cuda_result.txt

#!/bin/bash
HIP_VISIBLE_DEVICES=4,5,6,7

GPUS_PER_NODE=4
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 false \
    --bf16_full_eval false \
    --fp16 true \
    --fp16_full_eval true \
    --do_train \
    --do_eval \
    --tune_vision true \
    --tune_llm false \
    --use_lora true \
    --lora_target_modules "llm\..*layers\.\d+\.self_attn\.(q_proj|k_proj)" \
    --model_max_length 2048 \
    --max_slice_nums 9 \
    --max_steps 100 \
    --eval_steps 10 \
    --output_dir /home/wanglch/projects/saves/MiniCPM-Llama3-V-2_5/lora_train_dtk \
    --logging_dir /home/wanglch/projects/saves/MiniCPM-Llama3-V-2_5/lora_train_dtk \
    --logging_strategy "steps" \
    --per_device_train_batch_size 2 \
    --per_device_eval_batch_size 1 \
    --gradient_accumulation_steps 1 \
    --evaluation_strategy "steps" \
    --save_strategy "steps" \
    --save_steps 100 \
    --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" # wandb


{'train_runtime': 305.4486, 'train_samples_per_second': 2.619, 'train_steps_per_second': 0.327, 'train_loss': 0.06511555195844267, 'epoch': 100.0}
{'eval_loss': 0.389315664768219, 'eval_runtime': 0.2161, 'eval_samples_per_second': 4.628, 'eval_steps_per_second': 4.628, 'epoch': 100.0}