Knowledge catalogue
Banger report on building faster memory for AI Agents. If your agent's memory layer is slow, this design is worth a look. Great weekend read. It proposes a new agentic memory architecture, Jev-Mem, in
arXiv:2609.28991v1 Announce Type: cross Abstract: Video understanding is increasingly performed by multi-stage LLM agents that separate temporal grounding, visual observation, and reasoning. Yet these
arXiv:2609.30017v1 Announce Type: cross Abstract: Many LLM inference problems, including model routing, prefix-cache management, prompt trimming, and test-time search, can be viewed as optimization ov
arXiv:2609.29051v1 Announce Type: new Abstract: On-policy self-distillation (OPSD) has become a popular recipe for post-training LLM agents. It supervises the agent model at the token level with a str
arXiv:2609.29551v1 Announce Type: new Abstract: Sparse autoencoders (SAEs) decompose LLM activations into sparse dictionary atoms, so that each distinct concept gets its own feature. One recurring beh
arXiv:2601.06676v3 Announce Type: replace-cross Abstract: Large Language Model (LLM)-based deep research agents perform multi-step reasoning, web exploration, and long-form report generation. In these
arXiv:2609.29444v1 Announce Type: cross Abstract: Deep search requires LLM agents to decompose complex queries, search for evidence, and synthesize grounded answers, yet existing ReAct-style agents su
arXiv:2609.29410v1 Announce Type: new Abstract: Recent studies have shown that Large Language Models can effectively solve problems and fix bugs in diverse programming environments, including competit
arXiv:2602.08329v2 Announce Type: replace-cross Abstract: A core bottleneck in large language model (LLM) inference is the cost of attending over the ever-growing key-value (KV) cache. Although near-o
Gonna ironically preface this with: I haven't written the text below with an LLM. boxxy.io So I thought I'd show you what I've made over the last couple of months, because for me this has probably bee
arXiv:2508.18192v2 Announce Type: replace Abstract: Large Language Models (LLMs) have reshaped our world with significant advancements in science, engineering, and society through applications ranging
We announced our RSI Lab earlier this year: https://sakana.ai/rsi-lab/ Over the last two years, we have systematically shipped the foundations for autonomous R&D: ▪ LLM²: AI automating research to inv
arXiv:2609.27038v1 Announce Type: new Abstract: Chain-of-thought (CoT) monitoring assumes that the reasoning a model writes reflects the computation that directly produces its answer. Previous faithfu
arXiv:2607.22115v2 Announce Type: replace-cross Abstract: Given a database S and a natural language question Q, text-to-SQL systems aim to generate an SQL query that correctly answers Q when executed
arXiv:2609.28245v1 Announce Type: cross Abstract: Large language models (LLMs) have shown strong performance in creative text generation, yet their ability to produce culturally grounded and stylistic
arXiv:2609.27067v1 Announce Type: cross Abstract: Large language model (LLM)-based agents use Electronic Design Automation (EDA) tools to generate and revise register-transfer-level (RTL) designs unde
arXiv:2609.27014v1 Announce Type: cross Abstract: Legal contracts are structurally complex documents in which contradictions may emerge across distant and interconnected provisions. Although large lan
arXiv:2609.27869v1 Announce Type: new Abstract: Long-horizon multimodal agents rely on specialized capabilities for perception, retrieval, reasoning, verification, and execution. Existing designs typi
arXiv:2609.28263v1 Announce Type: new Abstract: The growth of large language model (LLM) inference and search services increases the scale of online linear programming problems, motivating computation
arXiv:2609.28274v1 Announce Type: new Abstract: The final safeguard against rogue AI behavior is the human ability to shut systems down. It has been theorized that when an AI is instructed to perform
Been building some multi-agent pipelines recently and the degradation in direct execution is insane. Whenever I pass a strict operational variable or hypothesis, the LLM triggers this 'Devil's Advocat
arXiv:2609.27155v1 Announce Type: cross Abstract: With recent advancements in large language models (LLMs) and LLM-based agents, these agents are becoming increasingly autonomous and gaining broader a
arXiv:2609.25804v2 Announce Type: replace Abstract: LLM agents increasingly work on long-horizon tasks, and the decisions they make along the way, such as which hypothesis to test or which implementat
arXiv:2609.26911v1 Announce Type: new Abstract: A single locally plausible tool call can derail an otherwise successful agent trajectory. Suspicion alone does not justify intervention, because the rep
arXiv:2609.27603v1 Announce Type: cross Abstract: In-Context Learning (ICL) has become a cornerstone of modern LLM deployment. However, existing ICL post-training methods have a critical blind spot: t
arXiv:2609.23894v1 Announce Type: cross Abstract: Agentic AI extends LLM security beyond generated content to persistent state, autonomous actions, tool use, and interactions with humans and other age
Hey everyone! I've been experimenting with Qwen Image 2.1 Prompt Enhancer (PE) lately and noticed an interesting detail in Qwen's official system prompts. The PE models output structured JSON containi
arXiv:2609.24115v1 Announce Type: new Abstract: Tool-calling LLM agents are increasingly deployed in enterprise applications. However, effective evaluation and optimization require high-quality, diver
arXiv:2609.25192v1 Announce Type: new Abstract: Financial search is a highly demanding task for LLM agents, requiring not only a correct final answer but also temporally valid information retrieval, a
arXiv:2609.24277v1 Announce Type: new Abstract: GUI agents predict click coordinates as digit-token sequences, but standard text-LLM confidence estimation methods rank correct clicks from wrong ones o
arXiv:2609.26667v1 Announce Type: new Abstract: Collaboration topology shapes both the performance and execution cost of LLM-based multi-agent systems. Because tasks differ in complexity and required
arXiv:2609.25054v1 Announce Type: new Abstract: For a long-horizon LLM agent, the memory question is not what was once recorded but what currently holds. Most designs answer it only indirectly: every
arXiv:2609.26355v1 Announce Type: new Abstract: Reinforcement learning has become a central component of large language model (LLM) post-training, yet token-level credit lacks a generally accepted mat
arXiv:2609.26629v1 Announce Type: new Abstract: Procedural character generation aims to populate games, simulations, and other virtual worlds with diverse characters. Large language models (LLMs) offe
arXiv:2609.23023v1 Announce Type: new Abstract: Physics-informed neural networks (PINNs) require coordinated choices over network representation, sampling, loss construction, and optimization, while e
arXiv:2609.24130v1 Announce Type: new Abstract: LLM agents can change their own future behavior, raising a basic control question of which self-generated changes should be allowed to persist. We formu
arXiv:2609.22792v1 Announce Type: cross Abstract: LLM agents are increasingly used for security tasks: vulnerability discovery, exploit reproduction, and patch generation. Improving them at the model
arXiv:2609.24847v1 Announce Type: cross Abstract: LLM inference on edge devices is constrained by computational and memory resources, making efficient autoregressive decoding challenging. Speculative
What it is. stuntd is a local proxy for typed LLM decisions: choice, yes/no, score. Your app keeps calling its provider through it. stuntd records the decisions, trains a small head on top of Laya (th
arXiv:2609.25678v1 Announce Type: cross Abstract: Large language model (LLM) agents must generalize from tools seen during training to unseen tools at deployment. A key challenge is tool trialing, i.e
arXiv:2609.26536v1 Announce Type: new Abstract: In LLM-based speech translation, transcription-based chain-of-thought (CoT) suffers from a mismatch between reference transcripts used in supervised fin
arXiv:2602.18307v2 Announce Type: replace-cross Abstract: Large language models have achieved striking results in interactive theorem proving, particularly in Lean. However, most benchmarks for LLM-ba
I've been comparing and working on Gemma 4, Phi-4-Mini, and Qwen3-4B for local LLM use on an 8 GB Raspberry Pi 5. The results were indeed close, but each model has a different strength: Gemma 4 E2B: a
arXiv:2609.22573v1 Announce Type: cross Abstract: LLM agents translate natural-language context, which may include attacker-controlled text, into privileged tool calls, so authorization must remain ef
arXiv:2609.24150v1 Announce Type: new Abstract: Speculative decoding accelerates large language model (LLM) inference by using a lightweight draft model to generate multiple candidate tokens that are
arXiv:2609.24698v1 Announce Type: new Abstract: Repeated execution of the target model during autoregressive decoding is a major source of LLM inference latency. Unlike linear speculation, which follo
arXiv:2609.22114v1 Announce Type: new Abstract: Context compression is widely proposed as a way to cut the token bill of LLM coding agents, and public benchmarks report that aggressive compression pre
arXiv:2609.22949v1 Announce Type: cross Abstract: Existing prompt injection research focuses on single-model chatbot scenarios, where an attacker manipulates one LLM through crafted input. Multi-agent
I have to say that I am quite impressed by recent progress in LLM use for coding. I've never really bothered setting up a local model. However now that I went trough some real $$ in Claude / gpt token
arXiv:2609.22138v1 Announce Type: new Abstract: Hidden-state hallucination probing - training a linear classifier on an LLM's internal activations to detect whether a generated answer is faithful to t
arXiv:2609.22231v1 Announce Type: new Abstract: Long-horizon interactions with LLM-based assistants require memory systems that preserve and update user states, preferences, and interaction histories.
arXiv:2609.24574v1 Announce Type: new Abstract: Computational social science increasingly relies on large language models for text annotation, and the validity of published findings now rests on the l
arXiv:2609.23808v1 Announce Type: new Abstract: While test-time scaling enhances Large Language Model (LLM) agents in long-horizon software engineering (SWE), sparse binary rewards (Pass/Fail) create
arXiv:2609.22538v1 Announce Type: new Abstract: Large language model (LLM) planners can decompose natural-language instructions and select reusable robot skills, but choosing the correct skill does no
arXiv:2609.22609v1 Announce Type: new Abstract: Insertion is a fundamental operation in robotic construction assembly, where variations in material properties and assembly conditions make it difficult
arXiv:2510.26915v2 Announce Type: replace Abstract: While heterogeneous teams have typically been designed for well-specified missions with known semantics, generative intelligence, i.e., large langua
arXiv:2609.22220v1 Announce Type: new Abstract: Benchmarks for LLM-generated GPU kernels decide correctness with a few random inputs and a loose floating-point tolerance, and their verdicts now feed l
arXiv:2609.24128v1 Announce Type: cross Abstract: Judge-specific sensitivity is useful for aggregating pairwise LLM evaluations, but its interpretation depends on which systematic presentation effects
arXiv:2609.22607v1 Announce Type: new Abstract: We argue here that the current dominant practice in LLM human simulation: prompting instruction-tuned assistant language models to role-play personas, i
arXiv:2609.24972v1 Announce Type: cross Abstract: An LLM agent's capability is largely magnified by its harness, namely the prompts, control flow, tooling, memory, and context management surrounding t