Update config.json
Browse files- config.json +5 -17
config.json
CHANGED
@@ -1,5 +1,6 @@
|
|
1 |
{
|
2 |
"model_type": "transformer",
|
|
|
3 |
"task_type": "CAUSAL_LM",
|
4 |
"peft_config": {
|
5 |
"lora_alpha": 16,
|
@@ -9,22 +10,9 @@
|
|
9 |
"task_type": "CAUSAL_LM",
|
10 |
"target_modules": ["q_proj", "k_proj", "v_proj", "o_proj", "gate_proj"]
|
11 |
},
|
12 |
-
"
|
13 |
-
"
|
14 |
-
"
|
15 |
-
"
|
16 |
-
"gradient_accumulation_steps": 1,
|
17 |
-
"optim": "paged_adamw_32bit",
|
18 |
-
"save_steps": 50,
|
19 |
-
"logging_steps": 1,
|
20 |
-
"learning_rate": 2e-4,
|
21 |
-
"weight_decay": 0.001,
|
22 |
-
"fp16": false,
|
23 |
-
"bf16": false,
|
24 |
-
"max_grad_norm": 0.3,
|
25 |
-
"max_steps": -1,
|
26 |
-
"warmup_ratio": 0.03,
|
27 |
-
"group_by_length": true,
|
28 |
-
"lr_scheduler_type": "constant"
|
29 |
}
|
30 |
}
|
|
|
1 |
{
|
2 |
"model_type": "transformer",
|
3 |
+
"trust_remote_code": true,
|
4 |
"task_type": "CAUSAL_LM",
|
5 |
"peft_config": {
|
6 |
"lora_alpha": 16,
|
|
|
10 |
"task_type": "CAUSAL_LM",
|
11 |
"target_modules": ["q_proj", "k_proj", "v_proj", "o_proj", "gate_proj"]
|
12 |
},
|
13 |
+
"quantization_config": {
|
14 |
+
"load_in_4bit": true,
|
15 |
+
"torch_dtype": "torch.bfloat16",
|
16 |
+
"device_map": "auto"
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
17 |
}
|
18 |
}
|