MLSys | CTO of @hpcaitech | Ex @Tencent Wechat AI.
-
http://hpcaitech.com/
- Beijing, China
-
11:06
(UTC +08:00) - https://fangjiarui.github.io/
Block or Report
Block or report feifeibear
Report abuse
Contact GitHub support about this user’s behavior. Learn more about reporting abuse.
Report abusePinned
-
hpcaitech/ColossalAI Public
Colossal-AI: A Unified Deep Learning System for Big Model Era
-
Tencent/PatrickStar Public
PatrickStar enables Larger, Faster, Greener Pretrained Models for NLP and democratizes AI for everyone.
-
Tencent/TurboTransformers Public
a fast and user-friendly runtime for transformer inference (Bert, Albert, GPT2, Decoders, etc) on CPU and GPU.
2,252 contributions in the last year
Less
More
Activity overview
Contributed to
hpcaitech/ColossalAI,
hpcaitech/CachedEmbedding,
hpcaitech/EnergonAI
and 47 other
repositories
Contribution activity
January 2023
Created 20 commits in 2 repositories
Created 1 repository
- feifeibear/transformers-rlfh Python
Created a pull request in hpcaitech/ColossalAI that received 4 comments
[builder] reconfig op_builder for pypi install
What's new remove duplicated op_builder. Make it work both for setup and runtime building.
+13
−332
•
4
comments
Opened 13 other pull requests in 1 repository
hpcaitech/ColossalAI
12
merged
1
closed
- [doc] hotfix #2377
- [hotfix] issue #2388
- [builder] correct readme
- [example] gpt, shard init on all processes
- [example] add google doc for benchmark results of GPT
- [example] make gpt example directory more clear
- [example] simplify opt example
- [example] update diffusion readme with official lightning
- [version] 0.1.14 -> 0.2.0
- [builder] MOE builder
- [example] GPT polish readme
- [Gemini] fix the convert_to_torch_module bug
- [NFC] fix typos
Reviewed 44 pull requests in 1 repository
hpcaitech/ColossalAI
25 pull requests
- Update python_ops.py
- [workflow] added missing file change detection output
- [device] find best logical mesh
- [hotfix] fix implement error in diffusers
- [Pipeline] Refine GPT PP Example
- [setup] support pre-build and jit-build of cuda kernels
- [examples] adding tflops to PaLM
- [NFC] polish code format
- [NFC] polish code format
- [NFC] polish batch_norm_handler.py code style
- [gemini] add get static torch model
- [example] upload auto parallel gpt2 demo
- [NFC] polish colossalai/auto_parallel/tensor_shard/deprecated/_utils.…
- Update roberta/README.md
- [examples]adding tp to PaLM
- [setup] make cuda extension build optional
- [setup] remove torch dependency
- Feature/diffusion update diffusion,Dreamblooth
- [NFC] polish colossalai/auto_parallel/tensor_shard/deprecated/op_hand…
- [workflow] removed unused assign reviewer workflow
- [workflow] rebuild cuda kernels when kernel-related files change
- fix dreamblooth format
- [builder] reconfig op_builder for pypi install
- [example] update gemini benchmark bash
- [NFC] polish colossalai/cli/benchmark/__init__.py code style
- Some pull request reviews not shown.
Answered 1 discussion in 1 repository
hpcaitech/ColossalAI
hpcaitech/ColossalAI
-
是否有更高效的大模型训练方法
This contribution was made on Jan 6
5
contributions
in private repositories
Jan 3 – Jan 6





