Pushed by DataDreamer
Browse filesUpdate training_args.json
- training_args.json +10 -10
training_args.json
CHANGED
|
@@ -1,25 +1,25 @@
|
|
| 1 |
{
|
| 2 |
-
"output_dir": "./output/train-styledistance-model/_checkpoints",
|
| 3 |
"overwrite_output_dir": false,
|
| 4 |
"do_train": false,
|
| 5 |
"do_eval": true,
|
| 6 |
"do_predict": false,
|
| 7 |
"evaluation_strategy": "steps",
|
| 8 |
"prediction_loss_only": false,
|
| 9 |
-
"per_device_train_batch_size":
|
| 10 |
-
"per_device_eval_batch_size":
|
| 11 |
"per_gpu_train_batch_size": null,
|
| 12 |
"per_gpu_eval_batch_size": null,
|
| 13 |
-
"gradient_accumulation_steps":
|
| 14 |
"eval_accumulation_steps": 1,
|
| 15 |
"eval_delay": 0,
|
| 16 |
-
"learning_rate": 0.
|
| 17 |
"weight_decay": 0.01,
|
| 18 |
"adam_beta1": 0.9,
|
| 19 |
"adam_beta2": 0.999,
|
| 20 |
"adam_epsilon": 1e-08,
|
| 21 |
"max_grad_norm": 1.0,
|
| 22 |
-
"num_train_epochs":
|
| 23 |
"max_steps": -1,
|
| 24 |
"lr_scheduler_type": "linear",
|
| 25 |
"lr_scheduler_kwargs": {},
|
|
@@ -28,13 +28,13 @@
|
|
| 28 |
"log_level": "passive",
|
| 29 |
"log_level_replica": "warning",
|
| 30 |
"log_on_each_node": true,
|
| 31 |
-
"logging_dir": "./output/train-styledistance-model/_checkpoints/runs/
|
| 32 |
"logging_strategy": "steps",
|
| 33 |
"logging_first_step": false,
|
| 34 |
"logging_steps": 1,
|
| 35 |
"logging_nan_inf_filter": true,
|
| 36 |
"save_strategy": "steps",
|
| 37 |
-
"save_steps":
|
| 38 |
"save_total_limit": 1,
|
| 39 |
"save_safetensors": false,
|
| 40 |
"save_on_each_node": false,
|
|
@@ -59,11 +59,11 @@
|
|
| 59 |
"tpu_metrics_debug": false,
|
| 60 |
"debug": [],
|
| 61 |
"dataloader_drop_last": false,
|
| 62 |
-
"eval_steps":
|
| 63 |
"dataloader_num_workers": 0,
|
| 64 |
"dataloader_prefetch_factor": null,
|
| 65 |
"past_index": -1,
|
| 66 |
-
"run_name": "DataDreamer - Train StyleDistance Model",
|
| 67 |
"disable_tqdm": true,
|
| 68 |
"remove_unused_columns": false,
|
| 69 |
"label_names": null,
|
|
|
|
| 1 |
{
|
| 2 |
+
"output_dir": "./output/train-wegmann--styledistance-model/_checkpoints",
|
| 3 |
"overwrite_output_dir": false,
|
| 4 |
"do_train": false,
|
| 5 |
"do_eval": true,
|
| 6 |
"do_predict": false,
|
| 7 |
"evaluation_strategy": "steps",
|
| 8 |
"prediction_loss_only": false,
|
| 9 |
+
"per_device_train_batch_size": 16,
|
| 10 |
+
"per_device_eval_batch_size": 16,
|
| 11 |
"per_gpu_train_batch_size": null,
|
| 12 |
"per_gpu_eval_batch_size": null,
|
| 13 |
+
"gradient_accumulation_steps": 8,
|
| 14 |
"eval_accumulation_steps": 1,
|
| 15 |
"eval_delay": 0,
|
| 16 |
+
"learning_rate": 0.001,
|
| 17 |
"weight_decay": 0.01,
|
| 18 |
"adam_beta1": 0.9,
|
| 19 |
"adam_beta2": 0.999,
|
| 20 |
"adam_epsilon": 1e-08,
|
| 21 |
"max_grad_norm": 1.0,
|
| 22 |
+
"num_train_epochs": 20,
|
| 23 |
"max_steps": -1,
|
| 24 |
"lr_scheduler_type": "linear",
|
| 25 |
"lr_scheduler_kwargs": {},
|
|
|
|
| 28 |
"log_level": "passive",
|
| 29 |
"log_level_replica": "warning",
|
| 30 |
"log_on_each_node": true,
|
| 31 |
+
"logging_dir": "./output/train-wegmann--styledistance-model/_checkpoints/runs/Jul21_05-54-20_nlpgpu04.seas.upenn.edu",
|
| 32 |
"logging_strategy": "steps",
|
| 33 |
"logging_first_step": false,
|
| 34 |
"logging_steps": 1,
|
| 35 |
"logging_nan_inf_filter": true,
|
| 36 |
"save_strategy": "steps",
|
| 37 |
+
"save_steps": 200,
|
| 38 |
"save_total_limit": 1,
|
| 39 |
"save_safetensors": false,
|
| 40 |
"save_on_each_node": false,
|
|
|
|
| 59 |
"tpu_metrics_debug": false,
|
| 60 |
"debug": [],
|
| 61 |
"dataloader_drop_last": false,
|
| 62 |
+
"eval_steps": 200,
|
| 63 |
"dataloader_num_workers": 0,
|
| 64 |
"dataloader_prefetch_factor": null,
|
| 65 |
"past_index": -1,
|
| 66 |
+
"run_name": "DataDreamer - Train Wegmann + StyleDistance Model",
|
| 67 |
"disable_tqdm": true,
|
| 68 |
"remove_unused_columns": false,
|
| 69 |
"label_names": null,
|