LSTM Finance Predictor

Following are the instructions on how to run the project, using docker to make it easier.

Prerequisites:

• Docker installed and running

Build (CPU — default)

docker build -t lstm-finance-predictor:latest .

Build (GPU)

docker build --build-arg TORCH=gpu -t lstm-finance-predictor:gpu .

Notes:

• The project's Dockerfile runs dependency sync, dataset gathering, and training during image build. The build is expected to take some minutes to finish as it will train the model, that was done to make all the other steps easier for the common user.
• The container exposes port 8080 and starts a FastAPI app via uv run uvicorn src.server.config:app.

Run (foreground)

docker run -p 8080:8080 lstm-finance-predictor:latest

Run (detached, with name)

docker run -d --name lstm-finance-predictor -p 8080:8080 lstm-finance-predictor:latest

Access:

• Open http://localhost:8080

Stop & remove (if run with --name lstm-finance-predictor):

docker stop lstm-finance-predictor && docker rm lstm-finance-predictor

Scripts

All the scripts have a simple --help parameter that you can use to see all allowed parameters.