Focusing
ONNX approver, working at OneFlow now.
PGP Key ID: 0xAE767ADF33D6DBC2746CE7EC845A581542488B59
-
OneFlow
- China
Highlights
- Pro
Block or Report
Block or report daquexian
Report abuse
Contact GitHub support about this user’s behavior. Learn more about reporting abuse.
Report abusePinned
-
Oneflow-Inc/oneflow Public
OneFlow is a deep learning framework designed to be user-friendly, scalable and efficient.
-
-
JDAI-CV/dabnn Public
dabnn is an accelerated binary neural networks inference framework for mobile platform
-
1,187 contributions in the last year
Less
More
Activity overview
Contributed to
Oneflow-Inc/oneflow,
daquexian/onnx-simplifier,
Oneflow-Inc/oneflow-cambricon
and 49 other
repositories
Contribution activity
March 2023
Created 22 commits in 5 repositories
Created a pull request in Oneflow-Inc/oneflow that received 22 comments
fix segfault and infinite loop in python stack getter
#4737 这个很早之前的 PR 在 python 退出后会跳过 tensor 的析构,导致 oom,当时这么做的原因已经不记得了,现在来看不跳过也想不到会有什么问题,所以本 PR 恢复为正常执行析构。
更新:CI 遇到问题,待进一步定位,这个 PR 先不包含这部分改动了。
原 oom 问题…
+23
−39
•
22
comments
Opened 10 other pull requests in 4 repositories
Oneflow-Inc/oneflow
1
open
6
merged
- release np array in main thread to eliminate the time-consuming gil request
- return False in dummy module __bool__ and add safetensors into hazard list
- unset lazy&verbose env var when disabling mock torch by command line
- [Resubmit] Refine tensor storage
- add .clangd for CUDA support by OneFlow's custom clangd
- give small_vector a default size kOpArgsReservedSize
- fix infinite loop on lazy mock torch dummy object
Oneflow-Inc/oneflow-cambricon
1
merged
onnx/optimizer
1
merged
Oneflow-Inc/diffusers
1
merged
Reviewed 49 pull requests in 5 repositories
Oneflow-Inc/oneflow
25 pull requests
- fix save and load
- fix map_location
- Add meta device and skip init function
- supprt fine-grained disable
- fix contiguous paramter buffer conflict
- GetDataDtype support uint64
- modify ParamGroup
- Add UninitializedParameter method to parameter.py
- modified python/oneflow/optim/optimizer.py
- Fix safetensors Typo
- this fixes the bug of jit.ScriptModule not found
- Dev embedding kwargs
- fix scatter reduce default value to None
- return False in dummy module __bool__ and add safetensors into hazard list
- Fix diffusers mock test
- check PYTHONPATH exist
- Support complex64 and complex128 datatype, construt ComplexFloatTensor and ComplexDoubleTensor, CompleDoubleAttr
- update mock disable behaviour
- Feat support host memory
- symbolic shape
- fix segfault and infinite loop in python stack getter
- fix closure make loss nondifferentiable
- update DummyModule
- Fix default generator
- add mock torch docs
- Some pull request reviews not shown.





