https://github.com/lucidrains/panoptic-transformer
Another attempt at a long-context / efficient transformer by me
https://github.com/lucidrains/panoptic-transformer
artificial-intelligence attention-mechanism deep-learning
Last synced: about 1 year ago
JSON representation
Another attempt at a long-context / efficient transformer by me
- Host: GitHub
- URL: https://github.com/lucidrains/panoptic-transformer
- Owner: lucidrains
- License: mit
- Created: 2021-11-22T19:44:28.000Z (over 4 years ago)
- Default Branch: main
- Last Pushed: 2022-04-11T15:39:01.000Z (about 4 years ago)
- Last Synced: 2025-03-30T17:46:14.412Z (over 1 year ago)
- Topics: artificial-intelligence, attention-mechanism, deep-learning
- Language: Python
- Homepage:
- Size: 47.9 KB
- Stars: 37
- Watchers: 7
- Forks: 2
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
- License: LICENSE
Awesome Lists containing this project
README
## Panoptic Transformer (wip)
Another attempt at a long-context / efficient transformer by me. This approach will completely generalize all multi-scale approaches of the past. I will be attempting the Pathfinder-X task, which so far has not been beat by a transformer.
Update: on track to solving path-x with transformers
## Training
The script will generate 25000 training samples (in paper they used 100k; you can change it to this number if you are willing to wait).
```bash
$ ./setup.sh
```