AI开发平台MODELARTS-MiniCPM-V2.6基于DevServer适配PyTorch NPU训练指导（6.3.909）:Step7 lora微调

时间：2024-11-15 09:44:20

AI开发平台MODELARTS

Step7 lora微调

bash finetune_lora.sh

修改模型权重路径${model_path}，保持其余参数一致。脚本参数设置如下：

#!/bin/bash 
GPUS_PER_NODE=8
NNODES=1
NODE_RANK=0
MASTER_ADDR=localhost 
MASTER_PORT=6001
export PYTORCH_NPU_ALLOC_CONF=expandable_segments:True  

MODEL=${mdoel_path} # or openbmb/MiniCPM-V-2, openbmb/MiniCPM-Llama3-V-2_5
# ATTENTION: specify the path to your training data, which should be a json file consisting of a list of conversations.
# See the section for finetuning in README for more information.
DATA="coco2014_train.json"
EVAL_DATA="coco2014_val.json"
LLM_TYPE="qwen2"  
# if use openbmb/MiniCPM-V-2, please set LLM_TYPE=minicpm#if use openbmb/MiniCPM-Llama3-V-2_5, please set LLM_TYPE=llama3 MODEL_MAX_Length=2048 
# if conduct multi-images sft, please set MODEL_MAX_Length=4096 

DISTRIBUTED_ARGS="
     --nproc_per_node $GPUS_PER_NODE \
     --nnodes $NNODES \
     --node_rank $NODE_RANK \
     --master_addr $MASTER_ADDR \
     --master_port $MASTER_PORT 
"
torchrun $DISTRIBUTED_ARGS finetune.py  \
    --model_name_or_path $MODEL \
    --llm_type $LLM_TYPE \
    --data_path $DATA \
    --eval_data_path $EVAL_DATA \
    --remove_unused_columns false \
    --label_names "labels" \
    --prediction_loss_only false \
    --bf16 true \
    --bf16_full_eval true \
    --fp16 false \
    --fp16_full_eval false \
    --do_train \
    --do_eval \
    --num_train_epochs 1 \
    --tune_vision true \
    --tune_llm false \
    --use_lora true \
    --lora_target_modules "llm\..*layers\.\d+\.self_attn\.(q_proj|k_proj|v_proj|o_proj)" \
    --model_max_length $MODEL_MAX_Length \
    --max_slice_nums 9 \
    --max_steps 1000 \
    --eval_steps 10000 \
    --output_dir output/output__lora \
    --logging_dir output/output_lora \
    --logging_strategy "steps" \
    --per_device_train_batch_size 1 \
    --per_device_eval_batch_size 1 \
    --gradient_accumulation_steps 1 \
    --evaluation_strategy "steps" \
    --save_strategy "steps" \
    --save_steps 10000 \
    --save_total_limit 10 \
    --learning_rate 1e-6 \
    --weight_decay 0.1 \
    --adam_beta2 0.95 \
    --warmup_ratio 0.01 \
    --lr_scheduler_type "cosine" \
    --logging_steps 1 \
    --gradient_checkpointing true \
    --deepspeed ds_config_zero2.json \
    --report_to "tensorboard

上一篇：AI开发平台MODELARTS-MiniCPM-V2.6基于DevServer适配PyTorch NPU训练指导（6.3.909）:Step6 监督微调

下一篇：AI开发平台MODELARTS-MiniCPM-V2.6基于DevServer适配PyTorch NPU训练指导（6.3.909）:Step6 监督微调