OpenGVLab/EfficientQAT

Issues

• Unable to reproduce the result in paper for 4 bit wikitext perplexity

  #24 opened 11 days ago by huoshenlaile
  1

• Is it possible to run e2e-qp process on a single 4090?

  #23 opened a month ago by sihouzi21c
  2

• Can the results of the quantification process be saved?

  #17 opened 2 months ago by QB-Chen
  1

• RuntimeError: Triton Error [CUDA]: device kernel image is invalid

  #20 opened 2 months ago by Niko-zyf
  1

• any experiments on qwen2-7b-instruct?

  #22 opened 2 months ago by brisker
  1

• Is 7B llama speed expected to be slow?

  #19 opened 3 months ago by w32zhong
  2

• Reproduce Llama3 8B Instruct results

  #21 opened 2 months ago by gdsaikrishna
  2

• Quip# inference speed in your paper is incorrect?

  #9 opened 3 months ago by sankeerth95
  2

• why model output is too much slow?

  #18 opened 3 months ago by dahwin
  1

• Question about calib datasets.

  #16 opened 3 months ago by mxjmtxrm
  3

• GGUF

  #3 opened 4 months ago by maxim-saplin
  8

• block_ap过程中似乎不能使用多卡并行，老是爆显存

  #15 opened 3 months ago by QB-Chen
  1

• Cohere Command R Plus

  #10 opened 3 months ago by Khaledhesham
  1

• Data arrangement

  #7 opened 3 months ago by yancaoweidaode
  4

• How to do e2e_qp with multi-GPU?

  #14 opened 3 months ago by mxjmtxrm
  1

• DATA FOR TRAINING

  #12 opened 3 months ago by LiMa-cas
  1

• Loss is nan

  #6 opened 3 months ago by mxjmtxrm
  4

• Reproduce Llama2-7b

  #13 opened 3 months ago by laomao0
  5

• Evaluation pipeline points to missing files

  #11 opened 4 months ago by snps-tonatiuh
  2

• question about real quant

  #4 opened 4 months ago by mxjmtxrm
  6

• Can not reproduce the results

  #8 opened 4 months ago by LiuSiQi-TJ
  2

• Any experiments on w4a16 with group_size=-1( no group)?

  #2 opened 4 months ago by brisker
  4

• dimension mismatch error.

  #5 opened 4 months ago by mxjmtxrm
  0