LLaMA-Factory/examples/README_zh.md

我们提供了多样化的大模型微调示例脚本。

请确保在 LLaMA-Factory 目录下执行下述命令。

目录

• 单 GPU LoRA 微调
• 单 GPU QLoRA 微调
• 多 GPU LoRA 微调
• 多 NPU LoRA 微调
• 多 GPU 全参数微调
• 合并 LoRA 适配器与模型量化
• 推理 LoRA 模型
• 杂项

示例

单 GPU LoRA 微调

（增量）预训练

CUDA_VISIBLE_DEVICES=0 llamafactory-cli train examples/lora_single_gpu/llama3_lora_pretrain.yaml

指令监督微调

CUDA_VISIBLE_DEVICES=0 llamafactory-cli train examples/lora_single_gpu/llama3_lora_sft.yaml

多模态指令监督微调

CUDA_VISIBLE_DEVICES=0 llamafactory-cli train examples/lora_single_gpu/llava1_5_lora_sft.yaml

奖励模型训练

CUDA_VISIBLE_DEVICES=0 llamafactory-cli train examples/lora_single_gpu/llama3_lora_reward.yaml

PPO 训练

CUDA_VISIBLE_DEVICES=0 llamafactory-cli train examples/lora_single_gpu/llama3_lora_ppo.yaml

DPO/ORPO/SimPO 训练

CUDA_VISIBLE_DEVICES=0 llamafactory-cli train examples/lora_single_gpu/llama3_lora_dpo.yaml

KTO 训练

CUDA_VISIBLE_DEVICES=0 llamafactory-cli train examples/lora_single_gpu/llama3_lora_kto.yaml

预处理数据集

对于大数据集有帮助，在配置中使用 tokenized_path 以加载预处理后的数据集。

CUDA_VISIBLE_DEVICES=0 llamafactory-cli train examples/lora_single_gpu/llama3_preprocess.yaml

在 MMLU/CMMLU/C-Eval 上评估

CUDA_VISIBLE_DEVICES=0 llamafactory-cli eval examples/lora_single_gpu/llama3_lora_eval.yaml

批量预测并计算 BLEU 和 ROUGE 分数

CUDA_VISIBLE_DEVICES=0 llamafactory-cli train examples/lora_single_gpu/llama3_lora_predict.yaml

单 GPU QLoRA 微调

基于 4/8 比特 Bitsandbytes 量化进行指令监督微调（推荐）

CUDA_VISIBLE_DEVICES=0 llamafactory-cli train examples/qlora_single_gpu/llama3_lora_sft_bitsandbytes.yaml

基于 4/8 比特 GPTQ 量化进行指令监督微调

CUDA_VISIBLE_DEVICES=0 llamafactory-cli train examples/qlora_single_gpu/llama3_lora_sft_gptq.yaml

基于 4 比特 AWQ 量化进行指令监督微调

CUDA_VISIBLE_DEVICES=0 llamafactory-cli train examples/qlora_single_gpu/llama3_lora_sft_awq.yaml

基于 2 比特 AQLM 量化进行指令监督微调

CUDA_VISIBLE_DEVICES=0 llamafactory-cli train examples/qlora_single_gpu/llama3_lora_sft_aqlm.yaml

多 GPU LoRA 微调

使用 Accelerate 进行单节点训练

CUDA_VISIBLE_DEVICES=0,1,2,3 llamafactory-cli train examples/lora_multi_gpu/llama3_lora_sft.yaml

使用 Accelerate 进行多节点训练

CUDA_VISIBLE_DEVICES=0,1,2,3 NNODES=2 RANK=0 MASTER_ADDR=192.168.0.1 MASTER_PORT=29500 llamafactory-cli train examples/lora_multi_gpu/llama3_lora_sft.yaml
CUDA_VISIBLE_DEVICES=0,1,2,3 NNODES=2 RANK=1 MASTER_ADDR=192.168.0.1 MASTER_PORT=29500 llamafactory-cli train examples/lora_multi_gpu/llama3_lora_sft.yaml

使用 DeepSpeed ZeRO-3 平均分配显存

CUDA_VISIBLE_DEVICES=0,1,2,3 llamafactory-cli train examples/lora_multi_gpu/llama3_lora_sft_ds.yaml

多 NPU LoRA 微调

使用 DeepSpeed ZeRO-0 训练

ASCEND_RT_VISIBLE_DEVICES=0,1,2,3 llamafactory-cli train examples/lora_multi_npu/llama3_lora_sft_ds.yaml

多 GPU 全参数微调

使用 DeepSpeed 进行单节点训练

CUDA_VISIBLE_DEVICES=0,1,2,3 llamafactory-cli train examples/full_multi_gpu/llama3_full_sft.yaml

使用 DeepSpeed 进行多节点训练

CUDA_VISIBLE_DEVICES=0,1,2,3 NNODES=2 RANK=0 MASTER_ADDR=192.168.0.1 MASTER_PORT=29500 llamafactory-cli train examples/full_multi_gpu/llama3_full_sft.yaml
CUDA_VISIBLE_DEVICES=0,1,2,3 NNODES=2 RANK=1 MASTER_ADDR=192.168.0.1 MASTER_PORT=29500 llamafactory-cli train examples/full_multi_gpu/llama3_full_sft.yaml

批量预测并计算 BLEU 和 ROUGE 分数

CUDA_VISIBLE_DEVICES=0,1,2,3 llamafactory-cli train examples/full_multi_gpu/llama3_full_predict.yaml

合并 LoRA 适配器与模型量化

合并 LoRA 适配器

注：请勿使用量化后的模型或 quantization_bit 参数来合并 LoRA 适配器。

CUDA_VISIBLE_DEVICES=0 llamafactory-cli export examples/merge_lora/llama3_lora_sft.yaml

使用 AutoGPTQ 量化模型

CUDA_VISIBLE_DEVICES=0 llamafactory-cli export examples/merge_lora/llama3_gptq.yaml

推理 LoRA 模型

使用 CUDA_VISIBLE_DEVICES=0,1 进行多卡推理。

使用命令行接口

CUDA_VISIBLE_DEVICES=0 llamafactory-cli chat examples/inference/llama3_lora_sft.yaml

使用浏览器界面

CUDA_VISIBLE_DEVICES=0 llamafactory-cli webchat examples/inference/llama3_lora_sft.yaml

启动 OpenAI 风格 API

CUDA_VISIBLE_DEVICES=0 llamafactory-cli api examples/inference/llama3_lora_sft.yaml

杂项

使用 GaLore 进行全参数训练

CUDA_VISIBLE_DEVICES=0 llamafactory-cli train examples/extras/galore/llama3_full_sft.yaml

使用 BAdam 进行全参数训练

CUDA_VISIBLE_DEVICES=0 llamafactory-cli train examples/extras/badam/llama3_full_sft.yaml

LoRA+ 微调

CUDA_VISIBLE_DEVICES=0 llamafactory-cli train examples/extras/loraplus/llama3_lora_sft.yaml

深度混合微调

CUDA_VISIBLE_DEVICES=0 llamafactory-cli train examples/extras/mod/llama3_full_sft.yaml

LLaMA-Pro 微调

bash examples/extras/llama_pro/expand.sh
CUDA_VISIBLE_DEVICES=0 llamafactory-cli train examples/extras/llama_pro/llama3_freeze_sft.yaml

FSDP+QLoRA 微调

bash examples/extras/fsdp_qlora/single_node.sh