xpflow

0.8 verified Sat May 09 auth: no python

A lightweight Python library for representing and managing machine learning experiments with class-based definitions. Current version 0.8, updated infrequently.

pip install xpflow

Common errors

error AttributeError: 'MyExperiment' object has no attribute 'run' ↓

cause Experiment subclass missing a 'run' method, or method signature mismatch.

fix

Define a 'run(self)' method in your experiment class.

error TypeError: run() missing 1 required positional argument: 'xp' ↓

cause run() called without an experiment instance.

fix

Call run(experiment_instance) instead of run().

Warnings

gotcha Experiment class attributes must be type-annotated; untyped attributes may be ignored or cause errors. ↓

fix Always annotate fields with types (e.g., int, str, float).

gotcha The run() function expects an instance of Experiment, not a class. Passing the class itself will raise an AttributeError. ↓

fix Instantiate the experiment class before passing to run: run(MyExperiment())

deprecated The 'sequential' decorator introduced in v0.3 was removed in v0.8. Code using it will break. ↓

fix Remove @sequential decorator; use class-based inheritance instead.

Imports

Experiment
```
from xpflow import Experiment
```
Xp
```
from xpflow import Xp
```
run
```
from xpflow import run
```

Quickstart

Define an experiment as a class inheriting from Experiment, then call run() to execute.

from xpflow import Experiment, Xp, run

class MyExperiment(Experiment):
    hyperparam1: int = 10
    hyperparam2: str = "default"

    def run(self):
        return self.hyperparam1 * len(self.hyperparam2)

if __name__ == "__main__":
    xp = MyExperiment()
    result = run(xp)
    print(result)