Package `phiml`

PhiML.

Project homepage: https://github.com/tum-pbs/PhiML

Documentation overview: https://tum-pbs.github.io/PhiML

PyPI: https://pypi.org/project/phiml/

This package can be used as convenience import, i.e. from phiml import *.

Sub-modules

phiml.backend: Low-level library wrappers for delegating vector operations.
phiml.dataclasses: PhiML makes it easy to work with custom classes. Any class decorated with @dataclass can be used with phiml.math functions, such as shape(), …
phiml.latent: Dimensionality reduction and latent space models …
phiml.math: Vectorized operations, tensors with named dimensions …
phiml.nn: Unified neural network library. Includes …
phiml.os: Adding vectorization to phiml.os functions …
phiml.parallel: This package contains utilities for running PhiML code in parallel. To use it, you need to declare the code to be parallelized as properties of a …

Functions

def set_logging_level(level='debug')

Expand source code

def set_logging_level(level='debug'):
    """
    Sets the logging level for Φ-ML functions.

    Args:
        level: Logging level, one of `'critical', 'fatal', 'error', 'warning', 'info', 'debug'`
    """
    from .backend import ML_LOGGER
    ML_LOGGER.setLevel(level.upper())

Sets the logging level for Φ-ML functions.

Args

level: Logging level, one of 'critical', 'fatal', 'error', 'warning', 'info', 'debug'

def verify()

Expand source code

def verify():
    """
    Checks your configuration for potential problems and prints a summary.

    To run verify without importing `phiml`, run the script `tests/verify.py` included in the source distribution.
    """
    import sys
    from ._troubleshoot import assert_minimal_config, troubleshoot
    try:
        assert_minimal_config()
    except AssertionError as fail_err:
        print("\n".join(fail_err.args), file=sys.stderr)
        return
    print(troubleshoot())

Checks your configuration for potential problems and prints a summary.

To run verify without importing phiml, run the script tests/verify.py included in the source distribution.