chaoyi-wu/Finetune_LLAMA

Issues

• 如何实现多节点fsdp

  #10 opened a year ago by boyue-jiang
  1

• test_sample.py运行报错，貌似是device不一样，untimeError: Expected all tensors to be on the same device, but found at least two devices, cpu and cuda:0! (when checking argument for argument index in method wrapper_CUDA_gather)

  #3 opened 2 years ago by WUHU-G
  2

• 你好，请问最低需要什么等级的显卡呢，必须A100吗，我的显存目前完全不够

  #5 opened a year ago by TangTZscut
  3

• 微调数据集切换成自己的中文数据集之后出现报错：RuntimeError: CUDA error: CUBLAS_STATUS_NOT_INITIALIZED when calling `cublasCreate(handle)`

  #9 opened a year ago by Doufanfan
  0

• 4090-24G 是否能微调呢

  #8 opened a year ago by zhongxiangboy
  1

• 咱们这个是否支持出LLAMA之外的其他模型呢

  #7 opened a year ago by xsailor511
  1

• 使用苹果Mac Studio 192G微调怎么样，性能速度是否推荐呢？

  #6 opened a year ago by xsailor511
  0

• convert_to_ds_params.py doesn't generate tokenizer

  #4 opened 2 years ago by tammypi
  3

• attention mask for different documents in dataset chunk

  #2 opened 2 years ago by waterhorse1
  3

• A100 Memory

  #1 opened 2 years ago by HuuuNan
  1