Search results

Discrepancy in rm_scores Length for Mistral-7B-SPPO Iter1 Dataset

Status: Open.

#27 In uclaml/SPPO;

· srzhu97 opened on Oct 9, 2024

Release training dataset and loss log to help reproduce results

Status: Open.

#26 In uclaml/SPPO;

· srzhu97 opened on Oct 9, 2024

Added a new PR to allow generation on fewer than 8 GPUs

Status: Open.

#25 In uclaml/SPPO;

· aman2304 opened on Sep 2, 2024

DPO baseline implementation

Status: Open.

#22 In uclaml/SPPO;

· yesiam-png opened on Jul 23, 2024

SPPO Implementation on Axolotl!

Status: Open.

#21 In uclaml/SPPO;

· kaykyr opened on Jul 11, 2024

Scores and probability calcuations

Status: Open.

#15 In uclaml/SPPO;

· namdw opened on Jul 9, 2024

What's the package configuration for reproduce SPPO-Gemma-2?

Status: Open.

#14 In uclaml/SPPO;

· Jackory opened on Jul 8, 2024

Ranking speed & training hyperparameters

Status: Open.

#10 In uclaml/SPPO;

· skramer-dev opened on Jul 3, 2024

Is it normal the pipeline start with a huge loss ?

Status: Open.

#8 In uclaml/SPPO;

· qy1026 opened on Jul 2, 2024

Some packages' version are too old

Status: Open.

#7 In uclaml/SPPO;

· qy1026 opened on Jul 2, 2024

Which version of vllm should be installed

Status: Open.

#5 In uclaml/SPPO;

· xinghuang2050 opened on Jul 1, 2024

ShareGPT appending

Status: Open.

#4 In uclaml/SPPO;

· Kquant03 opened on Jun 30, 2024