This repository contains the necessary files to deploy a sentiment analysis model based on the BERT architecture on the Akash Network. The model is capable of classifying text into five sentiment categories: very negative, negative, neutral, positive, and very positive.
The model used is the nlptown/bert-base-multilingual-uncased-sentiment model from Hugging Face. This model is capable of understanding and generating text in multiple languages, making it versatile for various use cases.
app.py: This is the main application file. It uses Flask to create a web application that takes user input, passes it to the model, and returns the sentiment prediction.Dockerfile: This file contains the instructions to build the Docker image for the application.requirements.txt: This file lists the Python libraries required by the application.deploy.yaml: This is the SDL (Stack Definition Language) file used for deploying the application on the Akash Network.index.html: This file contains the HTML code for the application's user interface.
To deploy the application on the Akash Network, you need to have an Akash account with sufficient AKT balance. Follow the steps below:
- Clone this repository.
- Build the Docker image and push it to a Docker registry.
- Update the
deploy.yamlfile with the correct Docker image path. - Use the Akash CLI or Akash Console to deploy the
deploy.yamlfile.
Please refer to the Akash Documentation for detailed instructions on deploying applications.
Once the application is deployed, you can test it by sending a POST request to the /predict endpoint with a JSON payload containing the text to be analyzed. For example:
curl -X POST -H "Content-Type: application/json" -d '{"text":"I love this product!"}' http://<your-akash-deployment-url>/predictThe application will return a JSON response with the sentiment prediction:
{"sentiment": "positive"}