Commit Graph

3764 Commits (cf519dac6a5799b8f314aac6f510e2a98d3af9c6)
 

Author SHA1 Message Date
Tong Li d08c99be0d
Merge branch 'main' into kto
4 months ago
Tong Li f585d4e38e
[ColossalChat] Hotfix for ColossalChat (#5910)
4 months ago
Edenzzzz 8cc8f645cd
[Examples] Add lazy init to OPT and GPT examples (#5924)
4 months ago
YeAnbang 544b7a38a1 fix style, add kto data sample
4 months ago
Guangyao Zhang 62661cde22
Merge pull request #5921 from BurkeHulk/fp8_fix
4 months ago
YeAnbang 845ea7214e Merge branch 'main' of https://github.com/hpcaitech/ColossalAI into kto
4 months ago
YeAnbang 09d5ffca1a add kto
4 months ago
Hongxin Liu e86127925a
[plugin] support all-gather overlap for hybrid parallel (#5919)
4 months ago
GuangyaoZhang 5b969fd831 fix shardformer fp8 communication training degradation
4 months ago
Guangyao Zhang d0bdb51f48
Merge pull request #5899 from BurkeHulk/SP_fp8
4 months ago
Hongxin Liu 73494de577
[release] update version (#5912)
4 months ago
GuangyaoZhang 6a20f07b80 remove all to all
4 months ago
GuangyaoZhang 5a310b9ee1 fix rebase
4 months ago
GuangyaoZhang 457a0de79f shardformer fp8
4 months ago
Hongxin Liu 27a72f0de1 [misc] support torch2.3 (#5893)
4 months ago
アマデウス 530283dba0 fix object_to_tensor usage when torch>=2.3.0 (#5820)
4 months ago
Guangyao Zhang 2e28c793ce [compatibility] support torch 2.2 (#5875)
4 months ago
Hanks 9470701110
Merge pull request #5885 from BurkeHulk/feature/fp8_comm
4 months ago
YeAnbang d8bf7e09a2
Merge pull request #5901 from hpcaitech/colossalchat
4 months ago
Guangyao Zhang 1c961b20f3
[ShardFormer] fix qwen2 sp (#5903)
4 months ago
Stephan Kö 45c49dde96
[Auto Parallel]: Speed up intra-op plan generation by 44% (#5446)
4 months ago
YeAnbang b3594d4d68 fix orpo cross entropy loss
4 months ago
pre-commit-ci[bot] 51f916b11d [pre-commit.ci] auto fixes from pre-commit.com hooks
5 months ago
BurkeHulk 1f1b856354 Merge remote-tracking branch 'origin/feature/fp8_comm' into feature/fp8_comm
5 months ago
BurkeHulk 66018749f3 add fp8_communication flag in the script
5 months ago
BurkeHulk e88190184a support fp8 communication in pipeline parallelism
5 months ago
BurkeHulk 1e1959467e fix scaling algorithm in FP8 casting
5 months ago
Hongxin Liu c068ef0fa0
[zero] support all-gather overlap (#5898)
5 months ago
YeAnbang 115c4cc5a4 hotfix citation
5 months ago
YeAnbang e7a8634636 fix eval
5 months ago
YeAnbang dd9e1cdafe
Merge pull request #5850 from hpcaitech/rlhf_SimPO
5 months ago
pre-commit-ci[bot] 8a9721bafe [pre-commit.ci] auto fixes from pre-commit.com hooks
5 months ago
YeAnbang 33f15203d3 Merge branch 'main' of https://github.com/hpcaitech/ColossalAI into rlhf_SimPO
5 months ago
YeAnbang f6ef5c3609 fix style
5 months ago
YeAnbang d888c3787c add benchmark for sft, dpo, simpo, orpo. Add benchmarking result. Support lora with gradient checkpoint
5 months ago
GuangyaoZhang dbfa7d39fc fix typo
5 months ago
Guangyao Zhang 669849d74b
[ShardFormer] Add Ulysses Sequence Parallelism support for Command-R, Qwen2 and ChatGLM (#5897)
5 months ago
YeAnbang 16f3451fe2 Merge branch 'main' of https://github.com/hpcaitech/ColossalAI into rlhf_SimPO
5 months ago
Edenzzzz fbf33ecd01
[Feature] Enable PP + SP for llama (#5868)
5 months ago
Runyu Lu 66abf1c6e8
[HotFix] CI,import,requirements-test for #5838 (#5892)
5 months ago
Runyu Lu cba20525a8
[Feat] Diffusion Model(PixArtAlpha/StableDiffusion3) Support (#5838)
5 months ago
Edenzzzz 8ec24b6a4d
[Hoxfix] Fix CUDA_DEVICE_MAX_CONNECTIONS for comm overlap
5 months ago
Haze188 3420921101
[shardformer] DeepseekMoE support (#5871)
5 months ago
pre-commit-ci[bot] e17f835df7 [pre-commit.ci] auto fixes from pre-commit.com hooks
5 months ago
Hanks 6991819a97
Merge branch 'hpcaitech:main' into feature/fp8_comm
5 months ago
pre-commit-ci[bot] 7997683aac
[pre-commit.ci] pre-commit autoupdate (#5878)
5 months ago
Hongxin Liu 7afbc81d62
[quant] fix bitsandbytes version check (#5882)
5 months ago
Wang Binluo 6cd4c32be4
[shardformer] fix the moe (#5883)
5 months ago
Edenzzzz eb24fcd914
[Hotfix] Fix OPT gradient checkpointing forward
5 months ago
Haze188 ea94c07b95
[hotfix] fix the bug that large tensor exceed the maximum capacity of TensorBucket (#5879)
5 months ago