https://github.com/pandede/metricmeter
MetricMeter: Recording the metric trend for deep learning model
https://github.com/pandede/metricmeter
Last synced: 3 months ago
JSON representation
MetricMeter: Recording the metric trend for deep learning model
- Host: GitHub
- URL: https://github.com/pandede/metricmeter
- Owner: Pandede
- Created: 2022-06-08T03:41:05.000Z (about 4 years ago)
- Default Branch: master
- Last Pushed: 2022-06-11T16:49:39.000Z (about 4 years ago)
- Last Synced: 2025-07-30T12:00:54.542Z (11 months ago)
- Language: Python
- Homepage:
- Size: 6.84 KB
- Stars: 0
- Watchers: 1
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
Awesome Lists containing this project
README
# MetricMeter
A package for which handling the trend of metrics while training a deep learning model, such as the trend of loss, accuracy, f1-score in each epoch or iteration etc.
## Version
*v0.0.1*
## Modules
### **Tracker**
`Tracker` is the core module which used for recording and summarizing the time series, such as running mean `AverageTracker` or running variance `VarTracker`.
For example, we can `append` the loss of each batch and `get` the running mean of loss. Concluded, the aim of `Tracker` is tracking the running metric in a single iteration.
### **Meter**
`Tracker` handles the running metric of each iteration, while `Meter` collects the final metric of `Tracker` in each epoch.
For example, after all iterations is over, `Meter` saves the mean of loss in this iteration, resets `Tracker` and goes forward to the next epoch.
## Note
:warning: The repository is heavily under constructing, thus welcome to provide any inspiration or thought, even make contributions!