Metadata-Version: 2.4
Name: azure-agents-learning-sdk
Version: 0.3.1
Summary: Native, in-process reinforcement learning SDK for AI agents — in-memory and local-file storage by default.
Author: Chris Tava
License: MIT License
        
        Copyright (c) 2026 Microsoft
        
        Permission is hereby granted, free of charge, to any person obtaining a copy
        of this software and associated documentation files (the "Software"), to deal
        in the Software without restriction, including without limitation the rights
        to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
        copies of the Software, and to permit persons to whom the Software is
        furnished to do so, subject to the following conditions:
        
        The above copyright notice and this permission notice shall be included in all
        copies or substantial portions of the Software.
        
        THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
        IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
        FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
        AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
        LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
        OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
        SOFTWARE.
        
Project-URL: Homepage, https://github.com/microsoft/azure-agents-learning-sdk
Project-URL: Repository, https://github.com/microsoft/azure-agents-learning-sdk
Project-URL: Issues, https://github.com/microsoft/azure-agents-learning-sdk/issues
Keywords: reinforcement-learning,ai-agents,azure,evaluation,rlhf
Classifier: Development Status :: 4 - Beta
Classifier: Intended Audience :: Developers
Classifier: License :: OSI Approved :: MIT License
Classifier: Programming Language :: Python :: 3
Classifier: Programming Language :: Python :: 3.10
Classifier: Programming Language :: Python :: 3.11
Classifier: Programming Language :: Python :: 3.12
Classifier: Topic :: Scientific/Engineering :: Artificial Intelligence
Requires-Python: >=3.10
Description-Content-Type: text/markdown
License-File: LICENSE
Requires-Dist: numpy>=1.24
Requires-Dist: azure-identity>=1.15
Requires-Dist: azure-ai-evaluation>=1.0.0
Requires-Dist: pydantic>=2.0
Provides-Extra: dev
Requires-Dist: pytest>=7.4; extra == "dev"
Requires-Dist: pytest-asyncio>=0.21; extra == "dev"
Requires-Dist: pytest-cov>=4.1; extra == "dev"
Requires-Dist: ruff>=0.1; extra == "dev"
Requires-Dist: mypy>=1.6; extra == "dev"
Provides-Extra: cosmos
Requires-Dist: azure-cosmos>=4.5; extra == "cosmos"
Provides-Extra: nlp
Requires-Dist: scikit-learn>=1.3; extra == "nlp"
Requires-Dist: scipy>=1.10; extra == "nlp"
Requires-Dist: joblib>=1.3; extra == "nlp"
Provides-Extra: examples
Requires-Dist: pyyaml>=6.0; extra == "examples"
Provides-Extra: slm
Requires-Dist: onnxruntime-genai>=0.5; extra == "slm"
Provides-Extra: llm
Requires-Dist: azure-ai-evaluation>=1.0.0; extra == "llm"
Requires-Dist: azure-identity>=1.15; extra == "llm"
Dynamic: license-file

# azure-agents-learning-sdk

Native reinforcement learning SDK for AI agents. An in-process learner optimizes a small, interpretable policy over discrete agent configuration choices (prompt variants, retrieval-k, tool selection strategies, …).

## How it works

The SDK improves agents without LLM weight fine-tuning. There are no GPU fine-tune jobs and no opaque update cycles — just three pieces that run in your existing Python process:

1. The **policy** is a softmax distribution over `N` discrete actions (e.g., "use prompt template A", "use template B"). It lives in Python and updates in milliseconds.

2. Each episode is **judged** by three Azure AI Evaluation evaluators — `IntentResolutionEvaluator`, `TaskAdherenceEvaluator`, and `TaskCompletionEvaluator` — whose scores are combined into a single scalar reward.

3. A **REINFORCE-with-baseline** learner updates the policy logits directly from logged episodes. Updates are tiny gradient steps that run on CPU and persist through a pluggable store — in-memory or local files by default, with Cosmos DB optional.

Every episode, reward, run, and deployment is captured by the configured store — in-memory or local files by default, or Cosmos DB — giving you a complete lineage and audit trail of how the policy evolved over time.
