Create trainer_config.json
Browse files- trainer_config.json +43 -0
trainer_config.json
ADDED
|
@@ -0,0 +1,43 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"output_dir": "./mineral-nano-1",
|
| 3 |
+
"overwrite_output_dir": true,
|
| 4 |
+
"do_train": true,
|
| 5 |
+
"do_eval": true,
|
| 6 |
+
"evaluation_strategy": "steps",
|
| 7 |
+
"eval_steps": 500,
|
| 8 |
+
"per_device_train_batch_size": 8,
|
| 9 |
+
"per_device_eval_batch_size": 8,
|
| 10 |
+
"gradient_accumulation_steps": 4,
|
| 11 |
+
"learning_rate": 5e-5,
|
| 12 |
+
"weight_decay": 0.01,
|
| 13 |
+
"adam_beta1": 0.9,
|
| 14 |
+
"adam_beta2": 0.999,
|
| 15 |
+
"adam_epsilon": 1e-8,
|
| 16 |
+
"max_grad_norm": 1.0,
|
| 17 |
+
"num_train_epochs": 3,
|
| 18 |
+
"max_steps": -1,
|
| 19 |
+
"lr_scheduler_type": "cosine",
|
| 20 |
+
"warmup_ratio": 0.1,
|
| 21 |
+
"warmup_steps": 0,
|
| 22 |
+
"logging_dir": "./logs",
|
| 23 |
+
"logging_strategy": "steps",
|
| 24 |
+
"logging_steps": 100,
|
| 25 |
+
"save_strategy": "steps",
|
| 26 |
+
"save_steps": 1000,
|
| 27 |
+
"save_total_limit": 3,
|
| 28 |
+
"seed": 42,
|
| 29 |
+
"fp16": false,
|
| 30 |
+
"bf16": true,
|
| 31 |
+
"dataloader_num_workers": 4,
|
| 32 |
+
"load_best_model_at_end": true,
|
| 33 |
+
"metric_for_best_model": "eval_loss",
|
| 34 |
+
"greater_is_better": false,
|
| 35 |
+
"group_by_length": false,
|
| 36 |
+
"report_to": ["tensorboard"],
|
| 37 |
+
"ddp_find_unused_parameters": false,
|
| 38 |
+
"optim": "adamw_torch",
|
| 39 |
+
"gradient_checkpointing": true,
|
| 40 |
+
"gradient_checkpointing_kwargs": {
|
| 41 |
+
"use_reentrant": false
|
| 42 |
+
}
|
| 43 |
+
}
|