Repository navigation
Closed
Description
Overall Goal
- Improve open-source project infrastructure
- Fix existing bugs and stabilize service/API
- Add quantifiable regression tests
- Support new feature integration, such as KV cache, streaming inference, and distributed inference
🧠 New Model Integrations
- SwiftVR feasibility study / initial integration feat(swiftvr): add real-time restoration pipeline #29
- Lingbot-videofeat(lingbot-video): add native Dense/MoE inference and four-GPU runtime #15
- Lingbot-world v2 [feat] Add LingBot-World v2 causal-fast streaming support #11
- Cosmos3 @Uxito-Ada
- Abot-World @Kaimary [FEATURE] Add ABot-World-0-5B-LF single-GPU interactive pipeline #26
- Lingbot-VLAv2 @HappyDog060713 feat(vla): integrate LingBot VLA v2 structured serving and benchmarks #28
- Mage
- TaoMate
- MiniMax-H3 [FEAT] Add MiniMax H3 joint audio-video generation #23
- Wan-Animate-2
- LTX-2.5 [FEATURE] Add LTX-2.5 distilled T2V/I2V inference #33
- QwenImage 2.1[FEATURE] Add native Qwen-Image 2.1 support #51
New Feature Support
- Chunk Batching @Kaimary
- Multi-instance KV cache with fp8 or fp4 quant @Uxito-Ada
- Disaggregated Inference
- Multi-instance server support
- world model stream server with livekit [FEAT] Unify WebRTC streaming on LiveKit and remove aiortc #19 feat(stream): time-slice retained LingBot sessions #21
- ROS2 adapter
⚡ Performance & Kernels
- NVF4 for world model @Uxito-Ada
- SVDQuant support @Uxito-Ada
- Sol-Attn feat(attention): integrate Sol-Attn for Wan2.1 #24
- optimize Ulysses SP alltoall transport [PERF] Add CUDA IPC Ulysses overlap and lossless LingBot streaming #27
- NPU support [FEATURE] Run Wan2.2 pipelines on Ascend NPU #41
🔗 Distributed Inference
- Ulysses SP for LingBot-World [REFACTOR] Refactor LingBot stateful generation and session lifecycle #8
🌐 API & Service
- Refactor service module and fix bugs [REFACTOR] Centralize local artifact and task route handling #6
- Stabilize task lifecycle and OpenAI-compatible API
🧪 Testing
- Add quantifiable regression test suite
- Add pipeline smoke tests
- Add service/API regression tests
- Add packaging/install validation tests
📚 Documentation & Tooling
- Support online document website
- Improve contributor docs for testing, service, and examples
Deployment
- Make pip install package available
Progress Tracking
Status:
| Area | Progress | Key PRs/Issues |
|---|---|---|
| New Models | ⬜ 7/11 | SwiftVR, Lingbot-video, Lingbot-world-v2, Cosmos3, LingBot VLA .etc |
| New Features | ⬜ 0/3 | KV streaming, disaggregated inference, multi-instance server |
| Performance | ⬜ 0/2 | NVF4, SVDQuant |
| Distributed | 🟩 1/1 | LingBot-World Ulysses SP |
| API & Service | ⬜ 0/2 | service refactor, API/task lifecycle stability |
| Testing | ⬜ 0/4 | regression suite, pipeline smoke, service/API, install validation |
| Docs & Tooling | 🟩 1/2 | online docs done; contributor docs pending |
| Deployment | ⬜ 0/1 | pip install package |
Metadata
Metadata
Assignees
Labels
No labels
Activity
ActivePeter commentedon Jun 13, 2026
Is there any difference between SGLang’s KV cache implementation and TeleFuser’s? Fluxon has been working on SGLang KV cache integration recently.
[-][ROADMAP] TeleFuser Q2 2026[/-][+][ROADMAP] TeleFuser Q3 2026[/+]