41 lines
1.1 KiB
JSON
41 lines
1.1 KiB
JSON
|
{
|
||
|
|
"model_name": "meta-llama/Llama-2-7b-hf",
|
||
|
|
"data_path": "RoversX/Samantha-EN-CN-Converted-Dataset-V1",
|
||
|
|
"train_split": "train",
|
||
|
|
"valid_split": null,
|
||
|
|
"text_column": "text",
|
||
|
|
"huggingface_token": null,
|
||
|
|
"learning_rate": 0.0002,
|
||
|
|
"num_train_epochs": 1,
|
||
|
|
"train_batch_size": 16,
|
||
|
|
"eval_batch_size": 4,
|
||
|
|
"warmup_ratio": 0.1,
|
||
|
|
"gradient_accumulation_steps": 1,
|
||
|
|
"optimizer": "adamw_torch",
|
||
|
|
"scheduler": "linear",
|
||
|
|
"weight_decay": 0.0,
|
||
|
|
"max_grad_norm": 1.0,
|
||
|
|
"seed": 42,
|
||
|
|
"add_eos_token": false,
|
||
|
|
"block_size": -1,
|
||
|
|
"use_peft": true,
|
||
|
|
"lora_r": 16,
|
||
|
|
"lora_alpha": 32,
|
||
|
|
"lora_dropout": 0.05,
|
||
|
|
"training_type": "generic",
|
||
|
|
"train_on_inputs": false,
|
||
|
|
"logging_steps": -1,
|
||
|
|
"project_name": "my-llm",
|
||
|
|
"evaluation_strategy": "epoch",
|
||
|
|
"save_total_limit": 1,
|
||
|
|
"save_strategy": "epoch",
|
||
|
|
"auto_find_batch_size": false,
|
||
|
|
"fp16": false,
|
||
|
|
"push_to_hub": true,
|
||
|
|
"use_int8": false,
|
||
|
|
"model_max_length": 1024,
|
||
|
|
"repo_id": "RoversX/llama-2-7b-hf-small-shards-Samantha-V1-SFT",
|
||
|
|
"use_int4": true,
|
||
|
|
"trainer": "sft",
|
||
|
|
"target_modules": null
|
||
|
|
}
|