xwc
xwc
> 参考 > > - issue: https://github.com/hiyouga/LLaMA-Factory/issues/226 > > - code: https://github.com/hiyouga/LLaMA-Factory/blob/182b42504399d2755897b9737db1d36655a0fa50/src/llmtuner/dsets/preprocess.py#L50 > > > > > > 在代码中,`labels`列表的构建方式如下: > > > > ```python > > labels += [IGNORE_INDEX] *...
机器资源:2节点16卡H20 命令: ```bash nnodes=2 nproc_per_node=8 export CUDA_LAUNCH_BLOCKING=1 export PYTORCH_CUDA_ALLOC_CONF=expandable_segments:True MODEL_PATH=$(readlink -f output/models/Qwen3-VL-32B-Instruct/sft) CUDA_VISIBLE_DEVICES=0,1,2,3,4,5,6,7 \ NPROC_PER_NODE=$nproc_per_node \ PYTORCH_CUDA_ALLOC_CONF=expandable_segments:True \ NNODES=$nnodes \ swift rlhf \ --rlhf_type grpo \ --model $MODEL_PATH \ --external_plugins...