Browse by topic
- LLM (9)
- AI Agents (9)
- Domain LLMs (1)
- Math (6)
- Analysis (1)
- Algebra & Foundations (0)
- Geometry & Topology (1)
- Probability & Finance (5)
- Physics (5)
- Reading (4)
- Books (0)
- Papers (4)
- Robotics & Embodied AI (2)
- Reinforcement Learning (2)
- Conferences (1)
- ICML 2026 (1)
-
Coconut: What It Actually Looks Like to Reason Without Words
CoGHP showed latent chain-of-thought works in a control domain; the faithfulness post argued verbalized CoT is a real liability. Coconut is the LLM-native mechanism that closes the loop — reasoning as fed-back hidden states instead of decoded tokens — plus the honest cost: what you give up to get there.
-
Does the Chain of Thought Say What the Model Actually Did? The Faithfulness Problem
Explicit chain-of-thought reads as a reasoning trace, but two papers show it can be a post-hoc story instead — and, more troublingly, that more capable models tend to produce less faithful explanations. What 'faithful' means, the two core results, and why this is the real argument for latent reasoning.
-
Ambient Attention for Long-Running Agent Sessions: kmux-observer
Dashboards fail for watching multiple long-running AI agent sessions — context-switch cost, staleness, visual noise, disconnection from the actual session. kmux-observer's answer: treat sessions like browser tabs, not dashboard rows. What it is, what's actually built (v0.1) vs. planned, and how it connects to the agent-loop escalation question.
-
Orchestrator-Workers + Augmented Function-Calling + Agent Loop: The Combination Nobody's Tried
Three agent-design patterns exist separately in the literature — task-decomposing orchestrators, tool-augmented reasoning, and iterative loops with escalation. Nobody has combined all three with fine-tuning specifically for working inside that loop. This is the general framing; the financial-LLM post is one worked instance of it.
-
Sector-Specialized Financial LLMs: Filing Evidence, Calculation Tools, and Verifiable Reasoning
A research-framing post: why nobody has combined fine-tuning, an agent loop, and escalation for evidence-grounded financial reasoning — and the two open questions that matter most (a chain-of-thought method for tool-verifiable calculation, and how to generate training data for it).