|
| 1 | +# Changelog |
| 2 | + |
| 3 | +## Agent-lightning v0.2.1 (10/30/2025) |
| 4 | + |
| 5 | +Agent-lightning v0.2.1 is a stabilization release for v0.2.0. It introduces several bug fixes and new features, plus a number of unlisted CI improvements. |
| 6 | + |
| 7 | +### Bug fixes |
| 8 | + |
| 9 | +* Fix LiteLLM issues when restarting the proxy multiple times in the same process (#174 #206) |
| 10 | +* Fix LiteLLM model name selection when multiple servers use the same model (#197) |
| 11 | +* Fix store port conflict handling (#227) |
| 12 | + |
| 13 | +### New Features |
| 14 | + |
| 15 | +* Add trainer port option for client-server strategies (#198) |
| 16 | + |
| 17 | +### Documentation |
| 18 | + |
| 19 | +* Add tutorial for launching workers on separate machines (#213) |
| 20 | +* Add link to VERL framework (#210) |
| 21 | +* Add link to vLLM blog (#215) |
| 22 | +* Fix a couple of typos and avoid emacs backup files (#237) |
| 23 | + |
| 24 | +### New Contributors |
| 25 | + |
| 26 | +A warm welcome to our first-time contributors: @scott-vsi, @ddsfda99, @jeis4wpi 🎉 |
| 27 | + |
| 28 | +**Full Changelog**: https://github.com/microsoft/agent-lightning/compare/v0.2.0...v0.2.1 |
| 29 | + |
| 30 | +--- |
| 31 | + |
| 32 | +## Agent-lightning v0.2.0 (10/22/2025) |
| 33 | + |
| 34 | +Agent-Lightning v0.2.0 introduces major framework improvements, new execution strategies, expanded documentation, and enhanced reliability across the agent training and deployment workflow. This release includes **78 pull requests** since v0.1.2. |
| 35 | + |
| 36 | +### Core Enhancements |
| 37 | + |
| 38 | +* **Lightning Store**: Added unified interface and implementation for Agent-lightning's core storage. |
| 39 | +* **Emitter**: Emitting any objects as spans to the store. |
| 40 | +* **Adapter** and **Tracer**: Adapting to OpenAI-like messages, and OpenTelemetry dummy tracer. |
| 41 | +* **LLM Proxy**: Added LLM Proxy as the first-class citizen in Agent-lightning. |
| 42 | +* **Agent Runner**: New version providing a more modular and robust runner design. |
| 43 | +* **Embedded Algorithms**: Algorithms are now embedded directly into trainers for simplicity. |
| 44 | +* **New Execution Strategies**: Introduced *Client-Server* and *Shared Memory* execution models. |
| 45 | +* **Trainer Updates**: Integrated v0.2 interfaces and FastAlgorithm validation. |
| 46 | + |
| 47 | +### Documentation & Examples |
| 48 | + |
| 49 | +* Revamped documentation with new guides for **agent creation**, **training**, **debugging**, and **store concepts**. |
| 50 | +* Improved quickstart tutorials, clarified installation and new deep-dive articles. |
| 51 | +* Added and updated examples: *SQL Agent*, *Calc-X*, *Local SFT*, *Search-R1*, and *APO algorithm*. |
| 52 | + |
| 53 | +### Developer Experience |
| 54 | + |
| 55 | +* Migrated build and CI pipelines to **1ES**, split workflows and aggregate badges for clarity. |
| 56 | +* Adopted **uv** as the dependency manager. |
| 57 | +* Added GPU-based pytest workflows for full test coverage. |
| 58 | +* Enhanced debugging UX, pre-commit configs, and linting (Pyright fixes, import sorting). |
| 59 | + |
| 60 | +### Ecosystem & Integrations |
| 61 | + |
| 62 | +* Added support for agents built with [**Agent-framework**](https://github.com/microsoft/agent-framework). |
| 63 | +* Added new community listings: [*DeepWerewolf*](https://github.com/af-74413592/DeepWerewolf) and [*AgentFlow*](https://agentflow.stanford.edu/). |
| 64 | + |
| 65 | +### New Contributors |
| 66 | + |
| 67 | +A warm welcome to our first-time contributors: |
| 68 | +@hzy46, @lunaqiu, @syeehyn, @linhx1999, @SiyunZhao, and @acured 🎉 |
| 69 | + |
| 70 | +**Full changelog:** [v0.1.2 → v0.2.0](https://github.com/microsoft/agent-lightning/compare/v0.1.2...v0.2.0) |
| 71 | + |
| 72 | +--- |
| 73 | + |
| 74 | +## Agent-lightning v0.1.2 (08/12/2025) |
| 75 | + |
| 76 | +### What's Changed |
| 77 | +* Add basic documentation in https://github.com/microsoft/agent-lightning/pull/33 |
| 78 | +* RAG example by @wizardlancet in https://github.com/microsoft/agent-lightning/pull/21 |
| 79 | + |
| 80 | +### New Contributors |
| 81 | +* @wizardlancet made their first contribution in https://github.com/microsoft/agent-lightning/pull/21 |
| 82 | + |
| 83 | +**Full Changelog**: https://github.com/microsoft/agent-lightning/compare/v0.1.1...v0.1.2 |
| 84 | + |
| 85 | +--- |
| 86 | + |
| 87 | +## Agent-lightning v0.1.1 (08/06/2025) |
| 88 | + |
| 89 | +### What's Changed |
| 90 | +* Disable HTTP tracer tests and bump to 0.1.1 in https://github.com/microsoft/agent-lightning/pull/26 |
| 91 | +* Fix trainer bugs in v0.1 in https://github.com/microsoft/agent-lightning/pull/24 |
| 92 | + |
| 93 | +**Full Changelog**: https://github.com/microsoft/agent-lightning/compare/v0.1...v0.1.1 |
| 94 | + |
| 95 | +--- |
| 96 | + |
| 97 | +## Agent-lightning v0.1.0 (08/04/2025) |
| 98 | + |
| 99 | +The first release of Agent-lightning! |
| 100 | + |
| 101 | +- Turn your agent into an optimizable beast with **ZERO CODE CHANGE** (almost)! 💤 |
| 102 | +- Build with **ANY** agent framework (LangChain, OpenAI Agent SDK, AutoGen, CrewAI, ...); or even WITHOUT agent framework (Python OpenAI). You name it! 🤖 |
| 103 | +- **Selectively** optimize one or more agents in a multi-agent system. 🎯 |
| 104 | +- Embraces Reinforcement Learning, Automatic Prompt Optimization and more **algorithms**. 🤗 |
| 105 | + |
| 106 | +Install via `pip install agentlightning`. |
0 commit comments