12 Commits

Author SHA1 Message Date
Yuge Zhang 5415d850af Migrate release and test workflows for v1 (#548) 2026-08-21 10:31:28 +08:00
Zhiyuan He bd80905120 Docs/add v1 technical report links (#545)
Co-authored-by: GitHub Actions <actions@github.com>
2026-08-19 12:24:04 +08:00
Zhiyuan He 2914f3e2ef Docs/update readme stable docs link (#544)
Co-authored-by: GitHub Actions <actions@github.com>
2026-08-17 15:02:48 +08:00
hzy46 d1d6895782 docs: update legacy branch links 2026-08-13 12:27:35 +08:00
hzy46 b2ca3a50af docs: update documentation for v1.0 2026-08-12 17:10:22 +08:00
Zhiyuan He ea4e49993d Docs, License, Copyright (#48)
* docs: add agl-lite documentation

* docs: refine project messaging

* docs: refresh project overview and examples

* Update README.md

* docs: adopt Agent Lightning v1.0 branding

* Update README.md

* docs & license

* Tune SWE-smith turn penalty

* Reorganize example documentation

* Refine README tagline

* docs: reorganize setup and configuration guides

* docs: clarify shared gateway key

* docs: simplify configuration and async training guides
2026-08-11 18:30:09 +08:00
Zhiyuan He fdf8ec957d Clean legacy example & tests (#37) 2026-06-10 15:12:32 +08:00
hzy46 d95c303a8b feat: refactor agl lite 2026-06-02 03:34:48 +00:00
SiweiPro 29b2e61cf3 feat(verl): add agl-lite PPO trainer integration 2026-05-09 07:23:01 -07:00
yuqyang 7dd323d0b3 update for documents 2026-03-28 00:04:06 -07:00
yuqyang d2a0b57705 docs: split Phase 4 into 4a (mock, CPU) and 4b (real vLLM, GPU)
Phase 4a: E2E validation with mock OpenAI server inside minikube.
  - Mock server (configurable: normal, 503, crash)
  - Example agent + algorithm script
  - Minikube setup + kr8s adapter
  - Full lifecycle tests (batch rollouts, cancel, retry, weight update)

Phase 4b: E2E with real vLLM on 4x A6000 GPUs.
  - vLLM deployment, real inference, weight update protocol
  - Performance baseline (gateway overhead)
  - Prerequisite for Phase 5 (VERL integration)
2026-03-20 20:00:30 -07:00
Yuqing bd565aaa90 docs: add architecture analysis and refactoring plan
- docs/refactor/0_architecture.md: comprehensive mapping of Agent Lightning
  architecture to agl-lite simplifications (core loop, data types, store,
  LLM proxy, execution strategies, VERL integration)
- dev/todo.md: phased refactoring plan (Phase 0-7) with task breakdown
- Covers all 4 key changes: replace litellm, remove OTEL, simplify
  trajectory format, adopt K8s runner
2026-03-18 14:34:13 +08:00