|
|
--- |
|
|
license: mit |
|
|
--- |
|
|
## Sentiment Analysis API |
|
|
|
|
|
This is the deployment part of the project. |
|
|
|
|
|
### Training: |
|
|
Run the Google Colab notebook (Runtime = "GPU") (https://colab.research.google.com/drive/1EuF5FDl1X8VnuOO5RxzmM0c9TbtQrVm9?usp=sharing) |
|
|
|
|
|
### Fine Tuning |
|
|
1) Increasing #epochs |
|
|
2) Increasing BATCH_SIZE to 32 |
|
|
3) Changing Adam optimizer rate |
|
|
|
|
|
### Usage |
|
|
1) Clone te repository |
|
|
2) Set up the conda environment with requirements.txt |
|
|
3) In terminal, run the command- uvivorn sentiment_analyzer.api:app --reload |
