You can not select more than 25 topics Topics must start with a chinese character,a letter or number, can include dashes ('-') and can be up to 35 characters long.
 
 
 
 
shibing624 ab6bc7ecee update dataset. 3 days ago
.github Update bug-report.md 9 months ago
data update reward data. 1 week ago
docs Fix a typo of s^2 attn. 1 week ago
.gitignore update trl. 2 months ago
CITATION.cff init project. 11 months ago
CONTRIBUTING.md init project. 11 months ago
DISCLAIMER add scripts of 1.pt; 11 months ago
LICENSE Initial commit 11 months ago
README.md update dataset. 3 days ago
README_EN.md update llama3 template. 1 week ago
_config.yml init project. 11 months ago
build_domain_tokenizer.py add medical data. 9 months ago
chatpdf.py update template for dpo. 6 days ago
convert_dataset.py update llama3 template. 5 days ago
deepspeed_zero_stage2_config.json update merge peft adapter. 5 months ago
deepspeed_zero_stage3_config.json update merge peft adapter. 5 months ago
dpo_training.py update bos token. 6 days ago
fastapi_server_demo.py update template for dpo. 6 days ago
gradio_demo.py update template for dpo. 6 days ago
inference.py update template for dpo. 6 days ago
inference_multigpu_demo.py update template for dpo. 6 days ago
merge_peft_adapter.py update export model to hf. 5 days ago
merge_tokenizers.py update merge. 9 months ago
openai_api.py update system tokens. 3 days ago
orpo_training.py update bos token. 6 days ago
ppo_training.py update bos token. 6 days ago
pretraining.py update quantization train. 3 months ago
requirements.txt update template for dpo. 6 days ago
reward_modeling.py update bos token. 6 days ago
run_dpo.sh update bos token. 6 days ago
run_orpo.sh update bos token. 6 days ago
run_ppo.sh update bos token. 6 days ago
run_pt.sh support qwen template. 2 weeks ago
run_rm.sh update bos token. 6 days ago
run_sft.sh update bos token. 6 days ago
run_training_dpo_pipeline.ipynb 1. Add scikit-learn to requirements; 2. Update deprecated API of peft; 3. set CUDA_VISIBLE_DEVICES=0 in ppo part of demo ipynb such that users with multi CUDA devices can run it smoothly; 4; Modify the test step in demo ipynb to non-interactive; 5. Copy INSTALL step to ENG doc. 2 weeks ago
run_training_ppo_pipeline.ipynb 1. Add scikit-learn to requirements; 2. Update deprecated API of peft; 3. set CUDA_VISIBLE_DEVICES=0 in ppo part of demo ipynb such that users with multi CUDA devices can run it smoothly; 4; Modify the test step in demo ipynb to non-interactive; 5. Copy INSTALL step to ENG doc. 2 weeks ago
supervised_finetuning.py update bos token. 6 days ago
template.py update llama3 system template. 5 days ago