Dan Saunders 80304c26a7 SP GRPO support + batch SP fixes (#2643) ...

* ctx manager for SP

* updates

* update

* further simplifying

* simplifying

* simplifying

* reorg

* batch api HF adapter for ring-flash-attn; cleanup and improvements

* update

* adding all batch ring-flash-attn methods via single adapter

* fix

* fixes for batch API funcs, simplify

* fix

* grpo sp support

* progress

* stronger subclassing of TRL GRPO trainer; custom distributed sampler

* subclassing constructor

* progress

* finalizing SP + GRPO trainer

* minimize diffs to GRPO trainer

* remove (most of) the custom GRPO trainer logic

* debug

* debug

* update

* update

* update

* progress

* cleanup

* cleanup

* minor changes

* update

* update

* update

* small changes

* updates

* cleanup; torch.compile ring_flash_attn functions to prevent numerical instability; lint

* spacing

* cleanup; log in pydantic model config only on main process

* remove comment

* fix sp sampler, update to latest upstream code, doc

* add docs

* update quartodoc autodoc contents

* fix, simplifications

* fixes + simplifications

* review comments

* lint

* removing main process only logs in favor of #2608

* fixes, additional smoke test

* updatse

* more tests

* update

* fix grad accum bug (sort of)

* lint, tests

* todo

2025-05-12 17:52:40 -04:00

..

integrations

don't sort multipack sampler (#2657)

2025-05-09 20:28:58 -04:00

kernels

adding pre-commit auto-update GH action and bumping plugin versions (#2428)

2025-03-21 11:02:43 -04:00

multigpu

SP GRPO support + batch SP fixes (#2643)

2025-05-12 17:52:40 -04:00

patched

SP GRPO support + batch SP fixes (#2643)

2025-05-12 17:52:40 -04:00

solo

make sure to download fixtures for kd test (#2541)

2025-04-21 10:31:50 -04:00

__init__.py

missing dunder-init

2023-11-06 18:33:01 -05:00

.gitignore

Support Sample packing for phi arch (#586)

2023-09-15 15:46:54 -04:00

test_deepseekv3.py

make e2e tests a bit faster by reducing test split size (#2522) [skip ci]

2025-04-12 07:24:43 -07:00

test_dpo.py

replace references to random 68m model w 135m smollm2 (#2570) [skip ci]

2025-04-28 10:08:07 -04:00

test_embeddings_lr.py

make sure to validate the config before normalizing so defaults get set (#2554)

2025-04-24 13:01:43 -04:00

test_evaluate.py

swap tinymodels that have safetensors for some ci tests (#2641)

2025-05-07 15:06:07 -04:00

test_falcon.py

swap tinymodels that have safetensors for some ci tests (#2641)

2025-05-07 15:06:07 -04:00

test_gemma2.py

make e2e tests a bit faster by reducing test split size (#2522) [skip ci]

2025-04-12 07:24:43 -07:00

test_gemma3_text.py

make e2e tests a bit faster by reducing test split size (#2522) [skip ci]

2025-04-12 07:24:43 -07:00

test_imports.py

adding pre-commit auto-update GH action and bumping plugin versions (#2428)

2025-03-21 11:02:43 -04:00

test_llama_pretrain.py

feat: add config for optional parameters in a chat message (#2260)

2025-02-18 09:59:27 +07:00

test_llama_vision.py

make sure to validate the config before normalizing so defaults get set (#2554)

2025-04-24 13:01:43 -04:00

test_llama.py

replace references to random 68m model w 135m smollm2 (#2570) [skip ci]

2025-04-28 10:08:07 -04:00

test_load_model.py

replace references to random 68m model w 135m smollm2 (#2570) [skip ci]

2025-04-28 10:08:07 -04:00

test_lora_llama.py

Add Post_model_load, post_lora_load, post_train, post_train_unload function calls (#2539)

2025-04-28 10:10:28 -04:00

test_mamba.py

feat: add config for optional parameters in a chat message (#2260)

2025-02-18 09:59:27 +07:00

test_mistral.py

swap tinymodels that have safetensors for some ci tests (#2641)

2025-05-07 15:06:07 -04:00

test_mixtral.py

make e2e tests a bit faster by reducing test split size (#2522) [skip ci]

2025-04-12 07:24:43 -07:00

test_optimizers.py

Add CAME Optimizer (#2385)

2025-05-07 10:31:46 -04:00

test_packing_loss.py

make sure to download fixtures for kd test (#2541)

2025-04-21 10:31:50 -04:00

test_phi.py

make sure to validate the config before normalizing so defaults get set (#2554)

2025-04-24 13:01:43 -04:00

test_process_reward_model_smollm2.py

make sure to validate the config before normalizing so defaults get set (#2554)

2025-04-24 13:01:43 -04:00

test_qwen.py

feat: add config for optional parameters in a chat message (#2260)

2025-02-18 09:59:27 +07:00

test_reward_model_smollm2.py

feat: add config for optional parameters in a chat message (#2260)

2025-02-18 09:59:27 +07:00

test_schedulers.py

replace references to random 68m model w 135m smollm2 (#2570) [skip ci]

2025-04-28 10:08:07 -04:00

utils.py

Add: Sparse Finetuning Integration with llmcompressor (#2479)

2025-05-01 12:25:16 -04:00