[fp8] support fp8 communication and fp8 training for Colossalai #6012

wangbluo · 2024-08-16T06:34:46Z

🚨 Issue number

#5873 #5916 #5926 #5964 #5886 #5960 #5943
#5927 #5999 #5972 #5974 #5915 #5996 #6004
#6005 #5955 #5971

📝 What does this PR do?

Supported FP8 communication and training for ColossalAI, including MoeHybridPlugin, HybridPlugin, Zero, and Gemini plugins. The model's loss convergence and accuracy have already tested.

@BurkeHulk @GuangyaoZhang @botbw @ver217 @flybird11111 Please have a look, thank you! :）

cast_to_fp8, cast_from_fp8, all_reduce_fp8

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

…p8_comm # Conflicts: # colossalai/quantization/fp8.py

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

Feature/fp8 comm

[Feature] FP8 communication in ShardFormer

[Shardformer] Fix Shardformer FP8 communication training accuracy degradation

[fp8] add fp8 comm for low level zero

* add llama shardformer fp8 * Llama Shardformer Parity * fix typo * fix all reduce * fix pytest failure * fix reduce op and move function to fp8.py * fix typo

* add SimPO * fix dataloader * remove debug code * add orpo * fix style * fix colossalai, transformers version * fix colossalai, transformers version * fix colossalai, transformers version * fix torch colossalai version * update transformers version * [shardformer] DeepseekMoE support (#5871) * [Feature] deepseek moe expert parallel implement * [misc] fix typo, remove redundant file (#5867) * [misc] fix typo * [pre-commit.ci] auto fixes from pre-commit.com hooks for more information, see https://pre-commit.ci --------- Co-authored-by: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com> * [Feature] deepseek support & unit test * [misc] remove debug code & useless print * [misc] fix typos (#5872) * [Feature] remove modeling file, use auto config. (#5884) * [misc] fix typos * [Feature] deepseek support via auto model, remove modeling file * [misc] delete useless file * [misc] fix typos * [Deepseek] remove redundant code (#5888) * [misc] fix typos * [Feature] deepseek support via auto model, remove modeling file * [misc] delete useless file * [misc] fix typos * [misc] remove redundant code * [Feature/deepseek] resolve comment. (#5889) * [misc] fix typos * [Feature] deepseek support via auto model, remove modeling file * [misc] delete useless file * [misc] fix typos * [misc] remove redundant code * [misc] mv module replacement into if branch * [misc] add some warning message and modify some code in unit test * [misc] fix typos --------- Co-authored-by: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com> * [Hoxfix] Fix CUDA_DEVICE_MAX_CONNECTIONS for comm overlap Co-authored-by: Edenzzzz <[email protected]> * [Feat] Diffusion Model(PixArtAlpha/StableDiffusion3) Support (#5838) * Diffusion Model Inference support * Stable Diffusion 3 Support * pixartalpha support * [HotFix] CI,import,requirements-test for #5838 (#5892) * [Hot Fix] CI,import,requirements-test --------- Co-authored-by: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com> * [Feature] Enable PP + SP for llama (#5868) * 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 * use a one cross entropy func for all shardformer models --------- Co-authored-by: Edenzzzz <[email protected]> Co-authored-by: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com> * [ShardFormer] Add Ulysses Sequence Parallelism support for Command-R, Qwen2 and ChatGLM (#5897) * add benchmark for sft, dpo, simpo, orpo. Add benchmarking result. Support lora with gradient checkpoint * fix style * [pre-commit.ci] auto fixes from pre-commit.com hooks for more information, see https://pre-commit.ci * fix eval * hotfix citation * [zero] support all-gather overlap (#5898) * [zero] support all-gather overlap * [zero] add overlap all-gather flag * [misc] fix typo * [zero] update api * fix orpo cross entropy loss * [Auto Parallel]: Speed up intra-op plan generation by 44% (#5446) * Remove unnecessary calls to deepcopy * Build DimSpec's difference dict only once This change considerably speeds up construction speed of DimSpec objects. The difference_dict is the same for each DimSpec object, so a single copy of it is enough. * Fix documentation of DimSpec's difference method * [ShardFormer] fix qwen2 sp (#5903) * [compatibility] support torch 2.2 (#5875) * Support Pytorch 2.2.2 * keep build_on_pr file and update .compatibility * fix object_to_tensor usage when torch>=2.3.0 (#5820) * [misc] support torch2.3 (#5893) * [misc] support torch2.3 * [devops] update compatibility ci * [devops] update compatibility ci * [devops] add debug * [devops] add debug * [devops] add debug * [devops] add debug * [devops] remove debug * [devops] remove debug * [release] update version (#5912) * [plugin] support all-gather overlap for hybrid parallel (#5919) * [plugin] fixed all-gather overlap support for hybrid parallel * add kto * fix style, add kto data sample * [Examples] Add lazy init to OPT and GPT examples (#5924) Co-authored-by: Edenzzzz <[email protected]> * [ColossalChat] Hotfix for ColossalChat (#5910) * add ignore and tiny llama * fix path issue * run style * fix issue * update bash * add ignore and tiny llama * fix path issue * run style * fix issue * update bash * fix ddp issue * add Qwen 1.5 32B * refactor tokenization * [FIX BUG] UnboundLocalError: cannot access local variable 'default_conversation' where it is not associated with a value (#5931) * cannot access local variable 'default_conversation' where it is not associated with a value set default value for 'default_conversation' * [pre-commit.ci] auto fixes from pre-commit.com hooks for more information, see https://pre-commit.ci --------- Co-authored-by: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com> * fix test data * refactor evaluation * remove real data path * remove real data path * Add n_fused as an input from native_module (#5894) * [FIX BUG] convert env param to int in (#5934) * [Hotfix] Fix ZeRO typo #5936 Co-authored-by: Edenzzzz <[email protected]> * [Feature] Add a switch to control whether the model checkpoint needs to be saved after each epoch ends (#5941) * Add a switch to control whether the model checkpoint needs to be saved after each epoch ends * [pre-commit.ci] auto fixes from pre-commit.com hooks for more information, see https://pre-commit.ci --------- Co-authored-by: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com> * fix style * fix style * fix style * [shardformer] hotfix attn mask (#5945) * [shardformer] hotfix attn mask (#5947) * [Feat] Distrifusion Acceleration Support for Diffusion Inference (#5895) * Distrifusion Support source * comp comm overlap optimization * sd3 benchmark * pixart distrifusion bug fix * sd3 bug fix and benchmark * generation bug fix * naming fix * add docstring, fix counter and shape error * add reference * readme and requirement * [zero] hotfix update master params (#5951) * [release] update version (#5952) * [Chat] Fix lora (#5946) * fix merging * remove filepath * fix style * Update README.md (#5958) * [hotfix] Remove unused plan section (#5957) * remove readme * fix readme * update * [test] add mixtral for sequence classification * [test] add mixtral transformer test * [moe] fix plugin * [test] mixtra pp shard test * [chore] handle non member group * [zero] solve hang * [test] pass mixtral shardformer test * [moe] implement transit between non moe tp and ep * [zero] solve hang * [misc] solve booster hang by rename the variable * solve hang when parallel mode = pp + dp * [moe] implement submesh initialization * [moe] add mixtral dp grad scaling when not all experts are activated * [chore] manually revert unintended commit * [chore] trivial fix * [chore] arg pass & remove drop token * [test] add mixtral modelling test * [moe] implement tp * [moe] test deepseek * [moe] clean legacy code * [Feature] MoE Ulysses Support (#5918) * moe sp support * moe sp bug solve * [pre-commit.ci] auto fixes from pre-commit.com hooks for more information, see https://pre-commit.ci --------- Co-authored-by: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com> * [chore] minor fix * [moe] init moe plugin comm setting with sp * moe sp + ep bug fix * [moe] finalize test (no pp) * [moe] full test for deepseek and mixtral (pp + sp to fix) * [chore] minor fix after rebase * [pre-commit.ci] auto fixes from pre-commit.com hooks for more information, see https://pre-commit.ci * [chore] solve moe ckpt test failure and some other arg pass failure * [moe] remove ops * [test] fix test: test_zero1_2 * [bug] fix: somehow logger hangs the program * [moe] deepseek moe sp support * [test] add check * [deepseek] replace attn (a workaround for bug in transformers) * [misc] skip redunant test * [misc] remove debug/print code * [moe] refactor mesh assignment * Revert "[moe] implement submesh initialization" This reverts commit 2f9bce6. * [chore] change moe_pg_mesh to private * [misc] remove incompatible test config * [misc] fix ci failure: change default value to false in moe plugin * [misc] remove useless condition * [chore] docstring * [moe] remove force_overlap_comm flag and add warning instead * [doc] add MoeHybridParallelPlugin docstring * [moe] solve dp axis issue * [chore] remove redundant test case, print string & reduce test tokens * [feat] Dist Loader for Eval (#5950) * support auto distributed data loader * [pre-commit.ci] auto fixes from pre-commit.com hooks for more information, see https://pre-commit.ci * support auto distributed data loader * [pre-commit.ci] auto fixes from pre-commit.com hooks for more information, see https://pre-commit.ci * fix tp error * remove unused parameters * remove unused * update inference * update docs * update inference --------- Co-authored-by: Michelle <[email protected]> Co-authored-by: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com> * [lora] lora support hybrid parallel plugin (#5956) * lora support hybrid plugin * fix * fix * fix * fix * fp8 operators for compressed communication cast_to_fp8, cast_from_fp8, all_reduce_fp8 * fix scaling algorithm in FP8 casting * support fp8 communication in pipeline parallelism * add fp8_communication flag in the script * [pre-commit.ci] auto fixes from pre-commit.com hooks for more information, see https://pre-commit.ci * fix typo * [pre-commit.ci] auto fixes from pre-commit.com hooks for more information, see https://pre-commit.ci * shardformer fp8 * fix rebase * remove all to all * fix shardformer fp8 communication training degradation * [fp8] support all-gather flat tensor (#5932) * [pre-commit.ci] auto fixes from pre-commit.com hooks for more information, see https://pre-commit.ci * fix * Update low_level_optim.py --------- Co-authored-by: YeAnbang <[email protected]> Co-authored-by: Haze188 <[email protected]> Co-authored-by: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com> Co-authored-by: Edenzzzz <[email protected]> Co-authored-by: Edenzzzz <[email protected]> Co-authored-by: Runyu Lu <[email protected]> Co-authored-by: Guangyao Zhang <[email protected]> Co-authored-by: YeAnbang <[email protected]> Co-authored-by: Hongxin Liu <[email protected]> Co-authored-by: Stephan Kö <[email protected]> Co-authored-by: アマデウス <[email protected]> Co-authored-by: Tong Li <[email protected]> Co-authored-by: zhurunhua <[email protected]> Co-authored-by: Insu Jang <[email protected]> Co-authored-by: Gao, Ruiyuan <[email protected]> Co-authored-by: hxwang <[email protected]> Co-authored-by: Michelle <[email protected]> Co-authored-by: Wang Binluo <[email protected]> Co-authored-by: HangXu <[email protected]>

* support all2all fp8 * fix * [pre-commit.ci] auto fixes from pre-commit.com hooks for more information, see https://pre-commit.ci * fix * fix * fix * [pre-commit.ci] auto fixes from pre-commit.com hooks for more information, see https://pre-commit.ci --------- Co-authored-by: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com>

* [fp8] add fp8 linear * [test] fix fp8 linear test condition * [test] fix fp8 linear test condition * [test] fix fp8 linear test condition

* [fp8] support fp8 amp for hybrid parallel plugin * [test] add fp8 hook test * [fp8] fix fp8 linear compatibility

* support fp8_communication in the Torch DDP grad comm, FSDP grad comm, and FSDP params comm * [pre-commit.ci] auto fixes from pre-commit.com hooks for more information, see https://pre-commit.ci * implement communication hook for FSDP params all-gather * added unit test for fp8 operators * support fp8 communication in GeminiPlugin * update training scripts to support fsdp and fp8 communication * fixed some minor bugs observed in unit test * add all_gather_into_tensor_flat_fp8 * [pre-commit.ci] auto fixes from pre-commit.com hooks for more information, see https://pre-commit.ci * [pre-commit.ci] auto fixes from pre-commit.com hooks for more information, see https://pre-commit.ci * add skip the test if torch < 2.2.0 * [pre-commit.ci] auto fixes from pre-commit.com hooks for more information, see https://pre-commit.ci * add skip the test if torch < 2.2.0 * add skip the test if torch < 2.2.0 * add fp8_comm flag * rebase latest fp8 operators * rebase latest fp8 operators * [pre-commit.ci] auto fixes from pre-commit.com hooks for more information, see https://pre-commit.ci --------- Co-authored-by: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com>

* fix * fix * fix

* [fp8] refactor hook * [fp8] support gemini plugin * [example] add fp8 option for llama benchmark

* [fp8] use torch compile (torch >= 2.4.0) * [fp8] set use_fast_accum in linear * [chore] formal version check * [chore] fix sig

* add SimPO * fix dataloader * remove debug code * add orpo * fix style * fix colossalai, transformers version * fix colossalai, transformers version * fix colossalai, transformers version * fix torch colossalai version * update transformers version * [shardformer] DeepseekMoE support (#5871) * [Feature] deepseek moe expert parallel implement * [misc] fix typo, remove redundant file (#5867) * [misc] fix typo * [pre-commit.ci] auto fixes from pre-commit.com hooks for more information, see https://pre-commit.ci --------- Co-authored-by: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com> * [Feature] deepseek support & unit test * [misc] remove debug code & useless print * [misc] fix typos (#5872) * [Feature] remove modeling file, use auto config. (#5884) * [misc] fix typos * [Feature] deepseek support via auto model, remove modeling file * [misc] delete useless file * [misc] fix typos * [Deepseek] remove redundant code (#5888) * [misc] fix typos * [Feature] deepseek support via auto model, remove modeling file * [misc] delete useless file * [misc] fix typos * [misc] remove redundant code * [Feature/deepseek] resolve comment. (#5889) * [misc] fix typos * [Feature] deepseek support via auto model, remove modeling file * [misc] delete useless file * [misc] fix typos * [misc] remove redundant code * [misc] mv module replacement into if branch * [misc] add some warning message and modify some code in unit test * [misc] fix typos --------- Co-authored-by: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com> * [Hoxfix] Fix CUDA_DEVICE_MAX_CONNECTIONS for comm overlap Co-authored-by: Edenzzzz <[email protected]> * [Feat] Diffusion Model(PixArtAlpha/StableDiffusion3) Support (#5838) * Diffusion Model Inference support * Stable Diffusion 3 Support * pixartalpha support * [HotFix] CI,import,requirements-test for #5838 (#5892) * [Hot Fix] CI,import,requirements-test --------- Co-authored-by: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com> * [Feature] Enable PP + SP for llama (#5868) * 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 * use a one cross entropy func for all shardformer models --------- Co-authored-by: Edenzzzz <[email protected]> Co-authored-by: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com> * [ShardFormer] Add Ulysses Sequence Parallelism support for Command-R, Qwen2 and ChatGLM (#5897) * add benchmark for sft, dpo, simpo, orpo. Add benchmarking result. Support lora with gradient checkpoint * fix style * [pre-commit.ci] auto fixes from pre-commit.com hooks for more information, see https://pre-commit.ci * fix eval * hotfix citation * [zero] support all-gather overlap (#5898) * [zero] support all-gather overlap * [zero] add overlap all-gather flag * [misc] fix typo * [zero] update api * fix orpo cross entropy loss * [Auto Parallel]: Speed up intra-op plan generation by 44% (#5446) * Remove unnecessary calls to deepcopy * Build DimSpec's difference dict only once This change considerably speeds up construction speed of DimSpec objects. The difference_dict is the same for each DimSpec object, so a single copy of it is enough. * Fix documentation of DimSpec's difference method * [ShardFormer] fix qwen2 sp (#5903) * [compatibility] support torch 2.2 (#5875) * Support Pytorch 2.2.2 * keep build_on_pr file and update .compatibility * fix object_to_tensor usage when torch>=2.3.0 (#5820) * [misc] support torch2.3 (#5893) * [misc] support torch2.3 * [devops] update compatibility ci * [devops] update compatibility ci * [devops] add debug * [devops] add debug * [devops] add debug * [devops] add debug * [devops] remove debug * [devops] remove debug * [release] update version (#5912) * [plugin] support all-gather overlap for hybrid parallel (#5919) * [plugin] fixed all-gather overlap support for hybrid parallel * add kto * fix style, add kto data sample * [Examples] Add lazy init to OPT and GPT examples (#5924) Co-authored-by: Edenzzzz <[email protected]> * [ColossalChat] Hotfix for ColossalChat (#5910) * add ignore and tiny llama * fix path issue * run style * fix issue * update bash * add ignore and tiny llama * fix path issue * run style * fix issue * update bash * fix ddp issue * add Qwen 1.5 32B * refactor tokenization * [FIX BUG] UnboundLocalError: cannot access local variable 'default_conversation' where it is not associated with a value (#5931) * cannot access local variable 'default_conversation' where it is not associated with a value set default value for 'default_conversation' * [pre-commit.ci] auto fixes from pre-commit.com hooks for more information, see https://pre-commit.ci --------- Co-authored-by: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com> * fix test data * refactor evaluation * remove real data path * remove real data path * Add n_fused as an input from native_module (#5894) * [FIX BUG] convert env param to int in (#5934) * [Hotfix] Fix ZeRO typo #5936 Co-authored-by: Edenzzzz <[email protected]> * [Feature] Add a switch to control whether the model checkpoint needs to be saved after each epoch ends (#5941) * Add a switch to control whether the model checkpoint needs to be saved after each epoch ends * [pre-commit.ci] auto fixes from pre-commit.com hooks for more information, see https://pre-commit.ci --------- Co-authored-by: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com> * fix style * fix style * fix style * [shardformer] hotfix attn mask (#5945) * [shardformer] hotfix attn mask (#5947) * [Feat] Distrifusion Acceleration Support for Diffusion Inference (#5895) * Distrifusion Support source * comp comm overlap optimization * sd3 benchmark * pixart distrifusion bug fix * sd3 bug fix and benchmark * generation bug fix * naming fix * add docstring, fix counter and shape error * add reference * readme and requirement * [zero] hotfix update master params (#5951) * [release] update version (#5952) * [Chat] Fix lora (#5946) * fix merging * remove filepath * fix style * Update README.md (#5958) * [hotfix] Remove unused plan section (#5957) * remove readme * fix readme * update * [test] add mixtral for sequence classification * [test] add mixtral transformer test * [moe] fix plugin * [test] mixtra pp shard test * [chore] handle non member group * [zero] solve hang * [test] pass mixtral shardformer test * [moe] implement transit between non moe tp and ep * [zero] solve hang * [misc] solve booster hang by rename the variable * solve hang when parallel mode = pp + dp * [moe] implement submesh initialization * [moe] add mixtral dp grad scaling when not all experts are activated * [chore] manually revert unintended commit * [chore] trivial fix * [chore] arg pass & remove drop token * [test] add mixtral modelling test * [moe] implement tp * [moe] test deepseek * [moe] clean legacy code * [Feature] MoE Ulysses Support (#5918) * moe sp support * moe sp bug solve * [pre-commit.ci] auto fixes from pre-commit.com hooks for more information, see https://pre-commit.ci --------- Co-authored-by: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com> * [chore] minor fix * [moe] init moe plugin comm setting with sp * moe sp + ep bug fix * [moe] finalize test (no pp) * [moe] full test for deepseek and mixtral (pp + sp to fix) * [chore] minor fix after rebase * [pre-commit.ci] auto fixes from pre-commit.com hooks for more information, see https://pre-commit.ci * [chore] solve moe ckpt test failure and some other arg pass failure * [moe] remove ops * [test] fix test: test_zero1_2 * [bug] fix: somehow logger hangs the program * [moe] deepseek moe sp support * [test] add check * [deepseek] replace attn (a workaround for bug in transformers) * [misc] skip redunant test * [misc] remove debug/print code * [moe] refactor mesh assignment * Revert "[moe] implement submesh initialization" This reverts commit 2f9bce6. * [chore] change moe_pg_mesh to private * [misc] remove incompatible test config * [misc] fix ci failure: change default value to false in moe plugin * [misc] remove useless condition * [chore] docstring * [moe] remove force_overlap_comm flag and add warning instead * [doc] add MoeHybridParallelPlugin docstring * [moe] solve dp axis issue * [chore] remove redundant test case, print string & reduce test tokens * [feat] Dist Loader for Eval (#5950) * support auto distributed data loader * [pre-commit.ci] auto fixes from pre-commit.com hooks for more information, see https://pre-commit.ci * support auto distributed data loader * [pre-commit.ci] auto fixes from pre-commit.com hooks for more information, see https://pre-commit.ci * fix tp error * remove unused parameters * remove unused * update inference * update docs * update inference --------- Co-authored-by: Michelle <[email protected]> Co-authored-by: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com> * [lora] lora support hybrid parallel plugin (#5956) * lora support hybrid plugin * fix * fix * fix * fix * Support overall loss, update KTO logging * [Docs] clarify launch port Co-authored-by: Edenzzzz <[email protected]> * [Hotfix] README link (#5966) * update ignore * update readme * run style * update readme * [Hotfix] Avoid fused RMSnorm import error without apex (#5985) Co-authored-by: Edenzzzz <[email protected]> * [Chat] fix readme (#5989) * fix readme * fix readme, tokenization fully tested * fix readme, tokenization fully tested * [pre-commit.ci] auto fixes from pre-commit.com hooks for more information, see https://pre-commit.ci --------- Co-authored-by: root <root@notebook-8f919155-6035-47b4-9c6f-1be133b9e2c9-0.notebook-8f919155-6035-47b4-9c6f-1be133b9e2c9.colossal-ai.svc.cluster.local> Co-authored-by: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com> * fix sync condition (#6000) * [plugin] add cast inputs option for zero (#6003) * [pre-commit.ci] pre-commit autoupdate (#5995) updates: - [github.com/psf/black-pre-commit-mirror: 24.4.2 → 24.8.0](psf/black-pre-commit-mirror@24.4.2...24.8.0) Co-authored-by: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com> * [misc] Bypass the huggingface bug to solve the mask mismatch problem (#5991) * [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 <[email protected]> Co-authored-by: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com> * [misc] update compatibility (#6008) * [misc] update compatibility * [misc] update requirements * [devops] disable requirements cache * [test] fix torch ddp test * [test] fix rerun on address in use * [test] fix lazy init * [pre-commit.ci] auto fixes from pre-commit.com hooks for more information, see https://pre-commit.ci * fix the merge * fix the merge * overlap kv comm with output rescale (#6017) Co-authored-by: Edenzzzz <[email protected]> * fix the merge * [pre-commit.ci] auto fixes from pre-commit.com hooks for more information, see https://pre-commit.ci * fix the merge * fix * fix * fix the merge * fix * [misc] Use dist logger in plugins (#6011) * use dist logger in plugins * remove trash * print on rank 0 --------- Co-authored-by: Edenzzzz <[email protected]> * fix * fix * fix * fix * fix the merge * fix * fix * fix * fix --------- Co-authored-by: YeAnbang <[email protected]> Co-authored-by: Haze188 <[email protected]> Co-authored-by: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com> Co-authored-by: Edenzzzz <[email protected]> Co-authored-by: Edenzzzz <[email protected]> Co-authored-by: Runyu Lu <[email protected]> Co-authored-by: Guangyao Zhang <[email protected]> Co-authored-by: YeAnbang <[email protected]> Co-authored-by: Hongxin Liu <[email protected]> Co-authored-by: Stephan Kö <[email protected]> Co-authored-by: アマデウス <[email protected]> Co-authored-by: Tong Li <[email protected]> Co-authored-by: zhurunhua <[email protected]> Co-authored-by: Insu Jang <[email protected]> Co-authored-by: Gao, Ruiyuan <[email protected]> Co-authored-by: hxwang <[email protected]> Co-authored-by: Michelle <[email protected]> Co-authored-by: root <root@notebook-8f919155-6035-47b4-9c6f-1be133b9e2c9-0.notebook-8f919155-6035-47b4-9c6f-1be133b9e2c9.colossal-ai.svc.cluster.local>

[fp8] merge

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

[fp8] merge

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

Update train_dpo.py

colossalai/shardformer/layer/attn.py

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

[fp8] fix the merge

BurkeHulk and others added 30 commits July 1, 2024 13:44

fp8 operators for compressed communication

f5a52e1

cast_to_fp8, cast_from_fp8, all_reduce_fp8

Merge branch 'hpcaitech:main' into feature/fp8_comm

6991819

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

e17f835

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

fix typo

dbfa7d3

fix scaling algorithm in FP8 casting

1e19594

support fp8 communication in pipeline parallelism

e881901

add fp8_communication flag in the script

6601874

Merge remote-tracking branch 'origin/feature/fp8_comm' into feature/f…

1f1b856

…p8_comm # Conflicts: # colossalai/quantization/fp8.py

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

51f916b

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

Merge pull request #5885 from BurkeHulk/feature/fp8_comm

9470701

Feature/fp8 comm

shardformer fp8

457a0de

fix rebase

5a310b9

remove all to all

6a20f07

Merge pull request #5899 from BurkeHulk/SP_fp8

d0bdb51

[Feature] FP8 communication in ShardFormer

fix shardformer fp8 communication training degradation

5b969fd

Merge pull request #5921 from BurkeHulk/fp8_fix

62661cd

[Shardformer] Fix Shardformer FP8 communication training accuracy degradation

[fp8] support all-gather flat tensor (#5932)

5fd0592

[fp8] add fp8 comm for low level zero

ae486ce

[test] add zero fp8 test case

91e596d

Merge pull request #5961 from ver217/feature/zeor-fp8

c297e21

[fp8] add fp8 comm for low level zero

[Feature] llama shardformer fp8 support (#5938)

53cb960

* add llama shardformer fp8 * Llama Shardformer Parity * fix typo * fix all reduce * fix pytest failure * fix reduce op and move function to fp8.py * fix typo

[fp8] add fp8 linear (#5967)

76ea164

* [fp8] add fp8 linear * [test] fix fp8 linear test condition * [test] fix fp8 linear test condition * [test] fix fp8 linear test condition

[fp8] support fp8 amp for hybrid parallel plugin (#5975)

ccabcf6

* [fp8] support fp8 amp for hybrid parallel plugin * [test] add fp8 hook test * [fp8] fix fp8 linear compatibility

fix (#5976)

7739629

[test ci]Feature/fp8 comm (#5981)

4b9bec8

* fix * fix * fix

[fp8] support gemini plugin (#5978)

8241c0c

* [fp8] refactor hook * [fp8] support gemini plugin * [example] add fp8 option for llama benchmark

[fp8] use torch compile (torch >= 2.3.0) (#5979)

e4aadee

* [fp8] use torch compile (torch >= 2.4.0) * [fp8] set use_fast_accum in linear * [chore] formal version check * [chore] fix sig

wangbluo and others added 23 commits August 19, 2024 09:23

fix

2eb3683

fix the merge

88b3f06

fix

1f703e0

fix

5382311

fix

f7acfa1

fix

2ee6235

fix

2e4cbe3

fix merge

2d362ac

fix the merge

eb5ba40

fix

193030f

fix

6aface9

fix

698c8b9

fix

8b8e282

Merge pull request #6023 from wangbluo/fp8_merge

d77e66a

[fp8] merge

fix

971b16a

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

a292554

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

Merge pull request #6024 from wangbluo/fix_merge

afe845f

[fp8] merge

Merge branch 'main' into feature/fp8_comm

caab4a3

Update train_dpo.py

0bc9a87

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

3b0df30

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

Update low_level_zero_plugin.py

9e76764

Merge pull request #6029 from hpcaitech/flybird11111-patch-1

0bf46c5

Update train_dpo.py

ver217 reviewed Aug 26, 2024

View reviewed changes

colossalai/shardformer/layer/attn.py Outdated Show resolved Hide resolved

wangbluo and others added 3 commits August 26, 2024 03:45

fix

dae3999

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

80d24ae

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

Merge pull request #6033 from wangbluo/fix

4a6f31e

[fp8] fix the merge

ver217 approved these changes Aug 27, 2024

View reviewed changes

ver217 merged commit 17904cb into main Aug 27, 2024
8 checks passed

ver217 deleted the feature/fp8_comm branch August 27, 2024 02:09

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

[fp8] support fp8 communication and fp8 training for Colossalai #6012

[fp8] support fp8 communication and fp8 training for Colossalai #6012

wangbluo commented Aug 16, 2024 •

edited

Loading

[fp8] support fp8 communication and fp8 training for Colossalai #6012

[fp8] support fp8 communication and fp8 training for Colossalai #6012

Conversation

wangbluo commented Aug 16, 2024 • edited Loading

🚨 Issue number

📝 What does this PR do?

wangbluo commented Aug 16, 2024 •

edited

Loading