Skip to content

armgit / Soup

Public Imported from GitHub
0

Fine-tune LLMs from one YAML. Layer streaming trains an 8B model on a 4 GB laptop GPU.

docs(contributors): #2116 [skip ci]

Alpamys2cbda053b238
diff --git a/CONTRIBUTORS.md b/CONTRIBUTORS.md
index 83e2536d..3df9d63c 100644
--- a/CONTRIBUTORS.md
+++ b/CONTRIBUTORS.md
@@ -1027,6 +1027,7 @@ Listed by first contribution. PR numbers link the work.
- Added native ORPO on `backend: mlx` (TRL 0.29.1's odds-ratio objective with the chosen-sequence NLL, reference-free, TRL-identical tokens on rows that fit), sharing MLX DPO's training loop through objective hooks ([#2004](https://github.com/MakazhanAlpamys/Soup/issues/2004), [#2073](https://github.com/MakazhanAlpamys/Soup/pull/2073))
- Made `backend: mlx` honour `training.eval_steps` for SFT, counted in completed optimizer updates (Soup runs the evaluation itself, over every validation row, once more after the last update) ([#1990](https://github.com/MakazhanAlpamys/Soup/issues/1990), [#2074](https://github.com/MakazhanAlpamys/Soup/pull/2074))
- Added native SimPO on `backend: mlx` (TRL 0.29.1's CPO `simpo` objective with the chosen-completion NLL, reference-free, beta fixed at TRL's default 0.1, TRL-identical tokens on rows that fit), sharing the MLX preference loop and ORPO's resume guard ([#1999](https://github.com/MakazhanAlpamys/Soup/issues/1999), [#2097](https://github.com/MakazhanAlpamys/Soup/pull/2097))
+ - Capped the prompt on the transformers SimPO path at `data.max_length // 2` (keeping its end) after TRL tokenizes, as DPO and ORPO do, and corrected the issue's own claim about the #1208 refusal against the TRL source, with tests on the real setup compared against TRL's tokenizer ([#2112](https://github.com/MakazhanAlpamys/Soup/issues/2112), [#2116](https://github.com/MakazhanAlpamys/Soup/pull/2116))
- **Tanmay Garg** ([@Tanmay9223](https://github.com/Tanmay9223))
- First contribution. Moved the PRM, MoLE routing and PPO wrappers onto the shared warmup resolution, and made PPO honour `--deepspeed`, take `response_length` from `data.max_length` and keep 3 saves ([#805](https://github.com/MakazhanAlpamys/Soup/issues/805), [#1775](https://github.com/MakazhanAlpamys/Soup/pull/1775))