pytorch/examples

storing the best model

Open

#232 opened on Oct 12, 2017

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

Description

https://github.com/pytorch/examples/blob/7d0d413425e2ee64fcd0e0de1b11c5cca1f79f4d/word_language_model/main.py#L171 doesn't look like it is actually saving the best model? It saves only the current model

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.
under 1 hour
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
PythonPyTorch basics
Newbie friendlinessA 1-100 score estimating how approachable this issue is for first-time contributors.
75
Research direction
The issue points to line 171 in word language model/main.py (https://github.com/pytorch/examples/blob/7d0d413425e2ee64fcd0e0de1b11c5cca1f79f4d/word language model/main.py#L171). The current code saves the model at the end of each epoch, but the issue states it saves the current model, not the best one. To fix, add logic to track the best validation loss and save the model only when a new best is achieved. Ensure the model state is properly saved using torch.save. The change should be localized to the training loop around that line. No linked PRs or recent comments indicate active work.