18907305772/FuseAI

Issues

• about gpt-4-0125-preview reference answer

  #21 opened 4 months ago by duguodong7
  4

• can use Qwen1.5-7B-Chat ?

  #13 opened 4 months ago by 18600709862
  11

• Starling-LM-7B-alpha tokenizer issues

  #19 opened 4 months ago by duguodong7
  4

• Evaluation

  #20 opened 5 months ago by ZLKong
  2

• Out of Memory Issue with OpenLLaMA-7B in Default FuseLLM Setting on A100 (80G)

  #16 opened 5 months ago by runtsang
  3

• how should I load minipile

  #15 opened 8 months ago by BlueCestbon
  0

• 使用fastchat加载对话模板是什么？

  #14 opened 9 months ago by BoFan-tunning
  1

• Purpose for the Split long text step

  #12 opened 10 months ago by ZLKong
  2

• Releasing FuseLLM for Korean

  #11 opened 10 months ago by sigridjineth
  0

• Encountering NaN grad_norm and loss values when training with DeepSpeed and OrionForCausalLM model

  #9 opened 10 months ago by sigridjineth
  19

• Out of Memory Issue with Blending for 14B Base Model

  #10 opened 10 months ago by sigridjineth
  3

• Getting flash_attn_2_cuda.cpython-310-x86_64-linux-gnu.so error

  #7 opened 10 months ago by sigridjineth
  3

• minipile_split issue

  #8 opened 10 months ago by Arbor334
  4

• Regarding MiniPile dataset splitting

  #5 opened a year ago by ZLKong
  6

• Comparison with merging LLaMA-2 CLM and LLaMA-2

  #4 opened a year ago by leegao
  1

• KeyError: 'per_step_logits' when running token_alignment.py

  #6 opened a year ago by ZLKong
  3

• Any plan to release the training code?

  #1 opened a year ago by VimukthiRandika1997
  2