微调1.5B OOM

[hackerhaiJu]

Reminder

• I have read the above rules and searched the existing issues.

System Info

llamafactory：0.9.2.dev0

Reproduction

两张2080TI，44G微调1.5B显存溢出，数据集3000条数据。是我配置的分布式deepspeed不对？

export CUDA_VISIBLE_DEVICES=2,3 && nohup llamafactory-cli train \
    --stage sft \
    --do_train True \
    --model_name_or_path /app/LLaMA-Factory/model/DeepSeek-R1-Distill-Qwen-1.5B \
    --preprocessing_num_workers 16 \
    --finetuning_type lora \
    --template deepseek3 \
    --flash_attn fa2 \
    --dataset_dir data \
    --dataset law_train \
    --cutoff_len 2048 \
    --learning_rate 3e-5 \
    --num_train_epochs 5 \
    --max_samples 100000 \
    --per_device_train_batch_size 1 \
    --gradient_accumulation_steps 8 \
    --lr_scheduler_type cosine \
    --max_grad_norm 1.0 \
    --logging_steps 5 \
    --save_steps 100 \
    --warmup_steps 0 \
    --packing False \
    --report_to none \
    --use_swanlab True \
    --output_dir saves/DeepSeek-R1-1.5B-Distill/lora/train_2025_law-6 \
    --bf16 True \
    --plot_loss True \
    --trust_remote_code True \
    --ddp_timeout 180000000 \
    --include_num_input_tokens_seen True \
    --optim adamw_torch \
    --lora_rank 8 \
    --lora_alpha 16 \
    --val_size 0.1 \
    --eval_strategy steps \
    --eval_steps 100 \
    --lora_dropout 0 \
    --loraplus_lr_ratio 16 \
    --lora_target all \
    --swanlab_project llamafactory \
    --swanlab_run_name deepseek-r1-1.5b_law-6 \
    --swanlab_mode local \
    --deepspeed cache/ds_z3_config.json > llama.log 2>&1 & disown && ps -ef |grep '/usr/local/bin/llamafactory-cli train' | grep -v 'grep' | awk '{print $2}' > llamafactory.pid

deepspeed

{
  "train_batch_size": "auto",
  "train_micro_batch_size_per_gpu": "auto",
  "gradient_accumulation_steps": "auto",
  "gradient_clipping": "auto",
  "zero_allow_untested_optimizer": true,
  "fp16": {
    "enabled": "auto",
    "loss_scale": 0,
    "loss_scale_window": 1000,
    "initial_scale_power": 16,
    "hysteresis": 2,
    "min_loss_scale": 1
  },
  "bf16": {
    "enabled": "auto"
  },
  "zero_optimization": {
    "stage": 3,
    "overlap_comm": true,
    "contiguous_gradients": true,
    "sub_group_size": 1000000000.0,
    "reduce_bucket_size": "auto",
    "stage3_prefetch_bucket_size": "auto",
    "stage3_param_persistence_threshold": "auto",
    "stage3_max_live_parameters": 1000000000.0,
    "stage3_max_reuse_distance": 1000000000.0,
    "stage3_gather_16bit_weights_on_model_save": true
  }
}

Others

No response