princeton-nlp/SimPO

Issues

• Question about RRHF

  #5 opened 24 days ago by SihengLi99
  2

• QLORA -4bit

  #14 opened 24 days ago by seroetr
  1

• Unable to reproduce the results of DPO

  #23 opened 2 months ago by AGTSAAA
  3

• simloss has no length normalization in simpo loss

  #32 opened a month ago by hjc3613
  1

• Yi-34B + Simpo + Full + Novel writing Task, fine-tuning results are unsatisfactory

  #33 opened a month ago by onlyfish79
  0

• Has anyone reproduced the results in the paper quantitatively?

  #21 opened 2 months ago by binzhwang
  5

• Unable to reproduce the results of SFT

  #27 opened 2 months ago by yujiaw98
  3

• Why is the AlpacaEval2 score of meta-llama/Meta-LLama-3-8B-Instruct in the paper higher than that on the leaderboard?

  #31 opened a month ago by eugene-yh
  1

• Usage on Custom Dataset

  #28 opened 2 months ago by ViperVille007
  1

• DPOTrainer.get_batch_logps() got an unexpected keyword argument 'average_log_prob'

  #29 opened 2 months ago by RAY2L
  2

• Question about apply_chat_template to prompt

  #30 opened a month ago by EganGu
  0

• Change of Mistral Chat template

  #26 opened 2 months ago by mianzhang
  2

• Your comparsion is unfair by using different chat templates

  #24 opened 2 months ago by AGTSAAA
  2

• Training leads to model collapse

  #12 opened 2 months ago by Ricardokevins
  13

• TRL version should be 0.8.6

  #25 opened 2 months ago by blakechi
  2

• Unable to reproduce the mt bench results in the paper

  #22 opened 2 months ago by lmx760581375
  1

• Length normalization in DPO and other variants

  #20 opened 2 months ago by yakazimir
  2

• For the Instruct setup, why do different models require different training datasets? Can the same dataset be used?

  #19 opened 2 months ago by qiuwenbogdut
  3

• ValueError: Unknown split "train". Should be one of ['train_iteration_1', 'test_iteration_1', 'train_iteration_2', 'test_iteration_2',

  #18 opened 2 months ago by qiuwenbogdut
  2

• the outputs of reproduced model has "<|start_header_id|>assistant<|end_header_id|>" at beginning

  #17 opened 2 months ago by binzhwang
  6

• accelerator.prepare() CUDA out of memory

  #10 opened 2 months ago by zouce
  2

• Question about the length-normalization

  #13 opened 2 months ago by yujiaw98
  1

• Confusing Code logic

  #15 opened 2 months ago by syboomsy
  2

• On Length Normalization in the Code

  #4 opened 2 months ago by jc-ryan
  4

• Mismatch of results

  #9 opened 2 months ago by AGTSAAA
  7

• About the version of alignment-handbook

  #6 opened 2 months ago by lucasliunju
  3

• Length normalization

  #7 opened 2 months ago by Michelleable
  0

• Repeated Addition of Assistant Turn in Prompt/Chosen/Rejected Text Using `apply_chat_template`

  #8 opened 2 months ago by iseesaw
  1

• Integrate upstream in alignment-handbook

  #2 opened 2 months ago by BramVanroy
  1

• Upstream `SimPOTrainer` to TRL

  #3 opened 2 months ago by philschmid
  1