Spend half decade on embedding then now I am unemployed bc of it 😭
Embeddings were a mistake.
We matched OpenAI’s best embedding model on FiQA without using a single embedding. No vector database. No chunking. No re-indexing.
Today, we're launching @wondersearchai searcheverything your company ever wrote and find the exact line.
@ycombinator @fdotinc @daltonmeon
Thanks for sharing been trying to add more subset in my work but found out is seem the idea I had was wrong
This is why I like diffusion model so much you need deep understanding of math to contribute the field it the makes the field enjoyable
I never imagined a lab that was not big ai lab will break the million context window
He truly choose the way of the Buddha
Hot take: the community keeps asking whether diffusion LLMs can beat AR. That's the wrong question. Diffusion is a sampling method, and it can sit on top of an AR model to make it faster instead of replacing it.
Speculative decoding already speeds up AR, but it has two long-standing problems. Training a drafter takes a long time, and you have to build a complete pretraining corpus for it. Even then, token acceptance drops sharply on out-of-domain data.
Uno @ssahoo_ offers an elegant way out: it decouples sampling training from capability training. There's no separate drafter. The model itself, plus one LoRA, drafts several tokens in a single forward pass, and the LoRA learns to match the model's own next-token distribution, so there's no drafter corpus to build and no labels to collect.
We applied this to the Looped Transformer Nanbeige4.2-3B: 1 hour of LoRA training gives a lossless 2× speedup on agentic task. We haven't optimized anything yet, and we believe the acceleration can go much further.
The takeaway: diffusion doesn't need to beat AR. Any model that can score its next token can get a fast diffusion mode in an hour.
Page: fengsxy.github.io/LoopUno
Lora Adapters: huggingface.co/liddlefish/Lo…
As person just keep prompting model to make as many subagent as you want this is a problem.
Anthropic is now recommending that companies REMOVE unnecessary subagents from their AI systems.
This is a pretty significant shift in agent architecture, with their team finding that a single agent equipped with skills can frequently outperform complex multi-agent systems, and one customer seeing roughly 2x faster execution.
The emerging rules of thumb:
→ Skills: Teach the agent HOW to do something, including procedures, expertise, and instructions.
→ Tools / MCP: Give the agent the ability to DO things through APIs, actions, and live data.
→ Subagents: Use for parallel work, isolated context, or tasks a single agent struggles with.
The interesting part is progressive disclosure. Instead of spawning specialized agents or stuffing everything into a giant prompt, one agent discovers and loads skills and tools as needed.
Their advice is to fold existing subagents into the main loop, turn their instructions into skills, expose their actions as tools, and run evals to see if performance holds up.
As models have gotten smarter, improving agent architecture is becoming less about what you can add and more about what you can delete.
claude.com/resources/article…
There have been so many paper showing the combining diffusion with other work improve reasoning. The future of diffusion is much brighter than autoregressive.
Chinese LLM are not slowing down, a lot people where coping with their success by saying distillations
Introducing Kimi K3.1 Pro: our most capable 3T-class open-weight model.
🔹 1M-token context, 2.8T total / 104B active params (MoE), Multimodal.
🔹 Full-project coding: games in Unreal, Unity & Godot.
🔹 Excels at parametric 3D modeling in Blender & CAD.
🔹 More reliable at finance work: Accurate modeling with hallucination down by 65%.
🔹 Matching the strongest proprietary models across coding, agentic and knowledge benchmarks.
Kimi K3.1 Pro and K3.1 Pro Highspeed (2–3× faster) are now live on Kimi.ai, Kimi Work, Kimi Code, and the Kimi API.
Open weights by Nov 4, 2026.
🔗 API: platform.kimi.ai
🔗 Tech blog: kimi.ai/blog/kimi-k3-1-pro
@GaryMarcus might be right
LLMs run on vectors, not symbols. A preprint led by Yale’s Tom McCoy (@RTomMcCoy) with CDS Assoc. Prof. @tallinzen found those vectors closely approximate symbolic structure in language, arithmetic, logic, and code.
arxiv.org/abs/2608.29530
news.yale.edu/2026/10/08/how…
I think it great other people are sharing their cold email approach, it give me hope a person looking for a job
Finally, someone out here using optimization theory for agent
1/8 New paper with Ivan Ilin: "Local Work, Global Progress: A Convergence Theory of AI Agent Collaboration".
We all know that teams of LLM agents can tackle big tasks. However, their collaboration protocols lack guarantees. We model collaboration as block optimization and, under suitable assumptions, prove convergence.
arxiv.org/abs/2610.28286
