ColossalAI/colossalai/shardformer/shard
Edenzzzz f5c84af0b0
[Feature] Zigzag Ring attention (#5905)
* halfway

* fix cross-PP-stage position id length diff bug

* fix typo

* fix typo

* [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci

* unified cross entropy func for all shardformer models

* remove redundant lines

* add basic ring attn; debug cross entropy

* fwd bwd logic complete

* fwd bwd logic complete; add experimental triton rescale

* precision tests passed

* precision tests passed

* fix typos and remove misc files

* [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci

* add sp_mode to benchmark; fix varlen interface

* update softmax_lse shape by new interface

* change tester name

* remove buffer clone; support packed seq layout

* add varlen tests

* fix typo

* all tests passed

* add dkv_group; fix mask

* remove debug statements

---------

Co-authored-by: Edenzzzz <wtan45@wisc.edu>
Co-authored-by: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com>
2024-08-16 13:56:38 +08:00
..
__init__.py [shardformer, pipeline] add `gradient_checkpointing_ratio` and heterogenous shard policy for llama (#5508) 2024-04-01 11:34:58 +08:00
grad_ckpt_config.py [shardformer] refactor pipeline grad ckpt config (#5646) 2024-04-25 15:19:30 +08:00
shard_config.py [Feature] Zigzag Ring attention (#5905) 2024-08-16 13:56:38 +08:00
sharder.py [nfc] fix typo colossalai/shardformer/ (#5133) 2024-01-04 16:21:55 +08:00
shardformer.py [Hoxfix] Fix CUDA_DEVICE_MAX_CONNECTIONS for comm overlap 2024-07-05 20:02:36 +08:00
utils.py [pipeline] update shardformer policy 2023-08-15 23:25:14 +08:00