https://github.com/osmr/utcte
Examples for Universal Training Control Tools
https://github.com/osmr/utcte
Last synced: about 2 months ago
JSON representation
Examples for Universal Training Control Tools
- Host: GitHub
- URL: https://github.com/osmr/utcte
- Owner: osmr
- License: mit
- Created: 2017-08-04T12:46:57.000Z (almost 9 years ago)
- Default Branch: master
- Last Pushed: 2017-12-19T08:26:47.000Z (over 8 years ago)
- Last Synced: 2025-12-27T13:35:14.950Z (6 months ago)
- Language: Python
- Size: 28.3 KB
- Stars: 0
- Watchers: 0
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
- License: LICENSE
Awesome Lists containing this project
README
# Examples for Universal Training Control Tools
A set examples of using UTCT for various classification/regression tasks.
## Remarks
All scripts must be runned from the root of the repository, e.g. for MXNet-version of MNIST training:
```
python mnist/MXNet/train.py
```
## Visualization of learning curves
Any training script usually creates a project subdirectory in the working director with all
working files. One of them is a `*_score.log` file with tabulated learning curves. You can visualize
this curves by using R-script `utct/common/observe.R`. This script starts a web-server and has two
arguments:
- the path to `score_ref.csv` file (this file is in the working directory),
- port for web-server (the defalut value is 6006).
After starting this script you can see curve plots in your web-browser by the address
`http://localhost:6006`.