- Model loading on server startup - dependent on language chosen as CLI argument
- Model fine tuning automatically and manually. Should be an argument
- Ability to choose between GPU and CPU (maybe even TPU)
- Anonymized documents / sentences returned with markings - including JSON format and the modified file
- Path to see loaded model and the summary
- Path to see version running
- Calculate F1 Score
- Connect to Azure to allocate a GPU on a VM to train. Should also be able to automatically delete the VM again
- Connect to Google Cloud to allocate a GPU on a VM to train. Should also be able to automatically delete the VM again
- Connect to AWS to allocate a GPU on a VM to train. Should also be able to automatically delete the VM again
- Store in a database(?)
- [ ]
- [ ]
- [ ]
This repository has been archived by the owner on Jan 29, 2024. It is now read-only.