Commits

Commits on Jan 8, 2024

Update README.md (CarperAI#586 )
StellaAthena
committedJan 8, 2024

Commits on Oct 30, 2023

fix model state_dict retrieving in zero3 (CarperAI#576 )

Jingru
and
聂靖入
committedOct 30, 2023

Commits on Oct 26, 2023

fix(modeling_base): partial loading of a sharded checkpoint (CarperAI#578 )
maxreciprocate
committedOct 26, 2023
support customized run_name in tracker (CarperAI#573 )

Jingru
and
聂靖入
committedOct 26, 2023

Commits on Oct 23, 2023

support extra model and tokenizer configs during loading by from_pretrained in accelerate trainer (CarperAI#568 )

committedOct 23, 2023
docs: update documentation (CarperAI#557 )
maxreciprocate
committedOct 23, 2023

Commits on Oct 20, 2023

fix(examples/t5_summarize_cnn): move labels into reward_fn kwargs (CarperAI#570 )
maxreciprocate
committedOct 20, 2023

Commits on Oct 17, 2023

fix(modeling_base): re-order model.forward_kwargs initialization (CarperAI#566 )
maxreciprocate
committedOct 17, 2023

Commits on Oct 11, 2023

feat: add rejection finetuning trainer (CarperAI#554 )
maxreciprocate
committedOct 11, 2023

Commits on Sep 25, 2023

add default llama2 config converter (CarperAI#563 )

PhungVanDuy
and
Duy Phung
committedSep 25, 2023

Commits on Sep 22, 2023

Llama NeMo support (CarperAI#542 )

committedSep 22, 2023

Commits on Sep 1, 2023

Update README.md (CarperAI#550 )
Dahoas
committedSep 1, 2023
fix(examples/hh): old gpt-j checkpoint loading (CarperAI#552 )
maxreciprocate
committedSep 1, 2023

Commits on Aug 31, 2023

chore(requirements.txt): update everything to the latest (CarperAI#547 )
maxreciprocate
committedAug 31, 2023

Commits on Aug 29, 2023

Add DS-Chat comparison (CarperAI#538 )

cat-state
and
cat-state
committedAug 29, 2023

Commits on Aug 28, 2023

revert(ppo_trainer): keep save_pretrained only over the base model (CarperAI#551 )
maxreciprocate
committedAug 28, 2023

Commits on Aug 21, 2023

mpt support (CarperAI#546 )

50m-regent
and
50mRegent
committedAug 21, 2023

Commits on Aug 8, 2023

fix(modeling): deepspeed checkpoint loading (CarperAI#482 )

maxreciprocate
and
jon-tow
committedAug 8, 2023

Commits on Aug 2, 2023

Value branch (CarperAI#530 )
Dahoas
committedAug 2, 2023
Freeze "output" embedding when using tied embeddings. (CarperAI#543 )

cat-state
and
cat-state
committedAug 2, 2023

Commits on Jul 31, 2023

fix reward model state dict loading (CarperAI#541 )
maxjeblick
committedJul 31, 2023
Fix LLaMA example (LLaMA 2) (CarperAI#539 )

committedJul 31, 2023
Fix ordering of ppo epoch iteration (CarperAI#522 )

RobertKirk
and
maxreciprocate
committedJul 31, 2023

Commits on Jul 24, 2023

fix(modeling_ppo): load reference head under zero3 (CarperAI#489 )
maxreciprocate
committedJul 24, 2023
Fix logging (CarperAI#526 )

Dahoas
and
maxreciprocate
committedJul 24, 2023

Commits on Jul 21, 2023

Update README.md (CarperAI#537 )
Dahoas
committedJul 21, 2023

Commits on Jul 20, 2023

Fix: rename model_tok to tokenizer is reward_fn arg (CarperAI#534 )
Dahoas
committedJul 20, 2023

Commits on Jul 19, 2023

Dense reward carper (Fine grained feedback) (CarperAI#514 )

committedJul 19, 2023

Commits on Jul 17, 2023

fix(ci): don't rely on cache in actions/setup-python (CarperAI#524 )
maxreciprocate
committedJul 17, 2023

Commits on Jul 13, 2023

feat(readme): add instructions to avoid OOMs with hyperparameters (CarperAI#470 )
maxreciprocate
committedJul 13, 2023

Commits on Jul 10, 2023

Update README.md
LouisCastricato
committedJul 10, 2023

Commits on Jul 5, 2023

fix peft config cause tensorboard type error (CarperAI#515 )
congchan
committedJul 5, 2023
fix(modeling_base): handle parent directory for peft head save (CarperAI#516 )
congchan
committedJul 5, 2023

Commits on Jun 26, 2023

feat(modeling_ppo): add GPTBigCodeModelBranch (CarperAI#507 )

congchan
and
maxreciprocate
committedJun 26, 2023

Commits on Jun 23, 2023

Release: v0.7.0 (CarperAI#511 )
jon-tow
committedJun 23, 2023