agent-learning 0.4.1__tar.gz

agent_learning-0.4.1/LICENSE

@@ -0,0 +1,21 @@
+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.

agent_learning-0.4.1/PKG-INFO

@@ -0,0 +1,82 @@
+Metadata-Version: 2.4
+Name: agent-learning
+Version: 0.4.1
+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/agent-learning
+Project-URL: Repository, https://github.com/microsoft/agent-learning
+Project-URL: Issues, https://github.com/microsoft/agent-learning/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
+
+# agent-learning
+
+Native reinforcement learning SDK for AI agents. An in-process learner optimizes a small, interpretable policy over discrete agent choices (understand intent and complete task by choosing the right outcome).
+
+## 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., "take action A", "take action B", "take action C"). It lives in Python and updates in milliseconds.
+
+2. Each episode is **evaluated** by three 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 local compute and persist through a pluggable store — in-memory or local files by default, with Azure Cosmos DB optional.
+
+Every episode, reward, run, and deployment is captured by the configured store — in-memory or local files by default, or Azure Cosmos DB — giving you a complete lineage and audit trail of how the policy evolved over time.

agent_learning-0.4.1/PYPI.md

@@ -0,0 +1,15 @@
+# agent-learning
+
+Native reinforcement learning SDK for AI agents. An in-process learner optimizes a small, interpretable policy over discrete agent choices (understand intent and complete task by choosing the right outcome).
+
+## 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., "take action A", "take action B", "take action C"). It lives in Python and updates in milliseconds.
+
+2. Each episode is **evaluated** by three 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 local compute and persist through a pluggable store — in-memory or local files by default, with Azure Cosmos DB optional.
+
+Every episode, reward, run, and deployment is captured by the configured store — in-memory or local files by default, or Azure Cosmos DB — giving you a complete lineage and audit trail of how the policy evolved over time.

agent_learning-0.4.1/README.md

@@ -0,0 +1,87 @@
+# agent-learning
+
+Native reinforcement learning SDK for AI agents. An in-process
+learner optimizes a small, interpretable policy over discrete agent choices (e.g., "take action A", "take action B", "take action C") using AI Evaluation scores as the reward
+signal.
+
+<p align="center">
+   <img src="images/agent-learning-loop.svg" alt="Animated loop: Policy chooses an action, Score evaluates the episode, and Learner updates the policy" width="960" style="max-width:100%; height:auto;" />
+</p>
+
+## 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., "take action A", "take action B", "take action C"). It lives in Python and updates in milliseconds.
+
+   <img src="images/0f85e08d0c47cd01.png" alt="Policy selects one of N discrete actions" width="360" style="max-width:100%; height:auto;" />
+
+2. Each episode is **evaluated** by three AI Evaluation
+   evaluators — `IntentResolutionEvaluator`, `TaskAdherenceEvaluator`,
+   and `TaskCompletionEvaluator` — whose scores are combined into a single scalar reward.
+
+   <img src="images/246d112f995b785a.png" alt="Three evaluator scores feed a single scalar reward" width="360" style="max-width:100%; height:auto;" />
+
+3. A **Reinforce-with-baseline** learner updates the policy logits
+   directly from stored episodes. Updates are tiny gradient steps
+   that run on local compute and persist through a pluggable store — in-memory
+   or local files by default, with Azure Cosmos DB optional.
+
+   <img src="images/cc970c453583c982.png" alt="Policy quality improves with every batch of episodes" width="360" style="max-width:100%; height:auto;" />
+
+Every episode, reward, run, and deployment is captured by the
+configured store — in-memory or local files by default, or Azure Cosmos DB —
+giving you a complete lineage and audit trail of how the policy
+evolved over time.
+
+## Install
+
+### Windows CLI
+
+For a Python-independent installation, download `agent-learn.exe` or the
+standalone installer from the
+[latest GitHub release](https://github.com/microsoft/agent-learning/releases/latest).
+The installer can add its installation directory to your user `PATH`, so
+`agent-learn` works from PowerShell or Command Prompt without Python or `pip`.
+
+```powershell
+agent-learn.exe --help
+```
+
+### Python SDK
+
+Released versions are published to PyPI:
+<https://pypi.org/project/agent-learning/>.
+
+```powershell
+py -m pip install agent-learning
+agent-learn.exe --help
+```
+
+`pip` installs `agent-learn.exe` into the active Python environment's
+`Scripts` directory.
+
+## Usage
+
+The `agent-learn` CLI provides the current task-learning-loop operations:
+
+```text
+agent-learn list
+agent-learn tasks-list <agent_id>
+agent-learn task-episodes-count <agent_id> [--task-id <task_id>]
+agent-learn task-episodes-list <agent_id> [--task-id <task_id>] [--limit <1-500>] [--include-incomplete]
+agent-learn task-policy-init --agent-id <agent_id> --task-id <task_id> --actions ./actions.json
+agent-learn task-episode-register --agent-id <agent_id> --task-id <task_id> --episode ./episode.json
+agent-learn score --agent-id <agent_id> [--task-id <task_id>] [--limit <1-500>]
+agent-learn train --agent-id <agent_id> [--task-id <task_id>] [--limit <1-500>] [--start-date <date>] [--end-date <date>] [--skip-scoring]
+agent-learn task-policy --agent-id <agent_id> --task-id <task_id>
+```
+
+The subprocess-level functional workflow uses an isolated local store. Run the
+interactive capture scenario first, followed by the offline batch update:
+
+```powershell
+python tests/functional_cli_interactive.py
+python tests/functional_cli_batch.py
+```

agent_learning-0.4.1/pyproject.toml

@@ -0,0 +1,79 @@
+[build-system]
+requires = ["setuptools>=68", "wheel"]
+build-backend = "setuptools.build_meta"
+
+[project]
+name = "agent-learning"
+version = "0.4.1"
+description = "Native, in-process reinforcement learning SDK for AI agents — in-memory and local-file storage by default."
+readme = "PYPI.md"
+requires-python = ">=3.10"
+license = { file = "LICENSE" }
+authors = [{ name = "Chris Tava" }]
+keywords = ["reinforcement-learning", "ai-agents", "azure", "evaluation", "rlhf"]
+classifiers = [
+    "Development Status :: 4 - Beta",
+    "Intended Audience :: Developers",
+    "License :: OSI Approved :: MIT License",
+    "Programming Language :: Python :: 3",
+    "Programming Language :: Python :: 3.10",
+    "Programming Language :: Python :: 3.11",
+    "Programming Language :: Python :: 3.12",
+    "Topic :: Scientific/Engineering :: Artificial Intelligence",
+]
+
+dependencies = [
+    "numpy>=1.24",
+    "azure-identity>=1.15",
+    "azure-ai-evaluation>=1.0.0",
+    "pydantic>=2.0",
+]
+
+[project.optional-dependencies]
+dev = [
+    "pytest>=7.4",
+    "pytest-asyncio>=0.21",
+    "pytest-cov>=4.1",
+    "ruff>=0.1",
+    "mypy>=1.6",
+]
+cosmos = [
+    "azure-cosmos>=4.5",
+]
+nlp = [
+    "scikit-learn>=1.3",
+    "scipy>=1.10",
+    "joblib>=1.3",
+]
+examples = [
+    "pyyaml>=6.0",
+]
+slm = [
+    "onnxruntime-genai>=0.5",
+]
+llm = [
+    "azure-ai-evaluation>=1.0.0",
+    "azure-identity>=1.15",
+]
+
+[project.scripts]
+agent-learn = "agent_learning.cli:main"
+
+[project.urls]
+Homepage = "https://github.com/microsoft/agent-learning"
+Repository = "https://github.com/microsoft/agent-learning"
+Issues = "https://github.com/microsoft/agent-learning/issues"
+
+[tool.setuptools.packages.find]
+where = ["src"]
+
+[tool.setuptools.package-data]
+agent_learning = ["py.typed"]
+
+[tool.ruff]
+line-length = 110
+target-version = "py310"
+
+[tool.pytest.ini_options]
+testpaths = ["tests"]
+asyncio_mode = "auto"

agent_learning-0.4.1/setup.cfg

@@ -0,0 +1,4 @@
+[egg_info]
+tag_build = 
+tag_date = 0
+

agent_learning-0.4.1/src/agent_learning/__init__.py

@@ -0,0 +1,160 @@
+"""Native reinforcement learning SDK for AI agents.
+
+Replaces the agent-lightning LLM fine-tuning loop with a fully
+native, in-process learner. The SDK is organised into five layers:
+
+- ``agent_learning.types``    - durable record types (``Episode``,
+  ``Reward``, ``PolicySnapshot``, ...).
+- ``agent_learning.storage``  - pluggable persistence (Cosmos DB,
+  local file system, and in-memory).
+- ``agent_learning.metrics``  - score-based metrics that wrap the
+  Azure AI Evaluation evaluators for Intent Resolution, Task
+  Adherence, and Task Completion.
+- ``agent_learning.rewards``  - reward shaping + persistence.
+- ``agent_learning.policy``   - discrete softmax bandit policy.
+- ``agent_learning.learners`` - REINFORCE-with-baseline learner.
+- ``agent_learning.training`` - end-to-end :class:`LearningRunner`.
+
+Quick start::
+
+    from agent_learning import (
+        Action,
+        EpisodeCapture,
+        LearningRunner,
+        SoftmaxPolicy,
+    )
+
+    actions = [Action(id="prompt_A"), Action(id="prompt_B")]
+    policy = SoftmaxPolicy.from_actions(actions, agent_id="dq", task_id="sales-summary")
+
+    # Capture
+    capture = EpisodeCapture()
+    decision = policy.choose()
+    ctx = capture.start(
+        "Tell me my Q3 sales summary",
+        task_id="sales-summary",
+        intent_summary="Summarize Q3 sales",
+        action_type="chat",
+        action_name=decision.action.id,
+        expected_outcome="An accurate Q3 sales summary",
+        policy_id=policy.snapshot().id,
+        policy_version=policy.snapshot().version,
+        action_id=decision.action.id,
+        action_logprob=decision.logprob,
+    )
+    # ... agent runs, records tool calls, produces output ...
+    episode = capture.end(
+        ctx,
+        assistant_output="...",
+        execution_status="completed",
+        result_summary="Returned the sales summary",
+    )
+
+    # Train
+    runner = LearningRunner(policy=policy)
+    run = runner.run_offline_batch("dq", task_id="sales-summary", episode_limit=200)
+"""
+
+from ._version import __version__
+from .capture import CaptureContext, EpisodeCapture, get_capture
+from .classifiers import (
+    AdherenceScorer,
+    Classifier,
+    ClassifierResult,
+    CompletionScorer,
+    IntentScorer,
+    RouterClassifier,
+)
+from .config import (
+    CaptureConfig,
+    CosmosConfig,
+    ScoreConfig,
+    LearnerConfig,
+    ShapingConfig,
+)
+from .learners import Learner, LearnerResult, ReinforceLearner
+from .metrics import (
+    IntentResolutionMetric,
+    MetricEvaluator,
+    MetricRequest,
+    TaskAdherenceMetric,
+    TaskCompletionMetric,
+    default_metrics,
+    evaluate_all,
+)
+from .policy import ContextualSoftmaxPolicy, Policy, SoftmaxPolicy
+from .rewards import RewardShaper, RewardWriter, shape_episode_reward
+from .storage import (
+    CosmosStore,
+    InMemoryStore,
+    LearningStore,
+    LocalFileStore,
+    get_default_store,
+)
+from .training import LearningRunner
+from .types import (
+    Action,
+    AgentSummary,
+    AgentTaskSummary,
+    Episode,
+    MetricName,
+    MetricResult,
+    PolicySnapshot,
+    Reward,
+    RewardSource,
+    ToolCall,
+    TrainingRun,
+    TrainingStatus,
+)
+
+__all__ = [
+    "Action",
+    "AdherenceScorer",
+    "AgentSummary",
+    "AgentTaskSummary",
+    "CaptureConfig",
+    "CaptureContext",
+    "Classifier",
+    "ClassifierResult",
+    "CompletionScorer",
+    "ContextualSoftmaxPolicy",
+    "CosmosConfig",
+    "CosmosStore",
+    "Episode",
+    "EpisodeCapture",
+    "InMemoryStore",
+    "IntentScorer",
+    "IntentResolutionMetric",
+    "ScoreConfig",
+    "Learner",
+    "LearnerConfig",
+    "LearnerResult",
+    "LearningRunner",
+    "LearningStore",
+    "LocalFileStore",
+    "MetricEvaluator",
+    "MetricName",
+    "MetricRequest",
+    "MetricResult",
+    "Policy",
+    "PolicySnapshot",
+    "ReinforceLearner",
+    "Reward",
+    "RewardShaper",
+    "RewardSource",
+    "RewardWriter",
+    "RouterClassifier",
+    "ShapingConfig",
+    "SoftmaxPolicy",
+    "TaskAdherenceMetric",
+    "TaskCompletionMetric",
+    "ToolCall",
+    "TrainingRun",
+    "TrainingStatus",
+    "__version__",
+    "default_metrics",
+    "evaluate_all",
+    "get_capture",
+    "get_default_store",
+    "shape_episode_reward",
+]

agent_learning-0.4.1/src/agent_learning/_version.py

@@ -0,0 +1,3 @@
+"""Package version."""
+
+__version__ = "0.4.1"