Installation¶

Requirements¶

Note

Python 3.12+ is not currently supported due to TensorFlow availability constraints.

We strongly recommend installing pranaam inside a Python virtual environment. (see venv documentation)

Install pranaam using pip:

pip install pranaam

This installs TensorFlow 2.14.1, which is known to work correctly with the models.

For development work:

pip install -e .[dev]

For testing:

pip install -e .[test]

For documentation building:

pip install -e .[docs]

For all optional dependencies:

pip install -e .[all]

The package requires TensorFlow 2.14.1 with Keras 2.14.0 for model compatibility. If you encounter compatibility issues:

pip install 'pranaam[tensorflow-compat]'

Models are automatically downloaded from Harvard Dataverse (306MB) and cached locally on first use. Ensure you have:

Test your installation:

import pranaam
result = pranaam.pred_rel("Shah Rukh Khan")
print(result)

If successful, you should see a pandas DataFrame with prediction results.

TensorFlow/Keras Compatibility Errors

Error: "Keras 3 only supports V3 .keras files and legacy H5 format files"

Solution: Install with pip install 'pranaam[tensorflow-compat]'

Model Download Issues

Error: Network timeouts or download failures

Solution: Check internet connection, models are large (306MB)

Import Errors

Error: pkg_resources deprecation warnings

Solution: Already fixed in v0.1.0 (uses importlib.resources)

For additional help, please check our GitHub Issues.