Trainy Policy Nightly

raw JSON →
0.1.0.dev20260430120117 verified Fri May 01 auth: no python

Nightly builds of Trainy's Skypilot policy package. Provides policy management for SkyPilot-based training workloads on cloud infrastructure. Current version is 0.1.0.dev, Python 3.9+ only, updated nightly.

pip install trainy-policy-nightly
error ModuleNotFoundError: No module named 'trainy_policy_nightly'
cause The package name includes hyphens, so the import path is not the same as the package name.
fix
Use 'from trainy import ...' instead of trying to import from 'trainy_policy_nightly'.
error ImportError: cannot import name 'Policy' from 'trainy'
cause The installed version of trainy may be too old or the nightly package did not update the installed trainy correctly.
fix
Ensure you have installed the latest nightly: 'pip install --upgrade trainy-policy-nightly'. Also verify that only one version of trainy is installed.
error ValueError: AUTH_KEY environment variable not set
cause The Policy object requires authentication key via environment variable AUTH_KEY.
fix
Set the environment variable before running: export AUTH_KEY='your-key-here'
breaking This is a pre-release nightly version (0.1.0.dev). APIs are unstable and may change without notice. Do not use in production.
fix Pin to a specific stable release once available, or use trainy (stable) instead.
gotcha The package name (trainy-policy-nightly) is not a Python import name. It installs the 'trainy' package which may already be present in your environment. Conflicts can occur if a different version of trainy is already installed.
fix Install this package in a virtual environment to isolate from other versions of trainy.
deprecated Nightly builds may be replaced by stable releases. Check for a non-dev version of trainy-policy (or trainy) before relying on this package long-term.
fix Monitor https://pypi.org/project/trainy-policy/ for stable releases.

Basic usage: import Policy from trainy (not this nightly package) and authenticate via environment variable.

import os
from trainy import Policy

# Ensure AUTH_KEY is set
auth_key = os.environ.get('AUTH_KEY', '')
policy = Policy(auth_key=auth_key)
print(policy.list())