Create your own GitHub profile
Sign up for your own profile on GitHub, the best place to host code, manage projects, and build software alongside 50 million developers.
Sign up
-
Facebook
- Palo Alto
- Sign in to view email
Popular repositories
-
-
-
Forked from ccheever/bunny1
bunny1 is a tool that lets you write smart bookmarks in python and then share them across all your browsers and with a group of people or the whole world. It was developed at Facebook and is widely…
-
-
454 contributions in the last year
Contribution activity
May 2020
Created a pull request in pytorch/pytorch that received 2 comments
Remove __future__ imports from many files
Stack from ghstack: #38778 Remove future imports from many files #38777 Remove future import from torch.nn.functional Summary: Python 2 has reach…
+0
−171
•
2
comments
- Add support for float[]? arguments in native_functions.yaml
- Enabling lite interpreter in torch python API
- [vulkan] jni build support USE_VULKAN
- [vulkan] TensorConversions remove explicit vulkan ifs
- [Mobile GPU][Integration] Vulkan backend integration
- Some TODO fixes.
- Enabled per channel quantized static linear/conv
- Interface changes to enable per channel quant.
- Perf optimization for conv and gemm kernels.
- Added per channel separate test cases for fc and deconv tests.
- __torch_function__ overrides for torch.functional and torch.nn.functional
- Enable channelwise quantized test for dynamic linear.
- Changes to enable per channel support on dynamic linear.
- Changes to enable per channel requant.
- Enable per channel zero point.
- Unify PyTorch mobile's threadpool usage.
- [quant] Release qnnpack original weights for conv/linear