Making large AI models cheaper, faster and more accessible

aibig-modeldata-parallelismdeep-learningdistributed-computingfoundation-modelsheterogeneous-traininghpcinferencelarge-scalemodel-parallelismpipeline-parallelism

You can not select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.

Tree: f5c84af0b0

ColossalChat

ckpt

cloud/coati

colossalchat

colossalchat_upgrade

dev/gpt2_metainfo_patch

dev/zero-offload

develop

feat/moe

feat/online-serving

feat/speculative-decoding

feature/2-stage

feature/async-io

feature/colossal-infer

feature/elixir

feature/inference-refactor

feature/lora

feature/stable-diffusion

feature/zerobubble

fix-setup

fix/format

flybird11111-patch-1

hotfix/example_test

hotfix/kernel_build_before_load

llama3

main

moe_sp

pre-commit-ci-update-config

refactor/inference

supercooledith-patch-1

support-npu

v0.0.1-beta

v0.0.2

v0.1.0

v0.1.1

v0.1.10

v0.1.11rc1

v0.1.11rc2

v0.1.11rc3

v0.1.11rc4

v0.1.11rc5

v0.1.12

v0.1.13

v0.1.2

v0.1.3

v0.1.4

v0.1.5

v0.1.6

v0.1.7

v0.1.8

v0.1.9

v0.2.0

v0.2.1

v0.2.2

v0.2.3

v0.2.4

v0.2.5

v0.2.6

v0.2.7

v0.2.8

v0.3.0

v0.3.1

v0.3.2

v0.3.3

v0.3.4

v0.3.5

v0.3.6

v0.3.7

v0.3.8

v0.3.9

v0.4.0

v0.4.1

v0.4.2

v0.4.3

v0.4.4

v0.4.5

v0.4.6

Branches Tags

${ item.name }

Create tag ${ searchTerm }

Create branch ${ searchTerm }

from 'f5c84af0b0'

${ noResults }

ColossalAI/colossalai/shardformer/policies

History

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>		3 months ago
..
__init__.py	[shardformer] init shardformer code structure (#3731)	1 year ago
auto_policy.py	[lora] lora support hybrid parallel plugin (#5956)	4 months ago
base_policy.py	[Feature] Zigzag Ring attention (#5905)	3 months ago
bert.py	[shardformer]delete xformers (#5859)	5 months ago
blip2.py	[Shardformer] add assert for num of attention heads divisible by tp_size (#5670)	7 months ago
bloom.py	[shardformer]delete xformers (#5859)	5 months ago
chatglm2.py	[ShardFormer] Add Ulysses Sequence Parallelism support for Command-R, Qwen2 and ChatGLM (#5897)	5 months ago
command.py	[Feature] Zigzag Ring attention (#5905)	3 months ago
deepseek.py	[Feature] Zigzag Ring attention (#5905)	3 months ago
falcon.py	[Shardformer] Add parallel output for shardformer models(bloom, falcon) (#5702)	6 months ago
gpt2.py	change 'xxx if xxx else None' to 'xxx or None'	5 months ago
gptj.py	[shardformer] upgrade transformers to 4.39.3 (#5815)	5 months ago
llama.py	[Feature] Zigzag Ring attention (#5905)	3 months ago
mistral.py	[Feature] Zigzag Ring attention (#5905)	3 months ago
mixtral.py	[Feature] Zigzag Ring attention (#5905)	3 months ago
opt.py	[pre-commit.ci] auto fixes from pre-commit.com hooks	7 months ago
qwen2.py	[Feature] Zigzag Ring attention (#5905)	3 months ago
sam.py	[shardformer]delete xformers (#5859)	5 months ago
t5.py	[shardformer] Support the T5ForTokenClassification model (#5816)	5 months ago
vit.py	[Shardformer] add assert for num of attention heads divisible by tp_size (#5670)	7 months ago
whisper.py	[Shardformer] add assert for num of attention heads divisible by tp_size (#5670)	7 months ago