vtchg002

我没用DOCKER用本地电脑按说明搭了环境，测试torch.cuda.device_count()为1，显卡兼容没问题。但是又提示这样的错误： E:\eva\EVA-main\src>torchrun --master_port 1234 --nproc_per_node 1 E:/eva/EVA-main/src/eva_interactive.py --model-config E:/eva/EVA-main/src/configs/model/eva2.0_model_config.json --model-parallel-size 1 --load E:/eva/EVA-main/checkpoints/eva2.0 -- no_load_strict --distributed-backend nccl --weight-decay 1e-2 --clip-grad 1.0 --tokenizer-path E:/eva/EVA-main/bpe_dialog_new --temperature 0.9 --top_k 0 --top_p 0.9 --num-beams...

用huggingface没有其它报错了，但是提示显卡内存不够，我是2060 6G内存，这个跑起来最低需要多少G的内存呢？ RuntimeError: CUDA out of memory. Tried to allocate 16.00 MiB (GPU 0; 6.00 GiB total capacity; 5.32 GiB already allocated; 0 bytes free; 5.32 GiB reserved in total...

getattr(torch.optim,optimizer_name,None) is not None