search

KarenUllrich / Tutorial_BayesianCompressionForDL

A tutorial on "Bayesian Compression for Deep Learning" published at NIPS (2017).
MIT License
203 stars 48 forks source link

padding/string/dilation missing in testing the quantized network. #9

Closed aswin-raghavan closed 5 years ago

aswin-raghavan commented 6 years ago

https://github.com/KarenUllrich/Tutorial_BayesianCompressionForDL/blob/a7d3d83410788b2b8ebf76de948af07fbe4922e2/BayesianLayers.py#L350

clouizos commented 5 years ago

Apologies for the late reply; I just noticed the issues here. We only tested the fully connected layer so this fell through the cracks. It is now fixed!

aswin-raghavan commented 5 years ago

We used the solution posted before, but the LeNet model had lower performance than your paper. (Not able to reproduce) See https://arxiv.org/abs/1811.04985

I will try running this again. Aswin

On Dec 11, 2018 12:10 PM, Christos Louizos notifications@github.com wrote:

Closed #9https://na01.safelinks.protection.outlook.com/?url=https%3A%2F%2Fgithub.com%2FKarenUllrich%2FTutorial_BayesianCompressionForDL%2Fissues%2F9&data=01%7C01%7Caswin.raghavan%40sri.com%7Cc2d775af40f945e2fd8508d65f8b7f9b%7C40779d3379c44626b8bf140c4d5e9075%7C1&sdata=lOltFhC1pVlFu9MpJKqhC8Rb4S%2F0XHUF%2BOgHr5e7olY%3D&reserved=0.

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHubhttps://na01.safelinks.protection.outlook.com/?url=https%3A%2F%2Fgithub.com%2FKarenUllrich%2FTutorial_BayesianCompressionForDL%2Fissues%2F9%23event-2019315196&data=01%7C01%7Caswin.raghavan%40sri.com%7Cc2d775af40f945e2fd8508d65f8b7f9b%7C40779d3379c44626b8bf140c4d5e9075%7C1&sdata=D1qroOrPzL4ep07XX%2BbZByNRjArcqKshQehHHa8hkdk%3D&reserved=0, or mute the threadhttps://na01.safelinks.protection.outlook.com/?url=https%3A%2F%2Fgithub.com%2Fnotifications%2Funsubscribe-auth%2FAhK_yqwfkyhCFeeORi5d2u5fZWXv7OMeks5u3-btgaJpZM4WQW4v&data=01%7C01%7Caswin.raghavan%40sri.com%7Cc2d775af40f945e2fd8508d65f8b7f9b%7C40779d3379c44626b8bf140c4d5e9075%7C1&sdata=1Wv5gxs88%2B25ATDUpn5koyrQ3SNetAtliLEwAPbajIo%3D&reserved=0.