[moe] merge branch feature/pipeline #4435

oahzxl · 2023-08-14T14:32:53Z

📌 Checklist before creating the PR

I have created an issue for this PR for traceability
The title follows the standard format: [doc/gemini/tensor/...]: A concise description
I have added relevant tags if possible for us to better distinguish different PRs

🚨 Issue number

Link this PR to your issue with words like fixed to automatically close the linked issue upon merge

e.g. fixed #1234, closed #1234, resolved #1234

📝 What does this PR do?

Summarize your work here.
if you have any plots/diagrams/screenshots/tables, please attach them here.

💥 Checklist before requesting a review

I have linked my PR to an issue (instruction)
My issue clearly describes the problem/feature/proposal, with diagrams/charts/table/code if possible
I have performed a self-review of my code
I have added thorough tests.
I have added docstrings for all the functions/methods I implemented

⭐️ Do you enjoy contributing to Colossal-AI?

🌝 Yes, I do.
🌚 No, I don't.

Tell us more if you don't enjoy contributing to Colossal-AI.

* [cluster] add process group mesh * [test] add process group mesh test * force sync

* [pipeline] add stage manager * [test] add pipeline stage manager test * [pipeline] add docstring for stage manager

* [pipeline] add p2p communication * [test] add p2p communication test * [test] add rerun decorator * [test] rename to avoid conflict

* [api] update optimizer wrapper to fit pipeline * [pipeline] add base schedule * [pipeline] add 1f1b schedule * [test] add pipeline schedule utils test * [pipeline] fix import

* add pipeline policy and bert forward to be done * add bertmodel pipeline forward and make tests * add Bert_Policy and test for policy * update formatting * update formatting * update the code * fix bugs * fix name confilt

* [cluster] add process group mesh * [test] add process group mesh test * force sync

* [pipeline] add stage manager * [test] add pipeline stage manager test * [pipeline] add docstring for stage manager

* [pipeline] add p2p communication * [test] add p2p communication test * [test] add rerun decorator * [test] rename to avoid conflict

* [api] update optimizer wrapper to fit pipeline * [pipeline] add base schedule * [pipeline] add 1f1b schedule * [test] add pipeline schedule utils test * [pipeline] fix import

* add pipeline policy and bert forward to be done * add bertmodel pipeline forward and make tests * add Bert_Policy and test for policy * update formatting * update formatting * update the code * fix bugs * fix name confilt

[sync] update from main

…licy (hpcaitech#4161) * add pipeline policy and bert forward to be done * add bertmodel pipeline forward and make tests * add Bert_Policy and test for policy * update formatting * update formatting * update the code * fix bugs * fix name confilt * add bloom model and policy ,revise the base class of policy * revise * revision * add bert_for_pretraining

[pipeline] fit shardformer policy

…pcaitech#4172) * add pipeline policy and bert forward to be done * add bertmodel pipeline forward and make tests * add Bert_Policy and test for policy * update formatting * update formatting * update the code * fix bugs * fix name confilt * add bloom model and policy ,revise the base class of policy * revise * revision * add bert_for_pretraining * add bert_for_pretraining forward and policy * fix typos * cancel warning * change the imediate output to default dict * change the default output of get_shared_params

* [shardformer] added tests * [shardformer] vit test finish and support * fix attention dropout

…itech#4187) * move bert related pipeline components to shardformer * fix bugs * revision * fix bert model tests * fix bert_lm_head model tests * fix tests * fix tests * done checks * skip bloom

* [shardformer] support lazy init * [shardformer] linear support lazy init * [shardformer] embedding support lazy init * [shardformer] norm support lazy init * [shardformer] fused linear support lazy init * [test] update shardformer test layer * [test] shardformer with lazy init fit ddp * [lazy] hotfix deepcopy of param * [shardformer] fix bert policy and update test * [shardformer] fix bloom policy and update test * [shardformer] fix opt policy and update test * [shardformer] fix t5 policy and update test * [shardformer] fix gpt2 policy and update test * [shardformer] fix llama policy and update test

* add pipeline forward * complete pipeline forward check * fix bert forward without pipeline * fix comments * discard useless line * add todo * clean prints * fix distribute layers

* bloom policy * llama pipeline forward and tests * fix the output and attention_mask * fix name * bind argument to policy * Revert "bloom policy" This reverts commit 8dee68a. This policy should be revert and copied to feature/bloom * revert the bloom changes * cancel unneeded inputs * gpt

…line (hpcaitech#4208) * bloom policy * llama pipeline forward and tests * fix the output and attention_mask * fix name * bind argument to policy * Revert "bloom policy" This reverts commit 8dee68a. This policy should be revert and copied to feature/bloom * revert the bloom changes * cancel unneeded inputs * gpt * finish llama * causal lm and sequence classification * revision

* bloom policy * llama pipeline forward and tests * fix the output and attention_mask * fix name * bind argument to policy * finish bloom model * test shard gpt2 * clear cache

…4224) * * fix typehint & docstring in sharder.py * * update pipeline forward for GPT2Model * * add test for pipeline forward of GPT2Model * * add cache cleaning in gpt2 test * * change assert to raise command

* 1.support sam 2.add fused qkv for nn.Linear * update utils support set element in list * overtwrite SamVisionAttention foward to use DropoutForParallelInput * remove unused code

* [hotfix] fix gemini and zero test * [hotfix] fix lazy init test * [hotfix] fix lazy init test

* add unit test for 1f1b * polish code * polish code and update ut version * fix

hpcaitech#4324) * refactor tests * refactor bloom model * finish policy tests * refactor tests * fix test pure pipeline * remove test pipeline and cutdown launch process * refactor tests * refactor bloom model * finish policy tests * refactor tests * fix test pure pipeline * remove test pipeline and cutdown launch process

…/shardformer-models

…ipeline test into one file (hpcaitech#4354) * add naive optimizer for 3DPlugin/refactor gpt2 shardformer test * merge tests of PP/DP/TP combinations into one test file * fix bug when sync grad for dp in HybridPlugin * update supported precisions for 3DPlugin/fix bug when shifting tp_degree * improve the passing of lazy_init * modify lazy_init/use sync_shared_params

…models [Merge] Feature/shardformer models to feature/pipeline

…ech#4369) * fix llama test * fix test bug of bert, blip2, bloom, gpt2 * fix llama test * fix opt test * fix sam test * fix sam test * fix t5 test * fix vit test * fix whisper test * fix whisper test * polish code * adjust allclose parameter * Add mistakenly deleted code * addjust allclose * change loss function for some base model

…d_param (hpcaitech#4366) * add util functions for shardformer tests & rewrite gpt2 test * fix shared_params & embedding/merging * fix precision

* add pipeline policy and bert forward to be done * add bertmodel pipeline forward and make tests * add Bert_Policy and test for policy * update formatting * update formatting * update the code * fix bugs * fix name confilt * add bloom model and policy ,revise the base class of policy * revise * revision * add bert_for_pretraining * add bert_for_pretraining forward and policy * fix typos * cancel warning * change the imediate output to default dict * change the default output of get_shared_params * add chatglm * add * chatglm * chatglm * finish chatglm * deletes * fix rmsnorm * chatglm * fix chatglm shard * init

…ech#4362) * [shardformer] supported flash attention test dependency (hpcaitech#4158) * [shardformer] fix flash attention utils test (hpcaitech#4180) * [shardformer] opt support flash attention (hpcaitech#4163) * [shardformer] opt support flash attention * [shardformer] opt support flash attention * [shardformer] opt support flash attention * [shardformer] opt support flash attention * [shardformer] move to modeling * [shardformer] move to modeling * [shardformer] add performance benchmark of shardformer (hpcaitech#4175) * [shardformer] opt support flash attention * [shardformer] opt support flash attention * [shardformer] opt support flash attention * [shardformer] benchmark fix * [shardformer] benchmark fix * [shardformer] llama support flash attention (hpcaitech#4185) * [shardformer] opt support flash attention * [shardformer] opt support flash attention * [shardformer] opt support flash attention * [shardformer] opt support flash attention * [shardformer] move to modeling * [shardformer] move to modeling * [shardformer] llama support flash attention * [shardformer] llama support flash attention * [shardformer] Move the import statement for xformer outside the forward function. * [shardformer] gpt2 support flash attention. (hpcaitech#4191) * [shardformer] opt support flash attention * [shardformer] opt support flash attention * [shardformer] opt support flash attention * [shardformer] opt support flash attention * [shardformer] move to modeling * [shardformer] move to modeling * [shardformer] gpt2 support flash attention * [shardformer] gpt2 support flash attention * [shardformer] gpt2 support flash attention * [shardformer] bloom support flash attention (hpcaitech#4188) * [shardformer] opt support flash attention * [shardformer] opt support flash attention * [shardformer] opt support flash attention * [shardformer] opt support flash attention * [shardformer] move to modeling * [shardformer] move to modeling * [shardformer] bloom suport flash attention * [shardformer] add assert to sequence length * [shardformer] fix * [shardformer] fix * [shardformer] fix * [shardformer] bert support flash attention. (hpcaitech#4206) * [shardformer] opt support flash attention * [shardformer] opt support flash attention * [shardformer] opt support flash attention * [shardformer] opt support flash attention * [shardformer] move to modeling * [shardformer] move to modeling * [shardformer] bert support flash attention * [shardformer] t5 support flash attention. (hpcaitech#4216) * [shardformer] opt support flash attention * [shardformer] opt support flash attention * [shardformer] opt support flash attention * [shardformer] opt support flash attention * [shardformer] move to modeling * [shardformer] move to modeling * [shardformer] t5 support flash attention * [shardformer] t5 support flash attention * fix typo * fix typo * fix typo * fix typo * fix typo * fix typo * [shardformer] support 'paddedcausal' type of attention mask in Coloattention. (hpcaitech#4215) * added padded causal attn mask type for ColoAttention * [shardformer]t5 flash attention fix (hpcaitech#4239) * [shardformer] opt support flash attention * [shardformer] opt support flash attention * [shardformer] opt support flash attention * [shardformer] opt support flash attention * [shardformer] move to modeling * [shardformer] move to modeling * [shardformer] t5 flash attention fix * [shardformer] update gpt2 to use coloattention. (hpcaitech#4234) * [shardformer] opt support flash attention * [shardformer] opt support flash attention * [shardformer] opt support flash attention * [shardformer] opt support flash attention * [shardformer] move to modeling * [shardformer] move to modeling * [shardformer] update gpt2 to use coloattention * [shardformer] update gpt2 to use coloattention * [shardformer] update gpt2 to use coloattention * [shardformer] update gpt2 to use coloattention * [shardformer] update gpt2 * [shardformer] update opt and llama to use coloattention. (hpcaitech#4226) * [shardformer] opt support flash attention * [shardformer] opt support flash attention * [shardformer] opt support flash attention * [shardformer] opt support flash attention * [shardformer] move to modeling * [shardformer] move to modeling * update opt to use coloattention * [shardformer]update opt to use coloattention * [shardformer]update opt to use coloattention * [shardformer]update opt to use coloattention * [shardformer]update opt to use coloattention * [shardformer]update opt to use coloattention * [shardformer]update opt to use coloattention * [shardformer]update opt * [shardformer] shardformer support jit fused operator. (hpcaitech#4236) * [shardformer] opt support flash attention * [shardformer] opt support flash attention * [shardformer] opt support flash attention * [shardformer] opt support flash attention * [shardformer] move to modeling * [shardformer] move to modeling * [shardformer] bloom support jit fused operator * [shardformer] bloom support jit fused operator * [shardformer] bloom support jit fused operator * [shardformer] t5 support jit fused operator * [shardformer] t5 support jit fused operator * [shardformer] t5 support jit fused operator * [shardformer] add roadmap of flash attention * [shardformer] add roadmap of flash attention * [shardformer] add roadmap of flash attention * [shardformer] add type hint to 'self' param of forward * [shardformer] merge feature/shardformer-models branch to feature/flash-attention-shardformer branch. (hpcaitech#4290) * Feature/vit support (hpcaitech#4182) * [shardformer] added tests * [shardformer] vit test finish and support * fix attention dropout * [shardformer] support SAM (hpcaitech#4231) * 1.support sam 2.add fused qkv for nn.Linear * update utils support set element in list * overtwrite SamVisionAttention foward to use DropoutForParallelInput * remove unused code * [shardformer] support whisper (hpcaitech#4212) * support whisper * fix bug in vocabembedding * support downstream model of whisper * update readme * Feature/chatglm (hpcaitech#4240) * [shardformer] added tests * [shardformer] vit test finish and support * [shardformer] chatglm ready * import chatglm * [shardformer] add test kit in model zoo for chatglm * [sharformer] add first version of policy of chatglm * [shardformer] polish chatglm code * [shardformer] polish code * [shardformer] support chatglm without layernorm * [shardformer] chatglm shard without mlp sharding * [shardformer] delete some file * [shardformer] ChatGLM support layernorm sharding * [shardformer] register without auto policy * [shardformer] pre-commit check files * [shardformer] fix chatglm configuration with pre-commit --------- Co-authored-by: Kun Lin <[email protected]> Co-authored-by: FoolPlayer <[email protected]> * [shardformer] whisper support flash attention (hpcaitech#4301) * Feature/vit support (hpcaitech#4182) * [shardformer] added tests * [shardformer] vit test finish and support * fix attention dropout * [shardformer] support SAM (hpcaitech#4231) * 1.support sam 2.add fused qkv for nn.Linear * update utils support set element in list * overtwrite SamVisionAttention foward to use DropoutForParallelInput * remove unused code * [shardformer] support whisper (hpcaitech#4212) * support whisper * fix bug in vocabembedding * support downstream model of whisper * update readme * Feature/chatglm (hpcaitech#4240) * [shardformer] added tests * [shardformer] vit test finish and support * [shardformer] chatglm ready * import chatglm * [shardformer] add test kit in model zoo for chatglm * [sharformer] add first version of policy of chatglm * [shardformer] polish chatglm code * [shardformer] polish code * [shardformer] support chatglm without layernorm * [shardformer] chatglm shard without mlp sharding * [shardformer] delete some file * [shardformer] ChatGLM support layernorm sharding * [shardformer] register without auto policy * [shardformer] pre-commit check files * [shardformer] fix chatglm configuration with pre-commit * [shardformer] whisper support flash attention * [shardformer] whisper support flash attention * [shardformer]whisper support jit operator --------- Co-authored-by: Kun Lin <[email protected]> Co-authored-by: FoolPlayer <[email protected]> * [shardformer] sam support flash attention (hpcaitech#4316) * Feature/vit support (hpcaitech#4182) * [shardformer] added tests * [shardformer] vit test finish and support * fix attention dropout * [shardformer] support SAM (hpcaitech#4231) * 1.support sam 2.add fused qkv for nn.Linear * update utils support set element in list * overtwrite SamVisionAttention foward to use DropoutForParallelInput * remove unused code * [shardformer] support whisper (hpcaitech#4212) * support whisper * fix bug in vocabembedding * support downstream model of whisper * update readme * Feature/chatglm (hpcaitech#4240) * [shardformer] added tests * [shardformer] vit test finish and support * [shardformer] chatglm ready * import chatglm * [shardformer] add test kit in model zoo for chatglm * [sharformer] add first version of policy of chatglm * [shardformer] polish chatglm code * [shardformer] polish code * [shardformer] support chatglm without layernorm * [shardformer] chatglm shard without mlp sharding * [shardformer] delete some file * [shardformer] ChatGLM support layernorm sharding * [shardformer] register without auto policy * [shardformer] pre-commit check files * [shardformer] fix chatglm configuration with pre-commit * [shardformer] sam support flash attention --------- Co-authored-by: Kun Lin <[email protected]> Co-authored-by: FoolPlayer <[email protected]> * [shardformer] merge blip2/chatglm (hpcaitech#4321) * Feature/vit support (hpcaitech#4182) * [shardformer] added tests * [shardformer] vit test finish and support * fix attention dropout * [shardformer] support SAM (hpcaitech#4231) * 1.support sam 2.add fused qkv for nn.Linear * update utils support set element in list * overtwrite SamVisionAttention foward to use DropoutForParallelInput * remove unused code * [shardformer] support whisper (hpcaitech#4212) * support whisper * fix bug in vocabembedding * support downstream model of whisper * update readme * Feature/chatglm (hpcaitech#4240) * [shardformer] added tests * [shardformer] vit test finish and support * [shardformer] chatglm ready * import chatglm * [shardformer] add test kit in model zoo for chatglm * [sharformer] add first version of policy of chatglm * [shardformer] polish chatglm code * [shardformer] polish code * [shardformer] support chatglm without layernorm * [shardformer] chatglm shard without mlp sharding * [shardformer] delete some file * [shardformer] ChatGLM support layernorm sharding * [shardformer] register without auto policy * [shardformer] pre-commit check files * [shardformer] fix chatglm configuration with pre-commit * [shardformer] added tests * [shardformer] vit test finish and support * import chatglm * [shardformer] add test kit in model zoo for chatglm * [sharformer] add first version of policy of chatglm * [shardformer] polish chatglm code * [shardformer] polish code * [shardformer] support chatglm without layernorm * [shardformer] delete some file * [shardformer] ChatGLM support layernorm sharding * [shardformer] register without auto policy * [shardformer] pre-commit check files * [shardformer] support ChatGLMForConditionalGeneration & add fusedlayernorm for vit * [shardformer] support Blip2 (hpcaitech#4243) * support base blip2 * add support for downstream blip2 model * update readme * add forward injection * skip not compatible models test * fix test for gemini and low_level_zero_pugin --------- Co-authored-by: Kun Lin <[email protected]> Co-authored-by: FoolPlayer <[email protected]> Co-authored-by: klhhhhh <[email protected]> * [shardformer] blip2 support flash attention and jit operator (hpcaitech#4325) * Feature/vit support (hpcaitech#4182) * [shardformer] added tests * [shardformer] vit test finish and support * fix attention dropout * [shardformer] support SAM (hpcaitech#4231) * 1.support sam 2.add fused qkv for nn.Linear * update utils support set element in list * overtwrite SamVisionAttention foward to use DropoutForParallelInput * remove unused code * [shardformer] support whisper (hpcaitech#4212) * support whisper * fix bug in vocabembedding * support downstream model of whisper * update readme * Feature/chatglm (hpcaitech#4240) * [shardformer] added tests * [shardformer] vit test finish and support * [shardformer] chatglm ready * import chatglm * [shardformer] add test kit in model zoo for chatglm * [sharformer] add first version of policy of chatglm * [shardformer] polish chatglm code * [shardformer] polish code * [shardformer] support chatglm without layernorm * [shardformer] chatglm shard without mlp sharding * [shardformer] delete some file * [shardformer] ChatGLM support layernorm sharding * [shardformer] register without auto policy * [shardformer] pre-commit check files * [shardformer] fix chatglm configuration with pre-commit * [shardformer] added tests * [shardformer] vit test finish and support * import chatglm * [shardformer] add test kit in model zoo for chatglm * [sharformer] add first version of policy of chatglm * [shardformer] polish chatglm code * [shardformer] polish code * [shardformer] support chatglm without layernorm * [shardformer] delete some file * [shardformer] ChatGLM support layernorm sharding * [shardformer] register without auto policy * [shardformer] pre-commit check files * [shardformer] support ChatGLMForConditionalGeneration & add fusedlayernorm for vit * [shardformer] support Blip2 (hpcaitech#4243) * support base blip2 * add support for downstream blip2 model * update readme * add forward injection * skip not compatible models test * fix test for gemini and low_level_zero_pugin * [shardformer] blip2 support flash attention and jit operator * [shardformer] blip2 support flash attention and jit operator * [shardformer] blip2 support flash attention and jit operator --------- Co-authored-by: Kun Lin <[email protected]> Co-authored-by: FoolPlayer <[email protected]> Co-authored-by: klhhhhh <[email protected]> * [shardformer] chatglm support flash attention and jit operator (hpcaitech#4330) * Feature/vit support (hpcaitech#4182) * [shardformer] added tests * [shardformer] vit test finish and support * fix attention dropout * [shardformer] support SAM (hpcaitech#4231) * 1.support sam 2.add fused qkv for nn.Linear * update utils support set element in list * overtwrite SamVisionAttention foward to use DropoutForParallelInput * remove unused code * [shardformer] support whisper (hpcaitech#4212) * support whisper * fix bug in vocabembedding * support downstream model of whisper * update readme * Feature/chatglm (hpcaitech#4240) * [shardformer] added tests * [shardformer] vit test finish and support * [shardformer] chatglm ready * import chatglm * [shardformer] add test kit in model zoo for chatglm * [sharformer] add first version of policy of chatglm * [shardformer] polish chatglm code * [shardformer] polish code * [shardformer] support chatglm without layernorm * [shardformer] chatglm shard without mlp sharding * [shardformer] delete some file * [shardformer] ChatGLM support layernorm sharding * [shardformer] register without auto policy * [shardformer] pre-commit check files * [shardformer] fix chatglm configuration with pre-commit * [shardformer] added tests * [shardformer] vit test finish and support * import chatglm * [shardformer] add test kit in model zoo for chatglm * [sharformer] add first version of policy of chatglm * [shardformer] polish chatglm code * [shardformer] polish code * [shardformer] support chatglm without layernorm * [shardformer] delete some file * [shardformer] ChatGLM support layernorm sharding * [shardformer] register without auto policy * [shardformer] pre-commit check files * [shardformer] support ChatGLMForConditionalGeneration & add fusedlayernorm for vit * [shardformer] support Blip2 (hpcaitech#4243) * support base blip2 * add support for downstream blip2 model * update readme * add forward injection * skip not compatible models test * fix test for gemini and low_level_zero_pugin * [shardformer] chatglm support flash attention and jit operator * [shardformer] chatglm support flash attention and jit operator * [shardformer] chatglm support flash attention and jit operator * [shardformer] chatglm support flash attention and jit operator --------- Co-authored-by: Kun Lin <[email protected]> Co-authored-by: FoolPlayer <[email protected]> Co-authored-by: klhhhhh <[email protected]> * [shardformer] vit support flash attention and jit operator (hpcaitech#4334) * Feature/vit support (hpcaitech#4182) * [shardformer] added tests * [shardformer] vit test finish and support * fix attention dropout * [shardformer] support SAM (hpcaitech#4231) * 1.support sam 2.add fused qkv for nn.Linear * update utils support set element in list * overtwrite SamVisionAttention foward to use DropoutForParallelInput * remove unused code * [shardformer] support whisper (hpcaitech#4212) * support whisper * fix bug in vocabembedding * support downstream model of whisper * update readme * Feature/chatglm (hpcaitech#4240) * [shardformer] added tests * [shardformer] vit test finish and support * [shardformer] chatglm ready * import chatglm * [shardformer] add test kit in model zoo for chatglm * [sharformer] add first version of policy of chatglm * [shardformer] polish chatglm code * [shardformer] polish code * [shardformer] support chatglm without layernorm * [shardformer] chatglm shard without mlp sharding * [shardformer] delete some file * [shardformer] ChatGLM support layernorm sharding * [shardformer] register without auto policy * [shardformer] pre-commit check files * [shardformer] fix chatglm configuration with pre-commit * [shardformer] added tests * [shardformer] vit test finish and support * import chatglm * [shardformer] add test kit in model zoo for chatglm * [sharformer] add first version of policy of chatglm * [shardformer] polish chatglm code * [shardformer] polish code * [shardformer] support chatglm without layernorm * [shardformer] delete some file * [shardformer] ChatGLM support layernorm sharding * [shardformer] register without auto policy * [shardformer] pre-commit check files * [shardformer] support ChatGLMForConditionalGeneration & add fusedlayernorm for vit * [shardformer] support Blip2 (hpcaitech#4243) * support base blip2 * add support for downstream blip2 model * update readme * add forward injection * skip not compatible models test * fix test for gemini and low_level_zero_pugin * [shardformer] vit support flash attention and jit operator * [shardformer] vit support flash attention and jit operator --------- Co-authored-by: Kun Lin <[email protected]> Co-authored-by: FoolPlayer <[email protected]> Co-authored-by: klhhhhh <[email protected]> * [pipeline] merge flash attention branch * [pipeline] merge flash attention branch * [pipeline] merge flash attention branch * [pipeline] fix conflict * [pipeline] fix conflict * Merge branch 'feature/pipeline' into feature/pipeline * Merge branch 'feature/pipeline' into feature/pipeline * Merge branch 'feature/pipeline' into feature/pipeline * activate checks * activate checks * activate checks * activate checks * activate checks * activate checks * activate checks * activate checks * fix flash attention tests * gemini ignore whisper * fix vit * fix xformers import handle --------- Co-authored-by: Frank Lee <[email protected]> Co-authored-by: Kun Lin <[email protected]> Co-authored-by: FoolPlayer <[email protected]> Co-authored-by: klhhhhh <[email protected]>

…peline (hpcaitech#4388) * fix remaining t5 bugs/rewrite t5 tests * fix multi-tensor communication in pipeline * rearrange test_config * fix keyerror in sync_shared_params * fix get_held_layers & Randomnizer, complete t5 tests * erase printing * fix get_held_layers through modifying _release_unheld_layers * fix _get_recursive_held_layers bug

…4392) * cherry-pick flash attention 2 cherry-pick flash attention 2 * [shardformer] update shardformer to use flash attention 2 [shardformer] update shardformer to use flash attention 2, fix [shardformer] update shardformer to use flash attention 2, fix [shardformer] update shardformer to use flash attention 2, fix

[shardformer] test all optimizations [shardformer] test all optimizations [shardformer] test all optimizations

* add pipeline policy and bert forward to be done * add bertmodel pipeline forward and make tests * add Bert_Policy and test for policy * update formatting * update formatting * update the code * fix bugs * fix name confilt * add bloom model and policy ,revise the base class of policy * revise * revision * add bert_for_pretraining * add bert_for_pretraining forward and policy * fix typos * cancel warning * change the imediate output to default dict * change the default output of get_shared_params * rewrite bert test * rewrite bert test * fix some bugs * del pipeline tests * del pipeline tests * del useless print * del useless print * rewrite data repeats

* [shardformer] gpt2 tests fix [shardformer] test all optimizations (hpcaitech#4399) [shardformer] test all optimizations [shardformer] test all optimizations [shardformer] test all optimizations [shardformer] gpt2 tests fix * [shardformer] gpt2 tests fix

…h#4395) * rewrite opt tests * rewrite llama tests * rewrite bloom & vit tests * rewrite chatglm tests * fix LinearCol for classfiers * add judge for other tp layers, fix lazy init in util

[shardformer] update tests for all optimization

…4407) * [shardformer] gpt2 tests fix [shardformer] test all optimizations (hpcaitech#4399) [shardformer] test all optimizations [shardformer] test all optimizations [shardformer] test all optimizations [shardformer] gpt2 tests fix * [shardformer]update t5 to use all optimizations

[shardformer] update bloom/llama/vit/chatglm tests [shardformer] update opt tests [shardformer] update opt tests [shardformer] update bloom/llama/vit/chatglm tests [shardformer] update bloom/llama/vit/chatglm tests [shardformer] update bloom/llama/vit/chatglm tests

[sync] update pipeline branch with main

…to merge_pp

github-actions · 2023-08-14T15:16:07Z

The code coverage for the changed files is 85%.

Click me to view the complete report

Name                                                                    Stmts   Miss  Cover
-------------------------------------------------------------------------------------------
colossalai/amp/naive_amp/mixed_precision_optimizer.py                      98     20    80%
colossalai/booster/booster.py                                              66      9    86%
colossalai/booster/plugin/__init__.py                                      11      0   100%
colossalai/booster/plugin/hybrid_parallel_plugin.py                       152     17    89%
colossalai/booster/plugin/pp_plugin_base.py                                 9      1    89%
colossalai/cluster/__init__.py                                              5      0   100%
colossalai/cluster/process_group_mesh.py                                   72      1    99%
colossalai/engine/gradient_handler/__init__.py                              6      0   100%
colossalai/interface/optimizer.py                                          45      5    89%
colossalai/kernel/cuda_native/__init__.py                                   5      0   100%
colossalai/lazy/lazy_init.py                                              315     44    86%
colossalai/nn/layer/moe/experts.py                                        128    104    19%
colossalai/pipeline/p2p.py                                                102      7    93%
colossalai/pipeline/schedule/__init__.py                                    3      0   100%
colossalai/pipeline/schedule/_utils.py                                     50      5    90%
colossalai/pipeline/schedule/base.py                                       10      1    90%
colossalai/pipeline/schedule/one_f_one_b.py                               116      4    97%
colossalai/pipeline/stage_manager.py                                       68      4    94%
colossalai/shardformer/_utils.py                                           54     15    72%
colossalai/shardformer/layer/__init__.py                                    8      0   100%
colossalai/shardformer/layer/embedding.py                                 130     24    82%
colossalai/shardformer/layer/linear.py                                    181     53    71%
colossalai/shardformer/layer/normalization.py                              51     10    80%
colossalai/shardformer/layer/qkv_fused_linear.py                          292     70    76%
colossalai/shardformer/layer/utils.py                                      84     17    80%
colossalai/shardformer/modeling/bert.py                                   431    128    70%
colossalai/shardformer/modeling/blip2.py                                   53      1    98%
colossalai/shardformer/modeling/bloom.py                                  387    122    68%
colossalai/shardformer/modeling/chatglm2_6b/configuration_chatglm.py       30      0   100%
colossalai/shardformer/modeling/chatglm2_6b/modeling_chatglm.py           571    239    58%
colossalai/shardformer/modeling/chatglm.py                                149     34    77%
colossalai/shardformer/modeling/gpt2.py                                   293     83    72%
colossalai/shardformer/modeling/jit.py                                     19      3    84%
colossalai/shardformer/modeling/llama.py                                  204     65    68%
colossalai/shardformer/modeling/opt.py                                    285     65    77%
colossalai/shardformer/modeling/sam.py                                     94      6    94%
colossalai/shardformer/modeling/t5.py                                     297     74    75%
colossalai/shardformer/modeling/vit.py                                    149     23    85%
colossalai/shardformer/modeling/whisper.py                                 95     13    86%
colossalai/shardformer/policies/auto_policy.py                             27      2    93%
colossalai/shardformer/policies/base_policy.py                             87     11    87%
colossalai/shardformer/policies/bert.py                                   257      0   100%
colossalai/shardformer/policies/blip2.py                                   54      2    96%
colossalai/shardformer/policies/bloom.py                                  151      2    99%
colossalai/shardformer/policies/chatglm.py                                100      6    94%
colossalai/shardformer/policies/gpt2.py                                   181      1    99%
colossalai/shardformer/policies/llama.py                                  114      3    97%
colossalai/shardformer/policies/opt.py                                    140      2    99%
colossalai/shardformer/policies/sam.py                                     32      0   100%
colossalai/shardformer/policies/t5.py                                     182      5    97%
colossalai/shardformer/policies/vit.py                                    108      1    99%
colossalai/shardformer/policies/whisper.py                                 61      2    97%
colossalai/shardformer/shard/shard_config.py                               28      0   100%
colossalai/shardformer/shard/sharder.py                                    95      3    97%
colossalai/shardformer/shard/shardformer.py                                15      0   100%
colossalai/shardformer/shard/utils.py                                      11      0   100%
colossalai/tensor/d_tensor/api.py                                         149     24    84%
colossalai/tensor/moe_tensor/api.py                                         5      1    80%
colossalai/zero/low_level/low_level_optim.py                              330     30    91%
tests/kit/model_zoo/transformers/__init__.py                               12      0   100%
tests/kit/model_zoo/transformers/bert.py                                   50      0   100%
tests/kit/model_zoo/transformers/blip2.py                                  21      0   100%
tests/kit/model_zoo/transformers/bloom.py                                  36      0   100%
tests/kit/model_zoo/transformers/chatglm.py                                20      0   100%
tests/kit/model_zoo/transformers/gpt.py                                    39      0   100%
tests/kit/model_zoo/transformers/opt.py                                    32      0   100%
tests/kit/model_zoo/transformers/sam.py                                    14      0   100%
tests/kit/model_zoo/transformers/t5.py                                     25      0   100%
tests/kit/model_zoo/transformers/vit.py                                    24      0   100%
tests/kit/model_zoo/transformers/whisper.py                                23      0   100%
tests/test_booster/test_plugin/test_3d_plugin.py                           64      7    89%
tests/test_booster/test_plugin/test_gemini_plugin.py                       74     10    86%
tests/test_cluster/test_process_group_mesh.py                              86      1    99%
tests/test_fx/test_tracer/test_hf_model/hf_tracer_utils.py                 21      2    90%
tests/test_fx/test_tracer/test_hf_model/test_hf_bert.py                    17      1    94%
tests/test_fx/test_tracer/test_hf_model/test_hf_gpt.py                     17      1    94%
tests/test_lazy/test_models.py                                             14      1    93%
tests/test_pipeline/test_p2p_communication.py                              44      1    98%
tests/test_pipeline/test_schedule/test_oneF_oneB.py                        80      2    98%
tests/test_pipeline/test_schedule/test_pipeline_schedule_utils.py          40      0   100%
tests/test_pipeline/test_stage_manager.py                                  52      1    98%
tests/test_shardformer/test_layer/test_embedding.py                        37      1    97%
tests/test_shardformer/test_layer/test_gpt2_qkv_fused_linear_1d.py         89      1    99%
tests/test_shardformer/test_layer/test_layernorm.py                        35      1    97%
tests/test_shardformer/test_layer/test_linear_1d.py                       110      1    99%
tests/test_shardformer/test_layer/test_qkv_fused_linear_1d.py              51      1    98%
tests/test_shardformer/test_layer/test_vocab_parallel_embedding_1d.py      39      1    97%
tests/test_shardformer/test_model/_utils.py                               142     21    85%
tests/test_shardformer/test_model/test_shard_bert.py                       62      1    98%
tests/test_shardformer/test_model/test_shard_blip2.py                      40      1    98%
tests/test_shardformer/test_model/test_shard_bloom.py                      59      1    98%
tests/test_shardformer/test_model/test_shard_chatglm.py                    60      1    98%
tests/test_shardformer/test_model/test_shard_gpt2.py                       65      1    98%
tests/test_shardformer/test_model/test_shard_llama.py                      62      1    98%
tests/test_shardformer/test_model/test_shard_opt.py                        62      1    98%
tests/test_shardformer/test_model/test_shard_sam.py                        39      1    97%
tests/test_shardformer/test_model/test_shard_t5.py                         59      1    98%
tests/test_shardformer/test_model/test_shard_vit.py                        61      1    98%
tests/test_shardformer/test_model/test_shard_whisper.py                    46      1    98%
tests/test_shardformer/test_shard_utils.py                                 21      0   100%
tests/test_shardformer/test_with_torch_ddp.py                              52      1    98%
tests/test_utils/test_flash_attention.py                                   92      8    91%
-------------------------------------------------------------------------------------------
TOTAL                                                                    9333   1436    85%

ver217 and others added 30 commits June 20, 2023 10:39

[cluster] add process group mesh (hpcaitech#4039)

1015f04

* [cluster] add process group mesh * [test] add process group mesh test * force sync

[pipeline] add stage manager (hpcaitech#4093)

b10821a

* [pipeline] add stage manager * [test] add pipeline stage manager test * [pipeline] add docstring for stage manager

[pipeline] implement p2p communication (hpcaitech#4100)

bd6b0a3

* [pipeline] add p2p communication * [test] add p2p communication test * [test] add rerun decorator * [test] rename to avoid conflict

[pipeline] refactor 1f1b schedule (hpcaitech#4115)

faeac9d

* [api] update optimizer wrapper to fit pipeline * [pipeline] add base schedule * [pipeline] add 1f1b schedule * [test] add pipeline schedule utils test * [pipeline] fix import

[cluster] add process group mesh (hpcaitech#4039)

3be0c35

* [cluster] add process group mesh * [test] add process group mesh test * force sync

[pipeline] add stage manager (hpcaitech#4093)

18c7539

* [pipeline] add stage manager * [test] add pipeline stage manager test * [pipeline] add docstring for stage manager

[pipeline] implement p2p communication (hpcaitech#4100)

5a467e9

* [pipeline] add p2p communication * [test] add p2p communication test * [test] add rerun decorator * [test] rename to avoid conflict

[pipeline] refactor 1f1b schedule (hpcaitech#4115)

9526f44

* [api] update optimizer wrapper to fit pipeline * [pipeline] add base schedule * [pipeline] add 1f1b schedule * [test] add pipeline schedule utils test * [pipeline] fix import

Merge pull request hpcaitech#4166 from ver217/sync/main

ef1f972

[sync] update from main

[pipeline] update shardformer policy

15a4e82

[pipeline] update shardformer docstring

8b6679d

[test] update shardformer tests

9143556

[test] add shard util tests

0cbe423

[shardformer] rename policy file name

1a87dd7

[shardformer] fix type hint

d4b96ab

Merge pull request hpcaitech#4176 from ver217/feature/pipeline-policy

12e6d5d

[pipeline] fit shardformer policy

Feature/vit support (hpcaitech#4182)

ec217de

* [shardformer] added tests * [shardformer] vit test finish and support * fix attention dropout

[pipeline] move bert related pipeline components to shardformer (hpca…

c6f9c2c

…itech#4187) * move bert related pipeline components to shardformer * fix bugs * revision * fix bert model tests * fix bert_lm_head model tests * fix tests * fix tests * done checks * skip bloom

[pipeline] Bert pipeline for shardformer and its tests (hpcaitech#4197)

b30d1b9

* add pipeline forward * complete pipeline forward check * fix bert forward without pipeline * fix comments * discard useless line * add todo * clean prints * fix distribute layers

[pipeline] add bloom model pipeline (hpcaitech#4210)

3595eba

* bloom policy * llama pipeline forward and tests * fix the output and attention_mask * fix name * bind argument to policy * finish bloom model * test shard gpt2 * clear cache

[pipeline] Add Pipeline Forward for GPT2Model Shardformer (hpcaitech#…

236f294

…4224) * * fix typehint & docstring in sharder.py * * update pipeline forward for GPT2Model * * add test for pipeline forward of GPT2Model * * add cache cleaning in gpt2 test * * change assert to raise command

[shardformer] fix base policy (hpcaitech#4229)

ad2687c

[shardformer] support SAM (hpcaitech#4231)

ddecf73

* 1.support sam 2.add fused qkv for nn.Linear * update utils support set element in list * overtwrite SamVisionAttention foward to use DropoutForParallelInput * remove unused code

ver217 and others added 26 commits July 27, 2023 13:11

[hotfix] fix gemini and zero test (hpcaitech#4333)

8ad05d1

* [hotfix] fix gemini and zero test * [hotfix] fix lazy init test * [hotfix] fix lazy init test

[pipeline] fix return_dict/fix pure_pipeline_test (hpcaitech#4331)

d547377

[pipeline] add unit test for 1f1b (hpcaitech#4303)

b941e65

* add unit test for 1f1b * polish code * polish code and update ut version * fix

Merge remote-tracking branch 'upstream/feature/pipeline' into feature…

01ef6c5

…/shardformer-models

update some module with new api version

260df9e

[test] skip some not compatible models

5403578

Merge pull request hpcaitech#4358 from hpcaitech/feature/shardformer-…

b849657

…models [Merge] Feature/shardformer models to feature/pipeline

[shardformer] add util functions for shardformer tests/fix sync_share…

21c6bb0

…d_param (hpcaitech#4366) * add util functions for shardformer tests & rewrite gpt2 test * fix shared_params & embedding/merging * fix precision

[shardformer] test all optimizations (hpcaitech#4399)

ed2c229

[shardformer] test all optimizations [shardformer] test all optimizations [shardformer] test all optimizations

[shardformer] rewrite tests for opt/bloom/llama/vit/chatglm (hpcaitec…

1e518ae

…h#4395) * rewrite opt tests * rewrite llama tests * rewrite bloom & vit tests * rewrite chatglm tests * fix LinearCol for classfiers * add judge for other tp layers, fix lazy init in util

[shardformer] update tests for all optimization (hpcaitech#4413)

d4a3a10

[shardformer] update tests for all optimization

Merge branch 'main' into feature/pipeline

6990477

Merge pull request hpcaitech#4424 from ver217/sync/pipeline

60db2cc

[sync] update pipeline branch with main

[misc] resolve code factor issues (hpcaitech#4433)

9d1a6d2

Merge remote-tracking branch 'remotes/ColossalAI/feature/pipeline' in…

4f095e6

…to merge_pp

ver217 approved these changes Aug 15, 2023

View reviewed changes

ver217 merged commit 1c73ed6 into hpcaitech:feature/moe Aug 15, 2023
6 checks passed

oahzxl deleted the merge_pp branch August 16, 2023 03:45

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

[moe] merge branch feature/pipeline #4435

[moe] merge branch feature/pipeline #4435

oahzxl commented Aug 14, 2023

github-actions bot commented Aug 14, 2023

[moe] merge branch feature/pipeline #4435

[moe] merge branch feature/pipeline #4435

Conversation

oahzxl commented Aug 14, 2023

📌 Checklist before creating the PR

🚨 Issue number

📝 What does this PR do?

💥 Checklist before requesting a review

⭐️ Do you enjoy contributing to Colossal-AI?

github-actions bot commented Aug 14, 2023