jcyao

运行generate.sh加载模型正常，gradio发送消息报错，请问哪里出了问题吗？bitsandbytes有bug？ File "/home/yjc/LaWGPT/src/generate.py", line 131, in generate_with_callback model.generate(**kwargs) File "/home/yjc/anaconda3/lib/python3.9/site-packages/peft/peft_model.py", line 580, in generate return self.base_model.generate(**kwargs) File "/home/yjc/anaconda3/lib/python3.9/site-packages/torch/autograd/grad_mode.py", line 27, in decorate_context return func(*args, **kwargs) File "/home/yjc/anaconda3/lib/python3.9/site-packages/transformers/generation/utils.py", line 1604, in...

## error log | 日志或报错信息 | ログ 使用 pnnx 转 ncnn 模型后动态维度轴大小始终是1 ## model | 模型 | モデル 1. original model [nmt_enzh_enc.zip](https://github.com/Tencent/ncnn/files/13496853/nmt_enzh_enc.zip) ## how to reproduce | 复现步骤 | 再現方法...

请问，llama13b模型微调训练机器配置和时长大概是什么样的？

编译最后一步链接生成可执行程序的时候报 /usr/bin/ld: /root/Lightseq-ARM/3rdparty/protobuf-3.21.4/libs/linux-aarch64/libprotobuf-lite.a(arena.cc.o): relocation R_AARCH64_ADR_PREL_PG_HI21 against symbol `_ZNSt8ios_base4InitD1Ev@@GLIBCXX_3.4' which may bind externally can not be used when making a shared object; recompile with -fPIC libprotobuf-lite.a给错了吗

模型训练是用lightseq吗，能否提供训练文档，谢谢