Costa HuangandGitHub 6c6193b7cd Add optional r1-style thinking reward (#551) ...

* Add optional r1-style thinking reward

* quick change

* test

* quick change

* push latest change

* fix

* tested it to work

* Push changes

* push

2025-02-06 12:21:39 -05:00

..

olmo_adapter

Add dataset cache / mixing support (#522)

2025-01-28 09:02:38 -05:00

rejection_sampling

Synthetic preference dataset (#332)

2024-10-02 14:39:21 -04:00

__init__.py

add reward training script (#217)

2024-08-09 09:19:49 -07:00

dataset_processor.py

Clean up rlvr a lil, add base support (#538)

2025-02-01 10:37:25 -08:00

dataset_transformation.py

Add optional r1-style thinking reward (#551)

2025-02-06 12:21:39 -05:00

dpo_tune_cache.py

Remove unused dependencies (#529)

2025-01-28 16:27:41 -05:00

dpo_utils.py

Separate forward head (#401)

2024-10-29 17:56:57 -04:00

finetune.py

Remove unused dependencies (#529)

2025-01-28 16:27:41 -05:00

ground_truth_utils.py

Add optional r1-style thinking reward (#551)

2025-02-06 12:21:39 -05:00

grpo_vllm_thread_ray_gtrl.py

Add optional r1-style thinking reward (#551)

2025-02-06 12:21:39 -05:00

if_functions.py

Ground-Truth RL (#377)

2024-11-20 12:30:24 -05:00

math_utils.py

Ground-Truth RL (#377)

2024-11-20 12:30:24 -05:00

merge_lora.py

Add corresponding lora finetuning with config scripts (#343)

2024-10-23 14:16:45 -07:00

mix_data_preferences.py

store configs, minor improvements to pref data mixer (#457)

2024-12-02 16:20:41 -08:00

mix_data.py

update data dist plots (#410)

2024-11-10 20:42:20 -08:00

model_utils.py

Clean up rlvr a lil, add base support (#538)

2025-02-01 10:37:25 -08:00

online_dpo_vllm_thread.py

Clean up rlvr a lil, add base support (#538)

2025-02-01 10:37:25 -08:00

ppo_vllm_thread_ray_gtrl.py

Properly set max length for eval (#548)

2025-02-04 11:46:49 -08:00

ppo_vllm_thread_ray.py

Make the script more friendly to outside users and add docs. (#528)

2025-01-28 10:11:42 -05:00

ppo_vllm_thread.py

Clean up rlvr a lil, add base support (#538)

2025-02-01 10:37:25 -08:00

reward_modeling_eval.py

Support mixing dataset for reward modeling (#326)

2024-09-05 14:31:36 -04:00

reward_modeling.py

Olmo2 and olmoe support (#525)

2025-01-23 18:35:29 -05:00

test_utils.py

Fix auto eval parsing (#358)

2024-09-18 14:18:29 -07:00

utils.py

fix checkpointing (#541)

2025-02-01 10:40:46 -08:00

vllm_utils2.py

Fix vLLM worker for new version (#521)

2025-01-15 19:03:51 -08:00

vllm_utils.py

Add online trainers (#204)

2024-09-13 11:43:06 -04:00