OSSForks/pytorch
1
0
Fork 0
mirror of https://github.com/zebrajr/pytorch.git synced 2025-12-07 12:21:27 +01:00
Code Issues Actions 38 Packages Projects Releases Wiki Activity

Fix typos under caffe2 directory (#87840)

Browse Source 
This PR fixes typos in `.md` files under caffe2 directory

Pull Request resolved: https://github.com/pytorch/pytorch/pull/87840
Approved by: https://github.com/kit1980
This commit is contained in:
Kazuaki Ishizaki 2022-10-28 04:53:33 +00:00 committed by PyTorch MergeBot
parent e8a97a3721
commit daff5d3556
4 changed files with 5 additions and 5 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
caffe2/mobile/contrib/libopencl-stub/README.md
View File

@ -1,7 +1,7 @@
libopencl-stub
==============
A stub opecl library that dynamically dlopen/dlsyms opencl implementations at runtime based on environment variables. Will be useful when opencl implementations are installed in non-standard paths (say pocl on android)
A stub opencl library that dynamically dlopen/dlsyms opencl implementations at runtime based on environment variables. Will be useful when opencl implementations are installed in non-standard paths (say pocl on android)

2
caffe2/python/onnx/ONNXOpCoverage.md
View File

@ -19,7 +19,7 @@ This doc keeps tracking why operators are not covered by the testcases.
|Atan|||💚OK|
|AveragePool||OK|💚OK|
|BatchNormalization||OK|💚OK|
|Cast|Yes||💔Need extendtion|
|Cast|Yes||💔Need extension|
|Ceil|Yes||💚OK|
|Clip|Yes|OK|💚OK|
|Concat|Yes|OK|💚OK|

4
caffe2/quantization/server/README.md
View File

@ -19,8 +19,8 @@ To compute the quantization parameters of activation tensors, we need to know th
* Floating-point requantization
Unlike gemmlowp using fixed-point operations that emulates floating point operations of requantization, fbgemm just uses single-precison floating-point operations. This is because in x86 just using single-precision floating-point operations is faster. Probably, gemmlowp used pure fixed-point operations for low-end mobile processors. QNNPACK also has similar constraints as gemmlowp and provides multiple options of requantization implementations.
The users could modify the code to use a different requantization implementation to be bit-wise idential to the HW they want to emulate for example. If there're enough requests, we could consider implementing a few popular fixed-point requantization as QNNPACK did.
Unlike gemmlowp using fixed-point operations that emulates floating point operations of requantization, fbgemm just uses single-precision floating-point operations. This is because in x86 just using single-precision floating-point operations is faster. Probably, gemmlowp used pure fixed-point operations for low-end mobile processors. QNNPACK also has similar constraints as gemmlowp and provides multiple options of requantization implementations.
The users could modify the code to use a different requantization implementation to be bit-wise identical to the HW they want to emulate for example. If there're enough requests, we could consider implementing a few popular fixed-point requantization as QNNPACK did.
* 16-bit accumulation with outlier-aware quantization

2
caffe2/release-notes.md
View File

@ -133,7 +133,7 @@ If you're running this all on a cloud computer, you probably won't have a UI or
First configure your cloud server to accept port 8889, or whatever you want, but change the port in the following commands. On AWS you accomplish this by adding a rule to your server's security group allowing a TCP inbound on port 8889. Otherwise you would adjust iptables for this.
Next you launch the Juypter server.
Next you launch the Jupyter server.
```
jupyter notebook --no-browser --port=8889