AetherCortex/Llama-X

Issues

• support llama 3 models

  #34 opened 6 months ago by chauncygu
  0

• How to fix unstable loss. I am using wizardlm or Llama-X training code with vicuna style chat format for fine-tuning Llama-2-7b-hf model.

  #32 opened a year ago by apt-team-018
  1

• utils.py and openai==0.8.0 TypeError: 'type' object is not subscriptable

  #33 opened a year ago by zql022
  1

• Learning rate fixed at 0 during training via DeepSpeed

  #30 opened a year ago by anmolagarwal999
  2

• About llama-2-70B fine-tuning

  #31 opened a year ago by RickMeow
  0

• Have you ever tried PEFT on LLaMA?

  #29 opened a year ago by ZetangForward
  0

• How to run 13B LLaMA model for Full-parameters SFT ?

  #27 opened 2 years ago by jacklanda
  0

• how big is your RAM

  #15 opened 2 years ago by Sorezza
  6

• there is a endless loop in convert_tokens_to_ids(self.unk_token) and self._convert_token_to_id_with_added_voc(tokens)

  #23 opened 2 years ago by 2018211801
  0

• About the training strategy

  #3 opened 2 years ago by SparkJiao
  7

• good in llama-7b,but bad in llama-13B ?

  #22 opened 2 years ago by huiyangzhou
  1

• About Llama-X and Alpaca repo

  #20 opened 2 years ago by haorannlp
  0

• only load lm_head.weight and embed_tokens.weight parameters

  #19 opened 2 years ago by qwjaskzxl
  0

• extremely bad performance

  #17 opened 2 years ago by 152334H
  0

• Demo is dead

  #16 opened 2 years ago by fredi-python
  0

• improve LLaMA for visual understanding like GPT-4

  #13 opened 2 years ago by feizc
  1

• RuntimeError: CUDA out of memory.

  #12 opened 2 years ago by OpenSource-fan
  3

• Why use offload_param in CPU？

  #9 opened 2 years ago by xesdiny
  1

• Need optimization for mps

  #10 opened 2 years ago by yxKryptonite
  0

• Concern on the language

  #2 opened 2 years ago by zhhongzhi
  3

• Does this repository support training of the GPT series?

  #1 opened 2 years ago by Sierkinhane
  2