Training config
Browse files- training_config.json +35 -0
training_config.json
ADDED
|
@@ -0,0 +1,35 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"trainer": {
|
| 3 |
+
"evaluation_strategy": "steps",
|
| 4 |
+
"per_device_train_batch_size": 1,
|
| 5 |
+
"per_device_eval_batch_size": 1,
|
| 6 |
+
"gradient_accumulation_steps": 128,
|
| 7 |
+
"eval_steps": 50,
|
| 8 |
+
"save_steps": 50,
|
| 9 |
+
"logging_steps": 5,
|
| 10 |
+
"learning_rate": 0.0003,
|
| 11 |
+
"num_train_epochs": 5,
|
| 12 |
+
"lr_scheduler_type": "cosine",
|
| 13 |
+
"warmup_steps": 30,
|
| 14 |
+
"fp16": false,
|
| 15 |
+
"bf16": true,
|
| 16 |
+
"torch_compile": false,
|
| 17 |
+
"optim": "adamw_torch"
|
| 18 |
+
},
|
| 19 |
+
"lora": {
|
| 20 |
+
"r": 16,
|
| 21 |
+
"lora_alpha": 16,
|
| 22 |
+
"lora_dropout": 0.05,
|
| 23 |
+
"bias": "none",
|
| 24 |
+
"target_modules": ["q_proj", "v_proj", "k_proj", "o_proj"],
|
| 25 |
+
"task_type": "CAUSAL_LM"
|
| 26 |
+
},
|
| 27 |
+
"load_in_8bit": true,
|
| 28 |
+
"only_target_loss": true,
|
| 29 |
+
"mode": "chat",
|
| 30 |
+
"templates_path": "internal_prompts/saiga_v2.json",
|
| 31 |
+
"model_name": "models/llama2-13b",
|
| 32 |
+
"model_type": "causal",
|
| 33 |
+
"max_tokens_count": 2000
|
| 34 |
+
}
|
| 35 |
+
|