File size: 1,026 Bytes
7ed0fb5 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 |
./move.sh
cd src/r1-v
export DEBUG_MODE="true" # Enable Debug if you want to see the rollout of model during RL
export LOG_PATH="./debug_log_2b.txt"
CUDA_VISIBLE_DEVICES=0,1,2,3 torchrun --nproc_per_node="4" \
--nnodes="1" \
--node_rank="0" \
--master_addr="127.0.0.1" \
--master_port="12349" \
src/open_r1/sft_video.py \
--output_dir "./log/Qwen2.5-VL-3B-Video-3B-cot-sft" \
--model_name_or_path "Qwen/Qwen2.5-VL-3B-Instruct" \
--dataset_name "./Video-R1-data/Video-R1-COT-165k.json" \
--deepspeed local_scripts/zero2_1.json \
--per_device_train_batch_size 1 \
--gradient_accumulation_steps 2 \
--learning_rate 1e-6 \
--logging_steps 1 \
--bf16 \
--report_to wandb \
--gradient_checkpointing true \
--attn_implementation flash_attention_2 \
--num_train_epochs 1 \
--run_name Qwen2.5-VL-3B-Video-cot-sft \
--save_steps 1000 \
--max_grad_norm 5 \
--save_only_model true \
python /cq_1/share_1603164/user/zongxia/workspace/gpu_burn.py |