f7aecc0c6b
feat rmsnorm cuda kernel and add unittest, benchmark script ( #5417 )
2024-03-08 16:21:12 +08:00
0310b76e9d
Merge branch 'main' into sync/main
2024-03-04 10:09:36 +08:00
0aa27f1961
[Inference]Move benchmark-related code to the example directory. ( #5408 )
...
* move benchmark-related code to the example directory.
* fix bugs in test_fused_rotary_embedding.py
2024-02-28 16:46:03 +08:00
600881a8ea
[Inference]Add CUDA KVCache Kernel ( #5406 )
...
* add cuda KVCache kernel
* annotation benchmark_kvcache_copy
* add use cuda
* fix import path
* move benchmark scripts to example/
* rm benchmark codes in test_kv_cache_memcpy.py
* rm redundancy codes
* rm redundancy codes
* pr was modified according to the review
2024-02-28 14:36:50 +08:00
d882d18c65
[example] reuse flash attn patch ( #5400 )
2024-02-27 11:22:07 +08:00
bc1da87366
[Fix/Inference] Fix format of input prompts and input model in inference engine ( #5395 )
...
* Fix bugs in inference_engine
* fix bugs in engine.py
* rm CUDA_VISIBLE_DEVICES
* add request_ids in generate
* fix bug in engine.py
* add logger.debug for BatchBucket
2024-02-23 10:51:35 +08:00
2a718c8be8
Optimized the execution interval time between cuda kernels caused by view and memcopy ( #5390 )
...
* opt_view_and_memcopy
* fix bugs in ci
* fix ci bugs
* update benchmark scripts
* fix ci bugs
2024-02-21 13:23:57 +08:00
730103819d
[Inference]Fused kv copy into rotary calculation ( #5383 )
...
* revise rotary embedding
* remove useless print
* adapt
* fix
* add
* fix
* modeling
* fix
* fix
* fix
* fused kv copy
* fused copy
* colossalai/kernel/triton/no_pad_rotary_embedding.py
* del padding llama
* del
2024-02-21 11:31:48 +08:00
8c69debdc7
[Inference]Support vllm testing in benchmark scripts ( #5379 )
...
* add vllm benchmark scripts
* fix code style
* update run_benchmark.sh
* fix code style
2024-02-08 15:27:26 +08:00
8106ede07f
Revert "[Inference] Adapt to Fused rotary ( #5348 )" ( #5373 )
...
This reverts commit 9f4ab2eb92
2024-02-07 14:27:04 +08:00
9f4ab2eb92
[Inference] Adapt to Fused rotary ( #5348 )
...
* revise rotary embedding
* remove useless print
* adapt
* fix
* add
* fix
* modeling
* fix
* fix
* fix
2024-02-07 11:36:04 +08:00
631862f339
[Inference]Optimize generation process of inference engine ( #5356 )
...
* opt inference engine
* fix run_benchmark.sh
* fix generate in engine.py
* rollback tesh_inference_engine.py
2024-02-02 15:38:21 +08:00
21ad4a27f9
[Inference/opt]Optimize the mid tensor of RMS Norm ( #5350 )
...
* opt rms_norm
* fix bugs in rms_layernorm
2024-02-02 15:06:01 +08:00
249644c23b
[Inference]Repalce Attention layer and MLP layer by shardformer to optimize the weight transpose operation，add fused_qkv and fused linear_add ( #5340 )
...
* add fused qkv
* replace attn and mlp by shardformer
* fix bugs in mlp
* add docstrings
* fix test_inference_engine.py
* add optimize unbind
* add fused_addmm
* rm squeeze(1)
* refactor codes
* fix ci bugs
* rename ShardFormerLlamaMLP and ShardFormerLlamaAttention
* Removed the dependency on LlamaFlashAttention2
* rollback test_inference_engine.py
2024-02-01 15:49:39 +08:00
c565519913
merge commit
2024-01-31 10:41:47 +08:00
71321a07cf
fix typo change dosen't to doesn't ( #5308 )
2024-01-30 09:57:38 +08:00
8823cc4831
Merge pull request #5310 from hpcaitech/feature/npu
...
Feature/npu
2024-01-29 13:49:39 +08:00
4f28cb43c0
[inference]Optimize the usage of the mid tensors space in flash attn ( #5304 )
...
* opt flash attn
* opt tmp tensor
* fix benchmark_llama
* fix code style
* fix None logic for output tensor
* fix adapted to get_xine_cache
* add comment
* fix ci bugs
* fix some codes
* rm duplicated codes
* rm duplicated codes
* fix code style
* add _get_dtype in config.py
2024-01-26 14:00:10 +08:00
7cfed5f076
[feat] refactored extension module ( #5298 )
...
* [feat] refactored extension module
* polish
* polish
* polish
* polish
* polish
* polish
* polish
* polish
* polish
* polish
2024-01-25 17:01:48 +08:00
bce9499ed3
fix some typo ( #5307 )
2024-01-25 13:56:27 +08:00
bfff9254ac
[inference] Adapted to Rotary Embedding and RMS Norm ( #5283 )
...
* adapted to rotary_embedding
* adapted to nopad rms norm
* fix bugs in benchmark
* fix flash_decoding.py
2024-01-22 10:55:34 +08:00
f7e3f82a7e
fix llama pretrain ( #5287 )
2024-01-19 17:49:02 +08:00
9e2342bde2
[Hotfix] Fix bugs in testing continuous batching ( #5270 )
...
* fix bug
* fix bugs
* fix bugs
* fix bugs and add padding
* add funcs and fix bugs
* fix typos
* fix bugs
* add func
2024-01-18 16:31:14 +08:00
148469348a
Merge branch 'main' into sync/npu
2024-01-18 12:05:21 +08:00
86b63f720c
[Inference]Adapted to the triton attn kernels ( #5264 )
...
* adapted to the triton attn kernels
* fix pad input
* adapted to copy_kv_to_blocked_cache
* fix ci test
* update kv memcpy
* remove print
2024-01-17 16:03:10 +08:00
ef4f0ee854
[hotfix]: add pp sanity check and fix mbs arg ( #5268 )
...
* fix: fix misleading mbs arg
* feat: add pp sanity check
* fix: fix 1f1b sanity check
2024-01-15 15:57:40 +08:00
c174c4fc5f
[doc] fix doc typo ( #5256 )
...
* [doc] fix annotation display
* [doc] fix llama2 doc
2024-01-11 21:01:11 +08:00
d202cc28c0
[npu] change device to accelerator api ( #5239 )
...
* update accelerator
* fix timer
* fix amp
* update
* fix
* update bug
* add error raise
* fix autocast
* fix set device
* remove doc accelerator
* update doc
* update doc
* update doc
* use nullcontext
* update cpu
* update null context
* change time limit for example
* udpate
* update
* update
* update
* [npu] polish accelerator code
---------
Co-authored-by: Xuanlei Zhao <xuanlei.zhao@gmail.com>
Co-authored-by: zxl <43881818+oahzxl@users.noreply.github.com>
2024-01-09 10:20:05 +08:00
dd2c28a323
[npu] use extension for op builder ( #5172 )
...
* update extension
* update cpu adam
* update is
* add doc for cpu adam
* update kernel
* update commit
* update flash
* update memory efficient
* update flash attn
* update flash attention loader
* update api
* fix
* update doc
* update example time limit
* reverse change
* fix doc
* remove useless kernel
* fix
* not use warning
* update
* update
2024-01-08 11:39:16 +08:00
3c0d82b19b
[pipeline]: support arbitrary batch size in forward_only mode ( #5201 )
...
* fix: remove drop last in val & test dataloader
* feat: add run_forward_only, support arbitrary bs
* chore: modify ci script
2024-01-02 23:41:12 +08:00
4fa689fca1
[pipeline]: fix p2p comm, add metadata cache and support llama interleaved pp ( #5134 )
...
* test: add more p2p tests
* fix: remove send_forward_recv_forward as p2p op list need to use the same group
* fix: make send and receive atomic
* feat: update P2PComm fn
* feat: add metadata cache in 1f1b
* feat: add metadata cache in interleaved pp
* feat: modify is_xx_stage fn
* revert: add _broadcast_object_list
* feat: add interleaved pp in llama policy
* feat: set NCCL_BUFFSIZE in HybridParallelPlugin
2023-12-22 10:44:00 +08:00
21aa5de00b
[gemini] hotfix NaN loss while using Gemini + tensor_parallel ( #5150 )
...
* fix
aaa
fix
fix
fix
* fix
* fix
* test ci
* fix ci
fix
2023-12-08 11:10:51 +08:00
177c79f2d1
[doc] add moe news ( #5128 )
...
* [doc] add moe news
* [doc] add moe news
* [doc] add moe news
2023-11-28 17:44:06 +08:00
7172459e74
[shardformer]: support gpt-j, falcon, Mistral and add interleaved pipeline for bert ( #5088 )
...
* [shardformer] implement policy for all GPT-J models and test
* [shardformer] support interleaved pipeline parallel for bert finetune
* [shardformer] shardformer support falcon (#4883 )
* [shardformer]: fix interleaved pipeline for bert model (#5048 )
* [hotfix]: disable seq parallel for gptj and falcon, and polish code (#5093 )
* Add Mistral support for Shardformer (#5103 )
* [shardformer] add tests to mistral (#5105 )
---------
Co-authored-by: Pengtai Xu <henryxu880@gmail.com>
Co-authored-by: ppt0011 <143150326+ppt0011@users.noreply.github.com>
Co-authored-by: flybird11111 <1829166702@qq.com>
Co-authored-by: eric8607242 <e0928021388@gmail.com>
2023-11-28 16:54:42 +08:00
d5661f0f25
[nfc] fix typo change directoty to directory ( #5111 )
2023-11-27 18:25:53 +08:00
3acbf6d496
[npu] add npu support for hybrid plugin and llama ( #5090 )
...
* llama 3d
* update
* fix autocast
2023-11-22 19:23:21 +08:00
aae496631c
[shardformer]fix flash attention, when mask is casual, just don't unpad it ( #5084 )
...
* fix flash attn
* fix
fix
2023-11-22 16:00:07 +08:00
1cd7efc520
[inference] refactor examples and fix schedule ( #5077 )
...
* [setup] refactor infer setup
* [hotfix] fix infenrece behavior on 1 1 gpu
* [exmaple] refactor inference examples
2023-11-21 10:46:03 +08:00
4e3959d316
[hotfix/hybridengine] Fix init model with random parameters in benchmark ( #5074 )
...
* fix init model with random parameters
* fix example
2023-11-20 20:15:25 +08:00
8921a73c90
[format] applied code formatting on changed files in pull request 5067 ( #5072 )
...
Co-authored-by: github-actions <github-actions@github.com>
2023-11-20 19:46:43 +08:00
fb103cfd6e
[inference] update examples and engine ( #5073 )
...
* update examples and engine
* fix choices
* update example
2023-11-20 19:44:52 +08:00
e5ce4c8ea6
[npu] add npu support for gemini and zero ( #5067 )
...
* [npu] setup device utils (#5047 )
* [npu] add npu device support
* [npu] support low level zero
* [test] update npu zero plugin test
* [hotfix] fix import
* [test] recover tests
* [npu] gemini support npu (#5052 )
* [npu] refactor device utils
* [gemini] support npu
* [example] llama2+gemini support npu
* [kernel] add arm cpu adam kernel (#5065 )
* [kernel] add arm cpu adam
* [optim] update adam optimizer
* [kernel] arm cpu adam remove bf16 support
2023-11-20 16:12:41 +08:00
bce919708f
[Kernels]added flash-decoidng of triton ( #5063 )
...
* added flash-decoidng of triton based on lightllm kernel
* add req
* clean
* clean
* delete build.sh
---------
Co-authored-by: cuiqing.li <lixx336@gmail.com>
2023-11-20 13:58:29 +08:00
fd6482ad8c
[inference] Refactor inference architecture ( #5057 )
...
* [inference] support only TP (#4998 )
* support only tp
* enable tp
* add support for bloom (#5008 )
* [refactor] refactor gptq and smoothquant llama (#5012 )
* refactor gptq and smoothquant llama
* fix import error
* fix linear import torch-int
* fix smoothquant llama import error
* fix import accelerate error
* fix bug
* fix import smooth cuda
* fix smoothcuda
* [Inference Refactor] Merge chatglm2 with pp and tp (#5023 )
merge chatglm with pp and tp
* [Refactor] remove useless inference code (#5022 )
* remove useless code
* fix quant model
* fix test import bug
* mv original inference legacy
* fix chatglm2
* [Refactor] refactor policy search and quant type controlling in inference (#5035 )
* [Refactor] refactor policy search and quant type controling in inference
* [inference] update readme (#5051 )
* update readme
* update readme
* fix architecture
* fix table
* fix table
* [inference] udpate example (#5053 )
* udpate example
* fix run.sh
* fix rebase bug
* fix some errors
* update readme
* add some features
* update interface
* update readme
* update benchmark
* add requirements-infer
---------
Co-authored-by: Bin Jia <45593998+FoolPlayer@users.noreply.github.com>
Co-authored-by: Zhongkai Zhao <kanezz620@gmail.com>
2023-11-19 21:05:05 +08:00
bc09b95f50
[exampe] fix llama example' loss error when using gemini plugin ( #5060 )
...
fix llama example
2023-11-18 18:41:58 +08:00
b2ad0d9e8f
[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 )
...
* Use p2p
* Cannot bidirectonal send p2p
* Refactor tensor creation and serialization in P2P
communication
* Fix llama forward args in flash attention
* Add flop estimate from megatron
* Support loading weight not in weight_map when strict=False in hybrid_parallel
* Use send_forward_recv_backward, etc in 1f1b
* Use dataclass for metdata
Remove torch.cuda.synchronize() as suggested
* Add comment about the torch.cuda.synchronize for potential error
* Typo
* Update hybrid_parallel_checkpoint_io.py
* Update p2p.py
* Update one_f_one_b.py
* Update p2p.py
---------
Co-authored-by: flybird11111 <1829166702@qq.com>
2023-11-16 20:15:59 +08:00
28052a71fb
[Kernels]Update triton kernels into 2.1.0 ( #5046 )
...
* update flash-context-attention
* adding kernels
* fix
* reset
* add build script
* add building process
* add llama2 exmaple
* add colossal-llama2 test
* clean
* fall back test setting
* fix test file
* clean
* clean
* clean
---------
Co-authored-by: cuiqing.li <lixx336@gmail.com>
2023-11-16 16:43:15 +08:00
70885d707d
[hotfix] Suport extra_kwargs in ShardConfig ( #5031 )
...
* [refactor]: replace inference args with extra_kwargs in ShardConfig
* modify shardconfig
* polish code
* fix policy bug in llama
* fix bug in auto policy
* remove setattr in ShardConfig
2023-11-10 10:49:50 +08:00
724441279b
[moe]: fix ep/tp tests, add hierarchical all2all ( #4982 )
...
* fix: add warning for EP different behavior
* fix: use shard_data in ep & tp model
* to: add used_capacity
* fix: fix router test
* feat: add create_ep_node_group
* feat: add create_ep_hierarchical_group fn
* feat: add HierarchicalAllToAll
* test: add hierarchical all2all test
* fix: fix test errors
* fix: simplify create_ep_hierarchical_group
* fix: add hierarchical_alltoall arg
* fix: fix environ typo
* revert: revert process mesh order
* to: add todo mark
* fix: skip hierarchical_comm if torch < 1.13.1
2023-11-09 06:31:00 +00:00
f71e63b0f3
[moe] support optimizer checkpoint ( #5015 )
...
* Refactor MoE Manager setup method
* unshard optim ckpt
* optim io
* update transformer version
* update requirements
* update ckpt
* update ckpt
* update ckpt
* fix engine
* fix engine
2023-11-08 15:07:03 +00:00
Steve Luo