model checkpoint updated

README.md

@@ -49,7 +49,7 @@ The following hyperparameters were used during training:
 - optimizer: adamw_bnb_8bit
 - lr_scheduler_type: linear
 - lr_scheduler_warmup_steps: 4000
-- training_steps: 7080 (terminated upon convergence. Initially set to 21240 steps)
+- training_steps: 21240
 - mixed_precision_training: True
 
 ## Acknowledgement

pytorch_model.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:1fcbf405f4c04143e43e6dad2ec858b34607bde72dde81786bd2ea7f20a00643
+oid sha256:b6a70aa520d94ab03d43b72a5196e6c547a72eabb3f629bdf65b311e5d5e19bf
 size 3055748571