6505
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.4524657428264618,
+      "learning_rate": 9.936868250274059e-05,
+      "loss": 2.4108,
+      "step": 6510
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.4322394132614136,
+      "learning_rate": 9.936769930583359e-05,
+      "loss": 2.366,
+      "step": 6515
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.39550307393074036,
+      "learning_rate": 9.93667153487916e-05,
+      "loss": 2.4087,
+      "step": 6520
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.3773452341556549,
+      "learning_rate": 9.936573063162984e-05,
+      "loss": 2.4284,
+      "step": 6525
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.3868008553981781,
+      "learning_rate": 9.936474515436342e-05,
+      "loss": 2.439,
+      "step": 6530
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.38097408413887024,
+      "learning_rate": 9.936375891700753e-05,
+      "loss": 2.399,
+      "step": 6535
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.38632604479789734,
+      "learning_rate": 9.936277191957737e-05,
+      "loss": 2.4411,
+      "step": 6540
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.4030524492263794,
+      "learning_rate": 9.936178416208813e-05,
+      "loss": 2.3901,
+      "step": 6545
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.3928217589855194,
+      "learning_rate": 9.9360795644555e-05,
+      "loss": 2.453,
+      "step": 6550
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.38401854038238525,
+      "learning_rate": 9.935980636699322e-05,
+      "loss": 2.3374,
+      "step": 6555
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.4084552228450775,
+      "learning_rate": 9.935881632941801e-05,
+      "loss": 2.3915,
+      "step": 6560
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.42064937949180603,
+      "learning_rate": 9.935782553184463e-05,
+      "loss": 2.4401,
+      "step": 6565
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.42101553082466125,
+      "learning_rate": 9.935683397428833e-05,
+      "loss": 2.4856,
+      "step": 6570
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.39738377928733826,
+      "learning_rate": 9.935584165676436e-05,
+      "loss": 2.4845,
+      "step": 6575
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.4014604091644287,
+      "learning_rate": 9.935484857928801e-05,
+      "loss": 2.3948,
+      "step": 6580
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.3938162326812744,
+      "learning_rate": 9.935385474187459e-05,
+      "loss": 2.4364,
+      "step": 6585
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.4003547430038452,
+      "learning_rate": 9.93528601445394e-05,
+      "loss": 2.3553,
+      "step": 6590
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.4183579385280609,
+      "learning_rate": 9.935186478729771e-05,
+      "loss": 2.385,
+      "step": 6595
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.39463692903518677,
+      "learning_rate": 9.93508686701649e-05,
+      "loss": 2.4212,
+      "step": 6600
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.45294347405433655,
+      "learning_rate": 9.934987179315627e-05,
+      "loss": 2.3315,
+      "step": 6605
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.3868238031864166,
+      "learning_rate": 9.934887415628719e-05,
+      "loss": 2.3865,
+      "step": 6610
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.37825462222099304,
+      "learning_rate": 9.934787575957301e-05,
+      "loss": 2.4569,
+      "step": 6615
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.41850632429122925,
+      "learning_rate": 9.934687660302912e-05,
+      "loss": 2.4501,
+      "step": 6620
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.38461557030677795,
+      "learning_rate": 9.934587668667088e-05,
+      "loss": 2.478,
+      "step": 6625
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.4092370569705963,
+      "learning_rate": 9.93448760105137e-05,
+      "loss": 2.4806,
+      "step": 6630
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.3686850965023041,
+      "learning_rate": 9.934387457457299e-05,
+      "loss": 2.3072,
+      "step": 6635
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.4113618731498718,
+      "learning_rate": 9.934287237886417e-05,
+      "loss": 2.3938,
+      "step": 6640
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.3947383463382721,
+      "learning_rate": 9.934186942340265e-05,
+      "loss": 2.4572,
+      "step": 6645
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.4230484068393707,
+      "learning_rate": 9.934086570820392e-05,
+      "loss": 2.4857,
+      "step": 6650
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.4027727544307709,
+      "learning_rate": 9.933986123328337e-05,
+      "loss": 2.3235,
+      "step": 6655
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.3934069275856018,
+      "learning_rate": 9.933885599865651e-05,
+      "loss": 2.4808,
+      "step": 6660
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.3960352838039398,
+      "learning_rate": 9.933785000433882e-05,
+      "loss": 2.3921,
+      "step": 6665
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.3966562747955322,
+      "learning_rate": 9.933684325034577e-05,
+      "loss": 2.3695,
+      "step": 6670
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.419342041015625,
+      "learning_rate": 9.933583573669287e-05,
+      "loss": 2.317,
+      "step": 6675
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.39922794699668884,
+      "learning_rate": 9.933482746339564e-05,
+      "loss": 2.5177,
+      "step": 6680
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.4106009900569916,
+      "learning_rate": 9.933381843046959e-05,
+      "loss": 2.4415,
+      "step": 6685
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.38960400223731995,
+      "learning_rate": 9.933280863793026e-05,
+      "loss": 2.2901,
+      "step": 6690
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.4076553285121918,
+      "learning_rate": 9.933179808579321e-05,
+      "loss": 2.3963,
+      "step": 6695
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.3844624161720276,
+      "learning_rate": 9.933078677407399e-05,
+      "loss": 2.3735,
+      "step": 6700
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.37854811549186707,
+      "learning_rate": 9.932977470278818e-05,
+      "loss": 2.3691,
+      "step": 6705
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.4015577435493469,
+      "learning_rate": 9.932876187195135e-05,
+      "loss": 2.3824,
+      "step": 6710
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.3684917986392975,
+      "learning_rate": 9.932774828157911e-05,
+      "loss": 2.3444,
+      "step": 6715
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.4009644091129303,
+      "learning_rate": 9.932673393168705e-05,
+      "loss": 2.3994,
+      "step": 6720
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.4003559648990631,
+      "learning_rate": 9.932571882229081e-05,
+      "loss": 2.4226,
+      "step": 6725
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.40015944838523865,
+      "learning_rate": 9.932470295340599e-05,
+      "loss": 2.4982,
+      "step": 6730
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.381801038980484,
+      "learning_rate": 9.932368632504826e-05,
+      "loss": 2.3419,
+      "step": 6735
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.3856992721557617,
+      "learning_rate": 9.932266893723325e-05,
+      "loss": 2.4158,
+      "step": 6740
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.43591514229774475,
+      "learning_rate": 9.932165078997666e-05,
+      "loss": 2.4282,
+      "step": 6745
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.37896862626075745,
+      "learning_rate": 9.932063188329412e-05,
+      "loss": 2.3497,
+      "step": 6750
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.41021138429641724,
+      "learning_rate": 9.931961221720136e-05,
+      "loss": 2.3906,
+      "step": 6755
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.39199143648147583,
+      "learning_rate": 9.931859179171405e-05,
+      "loss": 2.4571,
+      "step": 6760
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.39339685440063477,
+      "learning_rate": 9.931757060684792e-05,
+      "loss": 2.4725,
+      "step": 6765
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.38259536027908325,
+      "learning_rate": 9.93165486626187e-05,
+      "loss": 2.4213,
+      "step": 6770
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.42154115438461304,
+      "learning_rate": 9.93155259590421e-05,
+      "loss": 2.4509,
+      "step": 6775
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.3899122476577759,
+      "learning_rate": 9.931450249613389e-05,
+      "loss": 2.3749,
+      "step": 6780
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.38905802369117737,
+      "learning_rate": 9.931347827390982e-05,
+      "loss": 2.3999,
+      "step": 6785
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.4075702726840973,
+      "learning_rate": 9.931245329238566e-05,
+      "loss": 2.4757,
+      "step": 6790
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.40164732933044434,
+      "learning_rate": 9.931142755157719e-05,
+      "loss": 2.4331,
+      "step": 6795
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.3900490701198578,
+      "learning_rate": 9.931040105150019e-05,
+      "loss": 2.4701,
+      "step": 6800
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.38056373596191406,
+      "learning_rate": 9.93093737921705e-05,
+      "loss": 2.3973,
+      "step": 6805
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.3809393048286438,
+      "learning_rate": 9.93083457736039e-05,
+      "loss": 2.3812,
+      "step": 6810
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.37728622555732727,
+      "learning_rate": 9.930731699581626e-05,
+      "loss": 2.3848,
+      "step": 6815
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.3866347670555115,
+      "learning_rate": 9.930628745882339e-05,
+      "loss": 2.4486,
+      "step": 6820
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.3657418191432953,
+      "learning_rate": 9.930525716264114e-05,
+      "loss": 2.431,
+      "step": 6825
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.38840287923812866,
+      "learning_rate": 9.93042261072854e-05,
+      "loss": 2.4069,
+      "step": 6830
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.3635327219963074,
+      "learning_rate": 9.930319429277202e-05,
+      "loss": 2.2626,
+      "step": 6835
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.37530097365379333,
+      "learning_rate": 9.93021617191169e-05,
+      "loss": 2.3998,
+      "step": 6840
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.37907060980796814,
+      "learning_rate": 9.930112838633592e-05,
+      "loss": 2.3942,
+      "step": 6845
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.41323164105415344,
+      "learning_rate": 9.930009429444503e-05,
+      "loss": 2.4419,
+      "step": 6850
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.38750505447387695,
+      "learning_rate": 9.929905944346012e-05,
+      "loss": 2.4203,
+      "step": 6855
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.40168970823287964,
+      "learning_rate": 9.929802383339713e-05,
+      "loss": 2.3819,
+      "step": 6860
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.41729095578193665,
+      "learning_rate": 9.9296987464272e-05,
+      "loss": 2.3809,
+      "step": 6865
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.37779906392097473,
+      "learning_rate": 9.929595033610072e-05,
+      "loss": 2.3974,
+      "step": 6870
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.4103444814682007,
+      "learning_rate": 9.929491244889921e-05,
+      "loss": 2.3925,
+      "step": 6875
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.3967428505420685,
+      "learning_rate": 9.929387380268349e-05,
+      "loss": 2.4103,
+      "step": 6880
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.38610827922821045,
+      "learning_rate": 9.929283439746955e-05,
+      "loss": 2.4244,
+      "step": 6885
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.372811496257782,
+      "learning_rate": 9.929179423327336e-05,
+      "loss": 2.3165,
+      "step": 6890
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.3871278464794159,
+      "learning_rate": 9.929075331011097e-05,
+      "loss": 2.413,
+      "step": 6895
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.5697849988937378,
+      "learning_rate": 9.928971162799838e-05,
+      "loss": 2.2404,
+      "step": 6900
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.3949141502380371,
+      "learning_rate": 9.928866918695165e-05,
+      "loss": 2.4574,
+      "step": 6905
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.4043257236480713,
+      "learning_rate": 9.928762598698684e-05,
+      "loss": 2.3821,
+      "step": 6910
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.3689405918121338,
+      "learning_rate": 9.928658202811999e-05,
+      "loss": 2.3155,
+      "step": 6915
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.3634750545024872,
+      "learning_rate": 9.928553731036719e-05,
+      "loss": 2.4689,
+      "step": 6920
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.4094627797603607,
+      "learning_rate": 9.928449183374451e-05,
+      "loss": 2.457,
+      "step": 6925
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.37538009881973267,
+      "learning_rate": 9.928344559826806e-05,
+      "loss": 2.4029,
+      "step": 6930
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.39394068717956543,
+      "learning_rate": 9.928239860395394e-05,
+      "loss": 2.4126,
+      "step": 6935
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.3731463551521301,
+      "learning_rate": 9.928135085081827e-05,
+      "loss": 2.3812,
+      "step": 6940
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.39725419878959656,
+      "learning_rate": 9.928030233887721e-05,
+      "loss": 2.3789,
+      "step": 6945
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.39634889364242554,
+      "learning_rate": 9.927925306814686e-05,
+      "loss": 2.5054,
+      "step": 6950
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.402915894985199,
+      "learning_rate": 9.927820303864341e-05,
+      "loss": 2.3843,
+      "step": 6955
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.38081100583076477,
+      "learning_rate": 9.927715225038303e-05,
+      "loss": 2.3644,
+      "step": 6960
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.4103027284145355,
+      "learning_rate": 9.927610070338187e-05,
+      "loss": 2.3557,
+      "step": 6965
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.39573001861572266,
+      "learning_rate": 9.927504839765615e-05,
+      "loss": 2.4449,
+      "step": 6970
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.3923678398132324,
+      "learning_rate": 9.927399533322206e-05,
+      "loss": 2.3331,
+      "step": 6975
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.39690858125686646,
+      "learning_rate": 9.92729415100958e-05,
+      "loss": 2.3571,
+      "step": 6980
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.3891814053058624,
+      "learning_rate": 9.927188692829363e-05,
+      "loss": 2.398,
+      "step": 6985
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.3998129665851593,
+      "learning_rate": 9.927083158783177e-05,
+      "loss": 2.3052,
+      "step": 6990
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.4100293219089508,
+      "learning_rate": 9.926977548872644e-05,
+      "loss": 2.4137,
+      "step": 6995
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.3871023952960968,
+      "learning_rate": 9.926871863099396e-05,
+      "loss": 2.3696,
+      "step": 7000
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.39212852716445923,
+      "learning_rate": 9.926766101465056e-05,
+      "loss": 2.4183,
+      "step": 7005
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.36996224522590637,
+      "learning_rate": 9.926660263971254e-05,
+      "loss": 2.326,
+      "step": 7010
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.3661785125732422,
+      "learning_rate": 9.92655435061962e-05,
+      "loss": 2.3455,
+      "step": 7015
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.3902539908885956,
+      "learning_rate": 9.926448361411784e-05,
+      "loss": 2.3519,
+      "step": 7020
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.3757844567298889,
+      "learning_rate": 9.926342296349378e-05,
+      "loss": 2.4038,
+      "step": 7025
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.3718620538711548,
+      "learning_rate": 9.926236155434034e-05,
+      "loss": 2.3237,
+      "step": 7030
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.41075217723846436,
+      "learning_rate": 9.926129938667388e-05,
+      "loss": 2.3832,
+      "step": 7035
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.39685356616973877,
+      "learning_rate": 9.926023646051075e-05,
+      "loss": 2.3643,
+      "step": 7040
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.4140591323375702,
+      "learning_rate": 9.925917277586734e-05,
+      "loss": 2.4511,
+      "step": 7045
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.3595258295536041,
+      "learning_rate": 9.925810833275998e-05,
+      "loss": 2.3692,
+      "step": 7050
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.37691229581832886,
+      "learning_rate": 9.925704313120508e-05,
+      "loss": 2.4487,
+      "step": 7055
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.3856925666332245,
+      "learning_rate": 9.925597717121906e-05,
+      "loss": 2.4495,
+      "step": 7060
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.39405855536460876,
+      "learning_rate": 9.925491045281832e-05,
+      "loss": 2.4122,
+      "step": 7065
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.3859237730503082,
+      "learning_rate": 9.925384297601928e-05,
+      "loss": 2.5056,
+      "step": 7070
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.39201733469963074,
+      "learning_rate": 9.925277474083835e-05,
+      "loss": 2.4606,
+      "step": 7075
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.3821859061717987,
+      "learning_rate": 9.925170574729204e-05,
+      "loss": 2.4014,
+      "step": 7080
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.37822872400283813,
+      "learning_rate": 9.925063599539677e-05,
+      "loss": 2.3365,
+      "step": 7085
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.39815863966941833,
+      "learning_rate": 9.924956548516901e-05,
+      "loss": 2.3008,
+      "step": 7090
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.5269124507904053,
+      "learning_rate": 9.924849421662528e-05,
+      "loss": 2.3662,
+      "step": 7095
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.39095330238342285,
+      "learning_rate": 9.924742218978203e-05,
+      "loss": 2.4377,
+      "step": 7100
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.3737209737300873,
+      "learning_rate": 9.924634940465577e-05,
+      "loss": 2.3583,
+      "step": 7105
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.38882026076316833,
+      "learning_rate": 9.924527586126305e-05,
+      "loss": 2.3119,
+      "step": 7110
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.4150848984718323,
+      "learning_rate": 9.924420155962037e-05,
+      "loss": 2.4388,
+      "step": 7115
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.3820304572582245,
+      "learning_rate": 9.924312649974427e-05,
+      "loss": 2.5435,
+      "step": 7120
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.39374080300331116,
+      "learning_rate": 9.924205068165134e-05,
+      "loss": 2.402,
+      "step": 7125
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.4140278100967407,
+      "learning_rate": 9.924097410535811e-05,
+      "loss": 2.4428,
+      "step": 7130
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.4072985053062439,
+      "learning_rate": 9.923989677088116e-05,
+      "loss": 2.4548,
+      "step": 7135
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.3700661063194275,
+      "learning_rate": 9.92388186782371e-05,
+      "loss": 2.4461,
+      "step": 7140
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.41510409116744995,
+      "learning_rate": 9.92377398274425e-05,
+      "loss": 2.4277,
+      "step": 7145
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.37451237440109253,
+      "learning_rate": 9.923666021851398e-05,
+      "loss": 2.4351,
+      "step": 7150
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.3952525854110718,
+      "learning_rate": 9.92355798514682e-05,
+      "loss": 2.3436,
+      "step": 7155
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.37874433398246765,
+      "learning_rate": 9.923449872632173e-05,
+      "loss": 2.4979,
+      "step": 7160
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.3876708149909973,
+      "learning_rate": 9.923341684309128e-05,
+      "loss": 2.4866,
+      "step": 7165
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.39724329113960266,
+      "learning_rate": 9.923233420179347e-05,
+      "loss": 2.3467,
+      "step": 7170
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.38195350766181946,
+      "learning_rate": 9.923125080244497e-05,
+      "loss": 2.3226,
+      "step": 7175
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.41608017683029175,
+      "learning_rate": 9.923016664506248e-05,
+      "loss": 2.4532,
+      "step": 7180
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.4102792739868164,
+      "learning_rate": 9.922908172966268e-05,
+      "loss": 2.3636,
+      "step": 7185
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.41052353382110596,
+      "learning_rate": 9.922799605626227e-05,
+      "loss": 2.3913,
+      "step": 7190
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.36142799258232117,
+      "learning_rate": 9.9226909624878e-05,
+      "loss": 2.3554,
+      "step": 7195
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.4051479697227478,
+      "learning_rate": 9.922582243552654e-05,
+      "loss": 2.4653,
+      "step": 7200
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.3846384882926941,
+      "learning_rate": 9.922473448822469e-05,
+      "loss": 2.3813,
+      "step": 7205
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.4064043164253235,
+      "learning_rate": 9.922364578298916e-05,
+      "loss": 2.3908,
+      "step": 7210
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.391489177942276,
+      "learning_rate": 9.922255631983673e-05,
+      "loss": 2.3749,
+      "step": 7215
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.3981476128101349,
+      "learning_rate": 9.922146609878418e-05,
+      "loss": 2.4254,
+      "step": 7220
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.37059667706489563,
+      "learning_rate": 9.922037511984829e-05,
+      "loss": 2.3285,
+      "step": 7225
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.40299201011657715,
+      "learning_rate": 9.921928338304584e-05,
+      "loss": 2.3693,
+      "step": 7230
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.36940956115722656,
+      "learning_rate": 9.921819088839369e-05,
+      "loss": 2.3173,
+      "step": 7235
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.37730613350868225,
+      "learning_rate": 9.92170976359086e-05,
+      "loss": 2.4302,
+      "step": 7240
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.3935486376285553,
+      "learning_rate": 9.921600362560744e-05,
+      "loss": 2.4331,
+      "step": 7245
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.41434401273727417,
+      "learning_rate": 9.921490885750706e-05,
+      "loss": 2.3471,
+      "step": 7250
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.38703855872154236,
+      "learning_rate": 9.92138133316243e-05,
+      "loss": 2.389,
+      "step": 7255
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.41050875186920166,
+      "learning_rate": 9.921271704797602e-05,
+      "loss": 2.4319,
+      "step": 7260
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.372763067483902,
+      "learning_rate": 9.921162000657913e-05,
+      "loss": 2.3255,
+      "step": 7265
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.4627690315246582,
+      "learning_rate": 9.921052220745048e-05,
+      "loss": 2.3642,
+      "step": 7270
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.37675344944000244,
+      "learning_rate": 9.920942365060702e-05,
+      "loss": 2.3658,
+      "step": 7275
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.3996916115283966,
+      "learning_rate": 9.920832433606564e-05,
+      "loss": 2.2863,
+      "step": 7280
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.3792148232460022,
+      "learning_rate": 9.920722426384326e-05,
+      "loss": 2.3772,
+      "step": 7285
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.42612460255622864,
+      "learning_rate": 9.920612343395683e-05,
+      "loss": 2.2908,
+      "step": 7290
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.3802616596221924,
+      "learning_rate": 9.92050218464233e-05,
+      "loss": 2.3494,
+      "step": 7295
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.3919801414012909,
+      "learning_rate": 9.920391950125963e-05,
+      "loss": 2.3347,
+      "step": 7300
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.3836115896701813,
+      "learning_rate": 9.920281639848278e-05,
+      "loss": 2.4234,
+      "step": 7305
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.37855634093284607,
+      "learning_rate": 9.920171253810975e-05,
+      "loss": 2.4398,
+      "step": 7310
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.4219200313091278,
+      "learning_rate": 9.920060792015754e-05,
+      "loss": 2.4211,
+      "step": 7315
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.4216143786907196,
+      "learning_rate": 9.919950254464316e-05,
+      "loss": 2.3711,
+      "step": 7320
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.44248148798942566,
+      "learning_rate": 9.919839641158361e-05,
+      "loss": 2.3994,
+      "step": 7325
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.37937596440315247,
+      "learning_rate": 9.919728952099593e-05,
+      "loss": 2.4436,
+      "step": 7330
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.42273813486099243,
+      "learning_rate": 9.919618187289717e-05,
+      "loss": 2.4245,
+      "step": 7335
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.392367422580719,
+      "learning_rate": 9.919507346730437e-05,
+      "loss": 2.4457,
+      "step": 7340
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.3875828981399536,
+      "learning_rate": 9.919396430423463e-05,
+      "loss": 2.3836,
+      "step": 7345
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.36270034313201904,
+      "learning_rate": 9.9192854383705e-05,
+      "loss": 2.4045,
+      "step": 7350
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.3954014480113983,
+      "learning_rate": 9.919174370573257e-05,
+      "loss": 2.3671,
+      "step": 7355
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.38273829221725464,
+      "learning_rate": 9.919063227033445e-05,
+      "loss": 2.4128,
+      "step": 7360
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.39100712537765503,
+      "learning_rate": 9.918952007752776e-05,
+      "loss": 2.3402,
+      "step": 7365
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.382375568151474,
+      "learning_rate": 9.91884071273296e-05,
+      "loss": 2.4169,
+      "step": 7370
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.4154987931251526,
+      "learning_rate": 9.918729341975712e-05,
+      "loss": 2.3415,
+      "step": 7375
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.40309762954711914,
+      "learning_rate": 9.918617895482749e-05,
+      "loss": 2.4395,
+      "step": 7380
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.41311776638031006,
+      "learning_rate": 9.918506373255784e-05,
+      "loss": 2.3301,
+      "step": 7385
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.38003242015838623,
+      "learning_rate": 9.918394775296536e-05,
+      "loss": 2.3283,
+      "step": 7390
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.38214972615242004,
+      "learning_rate": 9.91828310160672e-05,
+      "loss": 2.4016,
+      "step": 7395
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.3798430860042572,
+      "learning_rate": 9.918171352188062e-05,
+      "loss": 2.361,
+      "step": 7400
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.37754109501838684,
+      "learning_rate": 9.918059527042275e-05,
+      "loss": 2.4173,
+      "step": 7405
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.3851357102394104,
+      "learning_rate": 9.917947626171086e-05,
+      "loss": 2.4388,
+      "step": 7410
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.38617485761642456,
+      "learning_rate": 9.917835649576217e-05,
+      "loss": 2.3435,
+      "step": 7415
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.3717854619026184,
+      "learning_rate": 9.91772359725939e-05,
+      "loss": 2.4215,
+      "step": 7420
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.39697930216789246,
+      "learning_rate": 9.917611469222333e-05,
+      "loss": 2.4028,
+      "step": 7425
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.38296982645988464,
+      "learning_rate": 9.917499265466772e-05,
+      "loss": 2.3798,
+      "step": 7430
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.3896041810512543,
+      "learning_rate": 9.917386985994432e-05,
+      "loss": 2.3858,
+      "step": 7435
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.37736397981643677,
+      "learning_rate": 9.917274630807047e-05,
+      "loss": 2.3936,
+      "step": 7440
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.41972464323043823,
+      "learning_rate": 9.917162199906341e-05,
+      "loss": 2.3094,
+      "step": 7445
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.4049564003944397,
+      "learning_rate": 9.917049693294049e-05,
+      "loss": 2.3545,
+      "step": 7450
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.397005170583725,
+      "learning_rate": 9.916937110971902e-05,
+      "loss": 2.4189,
+      "step": 7455
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.39709025621414185,
+      "learning_rate": 9.916824452941635e-05,
+      "loss": 2.4572,
+      "step": 7460
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.3765231966972351,
+      "learning_rate": 9.916711719204981e-05,
+      "loss": 2.42,
+      "step": 7465
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.3726831078529358,
+      "learning_rate": 9.916598909763675e-05,
+      "loss": 2.3881,
+      "step": 7470
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.3879096508026123,
+      "learning_rate": 9.916486024619455e-05,
+      "loss": 2.4516,
+      "step": 7475
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.3731110990047455,
+      "learning_rate": 9.916373063774062e-05,
+      "loss": 2.3151,
+      "step": 7480
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.38409629464149475,
+      "learning_rate": 9.91626002722923e-05,
+      "loss": 2.3245,
+      "step": 7485
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.3826192319393158,
+      "learning_rate": 9.916146914986703e-05,
+      "loss": 2.2393,
+      "step": 7490
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.38986027240753174,
+      "learning_rate": 9.916033727048223e-05,
+      "loss": 2.3379,
+      "step": 7495
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.3654114902019501,
+      "learning_rate": 9.915920463415531e-05,
+      "loss": 2.3193,
+      "step": 7500
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.4038855731487274,
+      "learning_rate": 9.915807124090371e-05,
+      "loss": 2.4188,
+      "step": 7505
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.38425928354263306,
+      "learning_rate": 9.915693709074487e-05,
+      "loss": 2.4224,
+      "step": 7510
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.4007771909236908,
+      "learning_rate": 9.91558021836963e-05,
+      "loss": 2.3858,
+      "step": 7515
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.39653363823890686,
+      "learning_rate": 9.915466651977543e-05,
+      "loss": 2.332,
+      "step": 7520
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.3634871542453766,
+      "learning_rate": 9.915353009899976e-05,
+      "loss": 2.3991,
+      "step": 7525
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.4162198007106781,
+      "learning_rate": 9.915239292138678e-05,
+      "loss": 2.4485,
+      "step": 7530
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.36684197187423706,
+      "learning_rate": 9.915125498695402e-05,
+      "loss": 2.4191,
+      "step": 7535
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.3802706301212311,
+      "learning_rate": 9.9150116295719e-05,
+      "loss": 2.4006,
+      "step": 7540
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.4069001078605652,
+      "learning_rate": 9.914897684769921e-05,
+      "loss": 2.379,
+      "step": 7545
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.4017879068851471,
+      "learning_rate": 9.914783664291225e-05,
+      "loss": 2.3847,
+      "step": 7550
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.3683493137359619,
+      "learning_rate": 9.914669568137565e-05,
+      "loss": 2.4268,
+      "step": 7555
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.3779098689556122,
+      "learning_rate": 9.914555396310698e-05,
+      "loss": 2.4025,
+      "step": 7560
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.3903259336948395,
+      "learning_rate": 9.914441148812382e-05,
+      "loss": 2.385,
+      "step": 7565
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.3891471326351166,
+      "learning_rate": 9.914326825644376e-05,
+      "loss": 2.4169,
+      "step": 7570
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.3856396973133087,
+      "learning_rate": 9.91421242680844e-05,
+      "loss": 2.4059,
+      "step": 7575
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.3859300911426544,
+      "learning_rate": 9.914097952306338e-05,
+      "loss": 2.4231,
+      "step": 7580
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.36656346917152405,
+      "learning_rate": 9.913983402139828e-05,
+      "loss": 2.4235,
+      "step": 7585
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.41750648617744446,
+      "learning_rate": 9.913868776310677e-05,
+      "loss": 2.3355,
+      "step": 7590
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.3801565170288086,
+      "learning_rate": 9.913754074820649e-05,
+      "loss": 2.3763,
+      "step": 7595
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.3940177857875824,
+      "learning_rate": 9.913639297671511e-05,
+      "loss": 2.377,
+      "step": 7600
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.39858612418174744,
+      "learning_rate": 9.91352444486503e-05,
+      "loss": 2.373,
+      "step": 7605
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.4101460874080658,
+      "learning_rate": 9.913409516402974e-05,
+      "loss": 2.3576,
+      "step": 7610
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.3805282413959503,
+      "learning_rate": 9.913294512287112e-05,
+      "loss": 2.4052,
+      "step": 7615
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.38430556654930115,
+      "learning_rate": 9.913179432519215e-05,
+      "loss": 2.4088,
+      "step": 7620
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.40866968035697937,
+      "learning_rate": 9.913064277101055e-05,
+      "loss": 2.4026,
+      "step": 7625
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.3839671015739441,
+      "learning_rate": 9.912949046034406e-05,
+      "loss": 2.3412,
+      "step": 7630
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.3978743553161621,
+      "learning_rate": 9.912833739321042e-05,
+      "loss": 2.3052,
+      "step": 7635
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.3727779686450958,
+      "learning_rate": 9.912718356962738e-05,
+      "loss": 2.4965,
+      "step": 7640
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.3656705915927887,
+      "learning_rate": 9.91260289896127e-05,
+      "loss": 2.419,
+      "step": 7645
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.37696295976638794,
+      "learning_rate": 9.912487365318417e-05,
+      "loss": 2.3394,
+      "step": 7650
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.36101776361465454,
+      "learning_rate": 9.912371756035957e-05,
+      "loss": 2.4318,
+      "step": 7655
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.3893497884273529,
+      "learning_rate": 9.912256071115671e-05,
+      "loss": 2.3494,
+      "step": 7660
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.3917941153049469,
+      "learning_rate": 9.91214031055934e-05,
+      "loss": 2.2674,
+      "step": 7665
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.3854151964187622,
+      "learning_rate": 9.912024474368745e-05,
+      "loss": 2.3717,
+      "step": 7670
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.4030923545360565,
+      "learning_rate": 9.911908562545671e-05,
+      "loss": 2.3695,
+      "step": 7675
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.39796075224876404,
+      "learning_rate": 9.911792575091903e-05,
+      "loss": 2.3933,
+      "step": 7680
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.4091487526893616,
+      "learning_rate": 9.911676512009226e-05,
+      "loss": 2.3144,
+      "step": 7685
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.40143176913261414,
+      "learning_rate": 9.911560373299426e-05,
+      "loss": 2.4199,
+      "step": 7690
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.3907955288887024,
+      "learning_rate": 9.911444158964296e-05,
+      "loss": 2.2573,
+      "step": 7695
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.38906794786453247,
+      "learning_rate": 9.91132786900562e-05,
+      "loss": 2.3294,
+      "step": 7700
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.37860071659088135,
+      "learning_rate": 9.911211503425191e-05,
+      "loss": 2.3464,
+      "step": 7705
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.36610275506973267,
+      "learning_rate": 9.911095062224798e-05,
+      "loss": 2.3099,
+      "step": 7710
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.3735958933830261,
+      "learning_rate": 9.91097854540624e-05,
+      "loss": 2.3644,
+      "step": 7715
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.40103012323379517,
+      "learning_rate": 9.910861952971304e-05,
+      "loss": 2.4039,
+      "step": 7720
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.3786330223083496,
+      "learning_rate": 9.910745284921791e-05,
+      "loss": 2.3314,
+      "step": 7725
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.41469430923461914,
+      "learning_rate": 9.910628541259495e-05,
+      "loss": 2.3066,
+      "step": 7730
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.3950195610523224,
+      "learning_rate": 9.91051172198621e-05,
+      "loss": 2.3579,
+      "step": 7735
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.4292914867401123,
+      "learning_rate": 9.910394827103742e-05,
+      "loss": 2.4082,
+      "step": 7740
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.40993165969848633,
+      "learning_rate": 9.910277856613886e-05,
+      "loss": 2.2952,
+      "step": 7745
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.39674684405326843,
+      "learning_rate": 9.910160810518442e-05,
+      "loss": 2.419,
+      "step": 7750
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.36979225277900696,
+      "learning_rate": 9.910043688819218e-05,
+      "loss": 2.4668,
+      "step": 7755
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.4170330762863159,
+      "learning_rate": 9.909926491518012e-05,
+      "loss": 2.3376,
+      "step": 7760
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.42854490876197815,
+      "learning_rate": 9.909809218616629e-05,
+      "loss": 2.4495,
+      "step": 7765
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.38324177265167236,
+      "learning_rate": 9.909691870116877e-05,
+      "loss": 2.3291,
+      "step": 7770
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.37091174721717834,
+      "learning_rate": 9.909574446020561e-05,
+      "loss": 2.3859,
+      "step": 7775
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.3682883083820343,
+      "learning_rate": 9.90945694632949e-05,
+      "loss": 2.5034,
+      "step": 7780
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.35963916778564453,
+      "learning_rate": 9.909339371045474e-05,
+      "loss": 2.346,
+      "step": 7785
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.3968486487865448,
+      "learning_rate": 9.90922172017032e-05,
+      "loss": 2.3779,
+      "step": 7790
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.40308964252471924,
+      "learning_rate": 9.909103993705845e-05,
+      "loss": 2.3618,
+      "step": 7795
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.37105122208595276,
+      "learning_rate": 9.908986191653857e-05,
+      "loss": 2.3054,
+      "step": 7800
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.37456053495407104,
+      "learning_rate": 9.90886831401617e-05,
+      "loss": 2.368,
+      "step": 7805
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.3910890221595764,
+      "learning_rate": 9.908750360794603e-05,
+      "loss": 2.4364,
+      "step": 7810
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.3951781094074249,
+      "learning_rate": 9.908632331990968e-05,
+      "loss": 2.3467,
+      "step": 7815
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.37857773900032043,
+      "learning_rate": 9.908514227607084e-05,
+      "loss": 2.3598,
+      "step": 7820
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.37768179178237915,
+      "learning_rate": 9.90839604764477e-05,
+      "loss": 2.3696,
+      "step": 7825
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.4079396724700928,
+      "learning_rate": 9.908277792105845e-05,
+      "loss": 2.3256,
+      "step": 7830
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.4240504503250122,
+      "learning_rate": 9.90815946099213e-05,
+      "loss": 2.4263,
+      "step": 7835
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.3814687430858612,
+      "learning_rate": 9.908041054305446e-05,
+      "loss": 2.3754,
+      "step": 7840
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.36595654487609863,
+      "learning_rate": 9.907922572047618e-05,
+      "loss": 2.3369,
+      "step": 7845
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.36377689242362976,
+      "learning_rate": 9.90780401422047e-05,
+      "loss": 2.4015,
+      "step": 7850
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.3776533901691437,
+      "learning_rate": 9.907685380825825e-05,
+      "loss": 2.4075,
+      "step": 7855
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.37529441714286804,
+      "learning_rate": 9.907566671865512e-05,
+      "loss": 2.3543,
+      "step": 7860
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.3694101572036743,
+      "learning_rate": 9.907447887341359e-05,
+      "loss": 2.4515,
+      "step": 7865
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.4920283555984497,
+      "learning_rate": 9.907329027255194e-05,
+      "loss": 2.3422,
+      "step": 7870
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.3817356824874878,
+      "learning_rate": 9.907210091608848e-05,
+      "loss": 2.3584,
+      "step": 7875
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.157424807548523,
+      "learning_rate": 9.907091080404152e-05,
+      "loss": 2.3664,
+      "step": 7880
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.3837755620479584,
+      "learning_rate": 9.906971993642937e-05,
+      "loss": 2.4113,
+      "step": 7885
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.4128318727016449,
+      "learning_rate": 9.906852831327038e-05,
+      "loss": 2.433,
+      "step": 7890
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.36866334080696106,
+      "learning_rate": 9.906733593458289e-05,
+      "loss": 2.3063,
+      "step": 7895
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.3828504681587219,
+      "learning_rate": 9.906614280038529e-05,
+      "loss": 2.3844,
+      "step": 7900
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.392856627702713,
+      "learning_rate": 9.90649489106959e-05,
+      "loss": 2.3773,
+      "step": 7905
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.37735483050346375,
+      "learning_rate": 9.906375426553314e-05,
+      "loss": 2.3679,
+      "step": 7910
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.39767247438430786,
+      "learning_rate": 9.90625588649154e-05,
+      "loss": 2.358,
+      "step": 7915
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.3949820101261139,
+      "learning_rate": 9.906136270886107e-05,
+      "loss": 2.3867,
+      "step": 7920
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.37822195887565613,
+      "learning_rate": 9.90601657973886e-05,
+      "loss": 2.4013,
+      "step": 7925
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.37998852133750916,
+      "learning_rate": 9.905896813051636e-05,
+      "loss": 2.3363,
+      "step": 7930
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.37479186058044434,
+      "learning_rate": 9.905776970826286e-05,
+      "loss": 2.3779,
+      "step": 7935
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.37575533986091614,
+      "learning_rate": 9.905657053064651e-05,
+      "loss": 2.36,
+      "step": 7940
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.3584083616733551,
+      "learning_rate": 9.905537059768578e-05,
+      "loss": 2.2968,
+      "step": 7945
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.3788602650165558,
+      "learning_rate": 9.905416990939915e-05,
+      "loss": 2.4101,
+      "step": 7950
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.3688214123249054,
+      "learning_rate": 9.905296846580512e-05,
+      "loss": 2.348,
+      "step": 7955
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.38867130875587463,
+      "learning_rate": 9.905176626692217e-05,
+      "loss": 2.4043,
+      "step": 7960
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.3870021104812622,
+      "learning_rate": 9.905056331276884e-05,
+      "loss": 2.4879,
+      "step": 7965
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.40862980484962463,
+      "learning_rate": 9.904935960336359e-05,
+      "loss": 2.5296,
+      "step": 7970
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.3772049844264984,
+      "learning_rate": 9.904815513872503e-05,
+      "loss": 2.4187,
+      "step": 7975
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.37248101830482483,
+      "learning_rate": 9.904694991887164e-05,
+      "loss": 2.3217,
+      "step": 7980
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.3639294505119324,
+      "learning_rate": 9.904574394382203e-05,
+      "loss": 2.4743,
+      "step": 7985
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.40883296728134155,
+      "learning_rate": 9.904453721359474e-05,
+      "loss": 2.2816,
+      "step": 7990
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.40878501534461975,
+      "learning_rate": 9.904332972820837e-05,
+      "loss": 2.3922,
+      "step": 7995
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.37084394693374634,
+      "learning_rate": 9.904212148768148e-05,
+      "loss": 2.3847,
+      "step": 8000
+    }
+  ],
+  "logging_steps": 5,
+  "max_steps": 126689,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 1000,
+  "total_flos": 2.173076069626675e+19,
+  "train_batch_size": 16,
+  "trial_name": null,
+  "trial_params": null
+}