Yuanheng Zhao 3de2e62299 [Inference] Add CacheBlock and KV-Cache Manager (#5156) ...

* [Inference] Add KVCache Manager

* function refactored

* add test for KVCache Manager

* add attr beam width

* Revise alloc func in CacheManager

* Fix docs and pytests

* add tp slicing for head number

* optimize shapes of tensors used as physical cache

* Apply using InferenceConfig on KVCacheManager

* rm duplicate config file

* Optimize cache allocation: use contiguous cache

* Fix config in pytest (and config)

2024-01-11 13:39:29 +00:00

..

_analyzer

[misc] update pre-commit and run all files (#4752)

2023-09-19 14:20:26 +08:00

_C

[setup] support pre-build and jit-build of cuda kernels (#2374)

2023-01-06 20:50:26 +08:00

amp

[npu] add npu support for gemini and zero (#5067)

2023-11-20 16:12:41 +08:00

auto_parallel

[npu] add npu support for gemini and zero (#5067)

2023-11-20 16:12:41 +08:00

autochunk

[misc] update pre-commit and run all files (#4752)

2023-09-19 14:20:26 +08:00

booster

[ci] fix shardformer tests. (#5255)

2024-01-11 19:07:45 +08:00

checkpoint_io

[pipeline,shardformer] Fix p2p efficiency in pipeline, allow skipping loading weight not in weight_map when strict=False, fix llama flash attention forward, add flop estimation by megatron in llama benchmark (#5017)

2023-11-16 20:15:59 +08:00

cli

[bug] Fix the version check bug in colossalai run when generating the cmd. (#4713)

2023-09-22 10:50:47 +08:00

cluster

fix-test (#5210)

2024-01-03 14:26:13 +08:00

context

[moe] merge moe into main (#4978)

2023-11-02 02:21:24 +00:00

device

[npu] add npu support for hybrid plugin and llama (#5090)

2023-11-22 19:23:21 +08:00

fx

[misc] update pre-commit and run all files (#4752)

2023-09-19 14:20:26 +08:00

inference

[Inference] Add CacheBlock and KV-Cache Manager (#5156)

2024-01-11 13:39:29 +00:00

interface

[lazy] support from_pretrained (#4801)

2023-09-26 11:04:11 +08:00

kernel

[Inference/NFC] Clean outdated inference tests and deprecated kernels (#5159)

2024-01-11 13:39:29 +00:00

lazy

[doc] add lazy init docs (#4808)

2023-09-27 10:24:04 +08:00

legacy

[shardformer]: support gpt-j, falcon, Mistral and add interleaved pipeline for bert (#5088)

2023-11-28 16:54:42 +08:00

logging

[misc] update pre-commit and run all files (#4752)

2023-09-19 14:20:26 +08:00

moe

[hotfix]: modify create_ep_hierarchical_group and add test (#5032)

2023-11-17 10:53:00 +08:00

nn

[npu] add npu support for gemini and zero (#5067)

2023-11-20 16:12:41 +08:00

pipeline

[pipeline] A more general _communicate in p2p (#5062)

2024-01-08 15:37:27 +08:00

shardformer

[doc] fix doc typo (#5256)

2024-01-11 21:01:11 +08:00

tensor

fix (#5158)

2023-12-05 14:28:36 +08:00

testing

[npu] add npu support for hybrid plugin and llama (#5090)

2023-11-22 19:23:21 +08:00

utils

[shardformer]: support gpt-j, falcon, Mistral and add interleaved pipeline for bert (#5088)

2023-11-28 16:54:42 +08:00

zero

[shardformer]: support gpt-j, falcon, Mistral and add interleaved pipeline for bert (#5088)

2023-11-28 16:54:42 +08:00

__init__.py

[misc] update pre-commit and run all files (#4752)

2023-09-19 14:20:26 +08:00

initialize.py

[npu] add npu support for gemini and zero (#5067)

2023-11-20 16:12:41 +08:00