BaoWangMath / LaplacianSmoothing-GradientDescent

The code for the paper: https://arxiv.org/abs/1806.06317
24 stars 6 forks source link

Issue while pip installing fft package #1

Open Mrmoore98 opened 5 years ago

Mrmoore98 commented 5 years ago

I can't install this package in win 10(base) C:\Users\dell>pip install C:\Users\dell\Downloads\pytorch_fft-0.15.tar.gz Processing c:\users\dell\downloads\pytorch_fft-0.15.tar.gz Complete output from command python setup.py egg_info: error in pytorch_fft setup command: 'C:\\Users\\dell\\AppData\\Local\\Temp\\pip-req-build-kpn4v7h5\\build.py:ffi' must be of the form 'path/build.py:ffi_variable'

andrehuang commented 5 years ago

This issue should be raised in the pytorch fft thread.

Mrmoore98 commented 5 years ago

and I met another problem when I try to run this program Traceback (most recent call last): File "main.py", line 38, in <module> from utils import * File "C:\Users\dell\Downloads\LaplacianSmoothing-GradientDescent-master\LaplacianSmoothing-GradientDescent-master\Small-BatchSize\utils.py", line 29, in <module> _, term_width = os.popen('stty size', 'r').read().split() ValueError: not enough values to unpack (expected 2, got 0)

andrehuang commented 5 years ago

The code may not be mature (it actually looks like a pre-published version). But the core function is good to use. Try just use the LS_SGD function along with its class Optimizer. Just import the optimizer and use it like you use SGD in your own model.

Mrmoore98 commented 5 years ago

thx

BaoWangMath commented 5 years ago

@Mrmoore98 The updated code does not rely on external fft.