XCF-Mike
XCF-Mike
请问解决了吗，我遇到了类似问题 
> 没有CUDA 12.1的对应版本 安装了 flash_attn-2.6.3+cu123torch2.4cxx11abiTRUE-cp310-cp310-linux_x86_64.whl 但是报错 ImportError: /root/anaconda3/envs/loramoe/lib/python3.10/site-packages/flash_attn_2_cuda.cpython-310-x86_64-linux-gnu.so: undefined symbol: _ZN3c105ErrorC2ENS_14SourceLocationENSt7__cxx1112basic_stringIcSt11char_traitsIcESaIcEEE ImportError: FlashAttention-2 is not installed correctly. Please check the usage in https://github.com/Dao-AILab/flash-attention for more details. 我应该安装哪个版本呢？需要重新安装匹配的torch吗 请问你找到正确的版本了吗
Hello, I encountered the same issue when using the Lora finetune model for mmbench evaluation  did you solve it
> Hi, 是的 谢谢您的回复，还想再请问一下，我现在使用8张80G的H800推理InternVL3-78B，但是还是发生了OOM问题，使用的脚本为主页quick start中以InternVL2.5-8B为例子的脚本
> ### Model Series > Qwen2.5 > > ### What are the models used? > Qwen2.5-7B-Instruct-fp16-gguf > > ### What is the scenario where the problem happened? > ollama >...