Skip to content
Snippets Groups Projects
Commit 77fab8ab authored by Yifan Zhao's avatar Yifan Zhao
Browse files

Updated setup.py and removed outdated TODO

parent a2b3e99a
No related branches found
No related tags found
No related merge requests found
# What kind of models we should use
ResNet-50 -> as a start?
BERT
Mask R-CNN
# Questions from Sudipta
Graph optimization: operator fusions, data layout transformations
Vectorization/tensorization
Automatic scheduling
Handling dynamic control flow in model graph
Automatic differentiation
# Operators used in the BERT model available in the ONNX model zoo, organized by type and number of occurences
https://github.com/onnx/models/tree/master/text/machine_comprehension/bert-squad
(Opset 10 versions)
DATA MANIPULATION
Unsqueeze 191
Reshape 71
Cast 70
Transpose 62
Concat 56
Identity 28
Squeeze 7
Slice
Shape 5
ConstantOfShape 1
Gather 1
OneHot 1
Split 1
COMPUTATION
Mul 186
Add 185
MatMul 98
Sub 62
ReduceMean 50
Sqrt 25
Reciprocal 25
Softmax 12
Pow 12
Tanh 12
......@@ -4,11 +4,14 @@ setup(
name="torch2hpvm",
version="1.0",
description="PyTorch frontend for HPVM",
author="Yuanjing Shi, Yifan Zhao",
author_email="ys26@illinois.edu, yifanz16@illinois.edu",
author="Yifan Zhao, Yuanjing Shi",
author_email="yifanz16@illinois.edu, ys26@illinois.edu",
packages=["torch2hpvm"],
install_requires=[
"jinja2>=2.11", "networkx>=2.5", "onnx>=1.8.0", "torch", "onnx-simplifier>=0.2.27"
"jinja2>=2.11",
"networkx>=2.5",
"onnx>=1.8.0",
"torch>=1.5",
"onnx-simplifier>=0.2.27",
],
entry_points={"console_scripts": ["torch2hpvm=torch2hpvm:main"]},
)
0% Loading or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment