Wing Lian 76aeb16156 tiled_mlp supports single gpu (#2891) ...

* tiled_mlp supports single gpu

* use checkpoint offloading for arctic training

* patch torch checkpoint too

* support for single gpu zero3

* add linkback to where it was copied from

2025-07-09 12:48:22 -04:00

..

cli

Fix: do not call preprocess in multimodal or pretraining case (#2861)

2025-07-06 21:55:33 -04:00

common

don't fail during preprocess for sampling from iterable dataset (#2825) [skip ci]

2025-06-27 10:37:53 -04:00

core

update transformers to 4.53.1 (#2844) [skip ci]

2025-07-07 09:35:22 -04:00

integrations

release v0.11.0 (#2875)

2025-07-09 09:22:35 -04:00

kernels

chore: update pre-commit hooks (#2745)

2025-06-02 15:54:29 -07:00

loaders

tiled_mlp supports single gpu (#2891)

2025-07-09 12:48:22 -04:00

models

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

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

monkeypatch

tiled_mlp supports single gpu (#2891)

2025-07-09 12:48:22 -04:00

prompt_strategies

Feat: add devstral model support (#2880) [skip ci]

2025-07-08 11:01:19 -04:00

utils

tiled_mlp supports single gpu (#2891)

2025-07-09 12:48:22 -04:00

__init__.py

bump dev version (#2889) [skip ci]

2025-07-09 09:43:42 -04:00

convert.py

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

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

datasets.py

Feat: add devstral model support (#2880) [skip ci]

2025-07-08 11:01:19 -04:00

evaluate.py

Fix logging import in evaluate.py (#2782) (#2783)

2025-06-12 13:23:31 -04:00

logging_config.py

Fix: logging on py310 (#2802)

2025-06-18 15:46:27 -04:00

processing_strategies.py

feat: update CCE to use axolotl's fork (#2813) [skip ci]

2025-06-25 09:49:22 -04:00

prompt_tokenizers.py

Feat: add devstral model support (#2880) [skip ci]

2025-07-08 11:01:19 -04:00

prompters.py

Rank 0-only logging (#2608)

2025-05-28 14:57:30 +01:00

train.py

Accelerate 1.8.1 and BNB 0.46.0 update (#2815)

2025-06-28 15:29:19 -04:00