Create config.json
Browse files- config.json +47 -0
config.json
ADDED
|
@@ -0,0 +1,47 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"model_name": "unsloth/Phi-3-mini-4k-instruct",
|
| 3 |
+
"model_type": "phi-3",
|
| 4 |
+
|
| 5 |
+
"max_seq_length": 2048,
|
| 6 |
+
"load_in_4bit": true,
|
| 7 |
+
"lora_config": {
|
| 8 |
+
"r": 16,
|
| 9 |
+
"lora_alpha": 16,
|
| 10 |
+
"target_modules": [
|
| 11 |
+
"q_proj",
|
| 12 |
+
"k_proj",
|
| 13 |
+
"v_proj",
|
| 14 |
+
"o_proj",
|
| 15 |
+
"gate_proj",
|
| 16 |
+
"up_proj",
|
| 17 |
+
"down_proj"
|
| 18 |
+
],
|
| 19 |
+
"lora_dropout": 0,
|
| 20 |
+
"bias": "none",
|
| 21 |
+
"use_gradient_checkpointing": "unsloth",
|
| 22 |
+
"random_state": 3407,
|
| 23 |
+
"use_rslora": false,
|
| 24 |
+
"loftq_config": null
|
| 25 |
+
},
|
| 26 |
+
"training_arguments": {
|
| 27 |
+
"per_device_train_batch_size": 2,
|
| 28 |
+
"gradient_accumulation_steps": 4,
|
| 29 |
+
"warmup_steps": 5,
|
| 30 |
+
"max_steps": 60,
|
| 31 |
+
"learning_rate": 2e-4,
|
| 32 |
+
"fp16": true,
|
| 33 |
+
"bf16": false,
|
| 34 |
+
"logging_steps": 1,
|
| 35 |
+
"optim": "adamw_8bit",
|
| 36 |
+
"weight_decay": 0.01,
|
| 37 |
+
"lr_scheduler_type": "linear",
|
| 38 |
+
"seed": 3407,
|
| 39 |
+
"output_dir": "outputs"
|
| 40 |
+
},
|
| 41 |
+
"data_config": {
|
| 42 |
+
"dataset_text_field": "text",
|
| 43 |
+
"max_seq_length": 2048,
|
| 44 |
+
"dataset_num_proc": 2,
|
| 45 |
+
"packing": false
|
| 46 |
+
}
|
| 47 |
+
}
|