Tao Feng (Attending NeurIPS 2025 in San Diego)

🎨 Introducing ComfyUI Interface for LLMRouter - Build your entire LLM routing pipeline visually! We're excited to release a powerful visual interface that transforms how you work with LLMRouter. No more YAML configs or terminal scripts - just drag, drop, and connect nodes. 🔗 Project: github.com/ulab-uiuc/LLMR… 🎨 ComfyUI: github.com/ulab-uiuc/LLMR…

Continuous diffusion dominates image & video generation, but people used to believe that it inherently lags behind its discrete counterparts in language modeling. Today, we challenge this belief with LangFlow: the first continuous diffusion language model that rivals—and even beats—discrete diffusion. (1/7) Blog: caradryanl.github.io/blog/2026/lang… GitHub: github.com/nealchen2003/L… Arxiv: arxiv.org/abs/2604.11748

Huge thanks to our first author, @PeterLuo2077 and our amazing supervisors, @youjiaxuan and @GeLiuSaber.

🔥 Introducing MemReward — open-sourced! 📄 arxiv.org/abs/2603.19310 🚀 github.com/ulab-uiuc/MemR… 🤗 huggingface.co/datasets/ulab-… A graph-based experience memory framework that achieves near-Oracle RL fine-tuning with only 20% reward labels — and even surpasses full supervision on out-of-domain tasks. 💸 The Problem RL fine-tuning for LLMs requires reward labels for every rollout. But labels are expensive: math proofs need expert review, open-ended QA lacks ground truth, code verification is slow. Label everything? Too costly. Label only 20%? Performance drops sharply. 🧠 How MemReward Works We organize queries, chain-of-thought, and answers into a heterogeneous graph. A GNN trained on 20% labeled rollouts propagates reward signals to the remaining 80%. Labeled rollouts → ground-truth rewards Unlabeled rollouts → GNN-predicted rewards Combined → efficient GRPO training 📊 Key Results On Qwen2.5-3B / 1.5B across 13 benchmarks (math, QA, code): • 20% labels → 97.3% of Oracle performance (3B) • Surpasses Oracle on OOD tasks: 66.96 vs 66.07 (3B), 62.81 vs 62.00 (1.5B) • At 70% labels → 99.4% of Oracle • Math reasoning benefits most: GSM8K +11.6, GSM-Sym +14.9 (1.5B) ✅ Heterogeneous graph with query-query, query-thinking, thinking-answer edges ✅ Cross-domain GNN: joint training, zero-shot OOD generalization ✅ Plug-and-play: drop-in replacement for reward sources in GRPO ✅ Smooth scaling: more labels → better results, starting from just 20% Label less. Learn more. 🚀 If you find this useful, please give us a ⭐ on GitHub! #LLM #Memory #Opensource

You can do AI researach on JEPA on 1 GPU, no more "i don't have GPUs" excuse - github.com/facebookresear…

Yes, it's about 3-4 times faster than a single-round router. However, the training of an agency router can be optimized. If the user considers both latency and task performance, the agency router should learn to perform single-round routing for simple queries and routing based on the agency mechanism for complex queries. Stay tuned to our project; we will be releasing some of the latest experimental results soon.

@taofeng_uiuc llm routing at the agent level is underexplored. what's the latency overhead on the routing decisions themselves?

We release OpenClaw Router — Production-Ready LLM Routing 🚀 Code: github.com/ulab-uiuc/LLMR…📦 PyPI: pypi.org/project/llmrou… 🔥 Meet OpenClaw Router Deploy LLMRouter as an OpenAI-compatible API with one command. Seamlessly integrate with Slack, Discord, WhatsApp via OpenClaw. Support multimodal understanding — route based on images, audio, video, not just text. pip install llmrouter-lib && llmrouter serve 💸 Why OpenClaw Router? Why pay GPT-5 prices for "What's the weather?" Smart routing = Significant Token Savings. Simple query → Cheap model. Complex query → Powerful model. Save 30-50% on inference costs without sacrificing quality. 🧠 Train Your Own Router OpenClaw Router isn't just a server — it's a learning system. Train personalized routers on your own data, tailored to your domain. Every user feedback, every usage pattern feeds back into router training — continuously iterate toward a more user-friendly, cost-efficient OpenClaw. ✅ Routing Memory: RAG-powered decisions that learn from history ✅ Personalized Routing: Adapts to individual user preferences ✅ Feedback Loop: User interactions improve routing over time ✅ 16+ Strategies: KNN, SVM, MLP, BERT, Graph, RL, Agentic — switch with one flag Route smarter. Train your own. Save more. 🚀

Thanks to our main contributor @liangqi_yuan, and other contributors @haozhen_ntu, @lei_zijie. Special thanks to our advisors Prof. @GeLiuSaber and Prof. @youjiaxuan for their guidance and support! 🙏

5/5 ✨ Smart Features • Intelligent caching: Skips regeneration when config unchanged • Real-time monitoring via PreviewAny nodes • Pre-configured example workflow included • Multimodal support: Video understanding (Charades-Ego) • Works with any ComfyUI setup

🎨 Introducing ComfyUI Interface for LLMRouter - Build your entire LLM routing pipeline visually! We're excited to release a powerful visual interface that transforms how you work with LLMRouter. No more YAML configs or terminal scripts - just drag, drop, and connect nodes. 🔗 Project: github.com/ulab-uiuc/LLMR… 🎨 ComfyUI: github.com/ulab-uiuc/LLMR…

We just released a new survey of Agent Memory! we frame agent memory along three orthogonal axes: • Substrate — where memory lives • Cognitive mechanism — what role it plays (episodic/semantic/procedural) • Subject — who it serves (user/agent)

Thanks! Great questions 🙌 We’re currently running extensive user studies and experiments across different routing models to measure actual cost savings, latency impact, and user satisfaction. We’ll be releasing detailed benchmarks and aggregated statistics soon — please stay tuned and keep an eye on the project for updates!

@taofeng_uiuc @openclaw This looks very cool, juts a couple of questions: Have you benchmarked actual cost savings? What kind of reduction did you see in your own usage? and Does adding a routing layer slow down response times noticeably?

🦞 OpenClaw 2026.2.12 is out! 🔥 GLM-5 + MiniMax M2.5 💬 IRC channel — your bot fits right in with the old guard 🛡️ 40+ security fixes 📦 Custom provider onboarding, compaction improvements & more Your agent called. It wants an upgrade. github.com/openclaw/openc…

Feels like LLM memory, skills, and world models went from niche ideas to everywhere almost overnight 👀 Why do you think these directions suddenly exploded? Scaling limits? Agent autonomy? Learning from experience? Something else? And more fun question: what’s the next big breakout in LLM research? Drop your predictions 👇 #LLM #AIresearch #AIAgents #MachineLearning #ArtificialIntelligence #WorldModels #LLMMemory #AISkills #FutureOfAI

As a loyal OpenClaw user, I love how it powers my daily writing & coding workflows. But two pain points kept bugging me: 💸 excessive token costs and 🤖 model selection that doesn't really get my preferences. So we built something about it — 16+ LLM routers, fully open-sourced! Some are cost-aware to help you save money, while others are personalized routers that learn from your OpenClaw interaction history to adapt model selection to YOU — the more you use it, the better it knows you. Check it out & let us know what you think! We hope this makes everyone's OpenClaw experience smarter & more personal 🙌 🔗 OpenClaw Router: github.com/ulab-uiuc/LLMR… 🔗 Full Project: github.com/ulab-uiuc/LLMR… #OpenClaw #LLMRouter #OpenSource #LLM #AI #Personalization

Love OpenClaw but hate the token burn? 💸 Running a 24/7 agent on GPT-4/Claude is overkill. You don't need SOTA reasoning to handle a greeting or a simple lookup. LLMRouter 🩷 OpenClaw The first production-ready, agentic router designed to plug directly into OpenClaw. LLMRouter fully supports Multimodal, Memory-Equipped routing that adapts 100% to your needs—compatible with FREE open-source models. The Logic is Simple:🔹 Simple query → Cheap/Local model 🔹 Complex reasoning → SOTA model (GPT-4/Claude 3.5) 🔹 Multimodal input → Vision/Audio specialized model Why this isn't just a switch: 📉 30–50% drop in inference costs 🧠 Zero loss in response quality 🔓 100% compatible with OpenAI-style APIs 🚀 Deploy in Seconds General Usage: Get the library and serve any model: pip install llmrouter-lib llmrouter serve OpenClaw Native Integration: Want the full agent experience? LLMRouter built a dedicated integration for OpenClaw users: github.com/ulab-uiuc/LLMR… LLMRouter Resources: 🔗 Repo: github.com/ulab-uiuc/LLMR… 📦 PyPI: pypi.org/project/llmrou… 🤝 Works with: github.com/openclaw/openc… Route smarter. Train your own. Pay less. More on LLMRouter: Most routers are static if/else. LLMRouter is an intelligent, learning system. 🤖 Agentic & Memory-Aware: Decisions aren't stateless. We use RAG-powered memory to route based on context and history. 👤 Fully Personalized: It learns from your usage patterns via RL feedback loops. 🔬 Research-Grade: Switch between 16+ routing strategies (KNN, SVM, BERT, Graph, RL) with a single flag.

🙏 Acknowledgments Thanks to all contributors: @haozhen_ntu @lei_zijie @nehzux @liangqi_yuan . Special thanks to our advisor @youjiaxuan and @GeLiuSaber for guidance and support.