pytorch/examples

imagenet benchmark measure elapsed time function issue

Open

#741 opened on Mar 23, 2020

View on GitHub
 (0 comments) (0 reactions) (0 assignees)Python (21,634 stars) (9,429 forks)batch import
good first issue

Description

Hi Pytorch team, I think the measure elaspsed time function is not right. Both trainning and validate function show the progress if i %args.print_freq == 0 in magenet/main.py . currently, the batch_time only measure one loop. I think the batch_time should measure args.print_freq loop elaspsed time.

     # measure elapsed time
     batch_time.update(time.time() - end)
     end = time.time()

     if i % args.print_freq == 0:
         progress.display(i)

Contributor guide

Tech stack
python
Domain
machine learning
Issue type
bug
DifficultyEstimated implementation difficulty for a new contributor, from 1 for very small changes to 5 for expert-level work.
2
Estimated timeA rough time range for an experienced contributor to investigate, implement, test, and prepare a pull request.
1-3 hours
Activity statusHow available the issue appears right now: fresh, active, stale, blocked, or waiting on maintainer input.
stale
ClarityHow clearly the issue explains the expected change, acceptance criteria, and next step.
clear
Prerequisites
basic PythonPyTorch basics
Newbie friendlinessA 1-100 score estimating how approachable this issue is for first-time contributors.
85
Research direction
The issue is in imagenet/main.py. Currently, batch time.update is called every iteration, but it should average over args.print freq iterations. The fix involves moving the measurement to after the print freq loop or accumulating time over multiple iterations. Examine lines around batch time.update and the loop condition to implement the change.