flash-attention/flash_attn/models
Xuechen Li bb4cded17b
support when num_heads is not divisible by world_size; resolves #459 (#461)
* uneql rank.

* trim.

* enable passing in number of heads for each rank.

* simplify.

* simplify.

* cleanup.

* fix col parallel.

* fix bug with row parallel.

* fit out proj.

* refac.

* fix sharding logic.

* refac sharding.

* refac.

* support multiple of.

* make fn reuseable.

* fix bug in dimensions.

* scaffold.

* test uneven heads.

* fix test by adding barrier.

* refac.

* reuse code.

* clean up.
2023-08-18 14:10:35 -07:00
..
__init__.py Add __init__.py files to subdirectories for installation 2022-11-17 16:55:44 -08:00
bert.py Allow rotary embeddings for Bert (#363) 2023-07-23 00:21:45 -07:00
falcon.py [GPT] Implement Falcon 2023-07-23 10:32:29 -07:00
gpt_neox.py Implement GPT-NeoX 2023-03-29 01:21:25 -07:00
gpt.py support when num_heads is not divisible by world_size; resolves #459 (#461) 2023-08-18 14:10:35 -07:00
gptj.py Implement LLaMa 2023-04-18 21:51:35 -07:00
llama.py enable loading hf llama checkpoints for training (#446) 2023-08-15 08:33:15 -07:00
opt.py Implement LLaMa 2023-04-18 21:51:35 -07:00
vit.py [ViT] Run black on vit.py 2023-08-17 17:45:09 -07:00