Knowledge catalogue
arXiv:2610.06962v1 Announce Type: cross Abstract: In many review workflows the verdict is the only thing retained. The passages behind it are not marked, because that annotation costs far more than re
arXiv:2410.05479v2 Announce Type: replace Abstract: Uncertainty-aware explanation methods often produce several alternatives for the same prediction. Selecting among them requires a policy for balanci
arXiv:2610.08718v1 Announce Type: new Abstract: Knowledge that a language model appears to forget during finetuning often remains stored and can be recovered, a phenomenon called spurious forgetting.
arXiv:2610.08097v1 Announce Type: cross Abstract: Mathematical problem solving often requires deterministic computational steps that agents delegate to tools and implicitly trust. Yet tools can fail s
arXiv:2609.33155v2 Announce Type: replace Abstract: Test-time scaling and post-training have improved LLM performance in coding and mathematical reasoning, but their effectiveness for individual stanc
arXiv:2610.07365v1 Announce Type: new Abstract: As LLMs increasingly assist scientific writing and peer review, detecting who wrote the text is no longer sufficient: we need to determine who contribut
arXiv:2610.04364v1 Announce Type: new Abstract: Designing effective reward functions in RL typically requires substantial expertise and trial and error. Recent work automates this process with LLM-bas
Co-authors:Stenal Jolly, Strategic Cloud Engineer, GoogleAnubhav Dhawan, Software Engineer, Google Following the landmark announcement of MCP Toolbox v1.0, we're thrilled to announce that the MCP Tool
arXiv:2610.05863v1 Announce Type: new Abstract: High-quality child care in early life is a critical determinant of children's growth and development. Research on child care quality has been constraine
arXiv:2610.04286v1 Announce Type: cross Abstract: Distributed Energy Resource (DER) environments rely on network communication protocols to coordinate control commands, measurements, and device states
arXiv:2610.05572v1 Announce Type: new Abstract: Predicting whether an LLM agent will fail has emerged as a promising direction for supporting intervention during execution. Recent approaches report st
arXiv:2601.23027v2 Announce Type: replace Abstract: Long chain-of-thought reasoning (Long CoT) is now fundamental to state-of-the-art LLMs, especially in mathematical reasoning. However, LLM generatio
arXiv:2610.04283v1 Announce Type: new Abstract: Activation steering exploits interpretable directions in the residual stream to enable inference-time manipulation of model behavior. Composing steering
arXiv:2610.04862v1 Announce Type: cross Abstract: Long-horizon problem solving and scientific research require computation to accumulate across successive attempts. Partial solutions, experimental fin
arXiv:2610.04544v1 Announce Type: cross Abstract: Many groups now delegate label ontology and metadata harmonization to agentic LLM pipelines. We built one and audited it. Our aggregate scores looked
arXiv:2610.05564v1 Announce Type: cross Abstract: Text embedding models trained with contrastive learning learn to follow task instructions from instruction-paired data, while instruction-tuned LLMs a
arXiv:2610.05458v1 Announce Type: new Abstract: Running the same language model on different graphics processing unit (GPU) vendors can produce different logits, even when the model weights and inputs
Mistral AI SAS today opened access to Mistral Large 4, its most capable large language model to date. On launch, the LLM is available in public preview through the company’s cloud platform. Mistral pl
arXiv:2610.06360v1 Announce Type: cross Abstract: Reinforcement learning post-training for language models relies on two reward designs: human preferences (RLHF, DPO) and binary verifiers (RLVR). Clin
arXiv:2610.04976v1 Announce Type: new Abstract: We evaluate language models' capability to generate novel research ideas through the lens of compression. We aim to lower-bound the potentially tiny pro
arXiv:2610.06843v1 Announce Type: new Abstract: LLM agents that orchestrate frozen vision-language-action (VLA) policies improve across episodes through text memory, which records what the agent did b
Training a single LLM agent jointly across diverse interactive environments has attracted increasing attention as a route to generalist agents. Existing curriculum and data-selection strategies often
arXiv:2610.04403v1 Announce Type: cross Abstract: Post-training quantization to the GGUF format's mixed-precision K-quants is commonly how open-weight language models reach consumer hardware, yet its
arXiv:2610.04008v1 Announce Type: cross Abstract: Executable Agent Skills combine natural-language instructions and scripts into reusable packages for LLM agents, and revising them requires fixing err
arXiv:2610.04156v1 Announce Type: cross Abstract: LLM agents for clinical text-to-SQL applications reason autonomously over multiple steps but cannot assess whether their own reasoning or outputs can
arXiv:2610.02970v1 Announce Type: new Abstract: Large language models (LLMs) have shown promising potential for biomedical named entity recognition (BioNER) through instruction following and in-contex
arXiv:2610.03007v1 Announce Type: cross Abstract: Long-output reasoning shifts the KV-cache bottleneck from the fixed prompt to the generated trace. Existing reasoning-cache eviction methods largely t
llama: support both embd + raw tokens in batch (#29622) llama: support both embd + raw tokens in batch add to test-llama-archs also check case llm_arch_supports_mixed_batch = false constant graph topo
arXiv:2610.02376v1 Announce Type: cross Abstract: Co-designing ML models and the accelerators that run them is an unusual reasoning task: architects must draw confident, high-stakes conclusions about
arXiv:2610.02873v1 Announce Type: cross Abstract: The evolution of linguistic style in conversations is an underexplored issue in NLP. Most style-control datasets focus on sentences or assume a static
arXiv:2610.03361v1 Announce Type: cross Abstract: Critic-free reinforcement fine-tuning (RFT) for agentic large language models is often done through GRPO-style methods, which compute a group baseline
arXiv:2610.02697v1 Announce Type: cross Abstract: Recent vision-and-language navigation (VLN) systems increasingly adopt streaming Video-LLM policies that map egocentric RGB observations and instructi
Hot take: evals are the #1 thing keeping companies from becoming AI-native. If you want to understand evals, read this post & then the accompanying guide. Tweaking prompts and trying a few inputs by h
arXiv:2610.02772v1 Announce Type: cross Abstract: Large language models (LLMs) increasingly serve as general-purpose interfaces to factual knowledge, but their parameters do not automatically reflect
arXiv:2602.00066v1 Announce Type: cross Abstract: Large Language Models (LLMs) have shown strong capabilities in code generation, but their adherence to fine-grained user intent with multiple constrai
arXiv:2610.02381v1 Announce Type: new Abstract: On-policy distillation (OPD) trains a student on the responses it generates. Existing LLM multi-teacher OPD transfers what specialists predict through t
arXiv:2610.02792v1 Announce Type: new Abstract: Legal systems are increasingly implemented through software, yet scalable methods for translating legal texts into accurate symbolic representations rem
arXiv:2610.03312v1 Announce Type: new Abstract: Background: We address the problem of planning when the set of feasible states or actions changes over time. For example, in the problem of path plannin
arXiv:2610.03135v1 Announce Type: new Abstract: Serving long-context autoregressive language models is constrained by the key-value (KV) cache. Most dynamic eviction methods score token importance per
Over the summer I talked to the CEO of Springboards, a startup building an LLM that’s designed to come up with a wider variety of responses than its mainstream rivals do. At the start of the call, he
arXiv:2610.03667v1 Announce Type: new Abstract: Policy-gradient methods are central to modern reinforcement learning, including LLM post-training. When they struggle, the usual suspects are exploratio
arXiv:2610.02925v1 Announce Type: new Abstract: Safety monitors help safeguard language-model agents interacting with external tools and environments, but conservative monitoring can generate many fal
arXiv:2610.03082v1 Announce Type: new Abstract: Bridges contribute significantly to transportation connectivity and urban development. Therefore, reliable bridge monitoring is crucial for protecting p
arXiv:2610.03029v1 Announce Type: new Abstract: Gene set analysis is a cornerstone of functional genomics, yet it remains labor-intensive and heavily dependent on manual curation and expert biological
arXiv:2610.03585v1 Announce Type: cross Abstract: Security benchmarks for LLM-based agents often report the attack success rate (ASR) as a measure of model robustness and use these scores to compare d
arXiv:2610.03324v1 Announce Type: new Abstract: The scale of online content makes hate-speech moderation challenging, while Large Language Models (LLMs) enable harmful material to be produced and adap
Overview llama.cpp v0.6.0 introduces the new llama_batch_ext extended batch API (with llama_process) for mixed token/embedding inputs and MTP/deepstack state embeddings, adds support for the GLM-5.3-F
arXiv:2610.03565v1 Announce Type: new Abstract: We describe the Writerslogic systems for three PAN at CLEF 2026 shared tasks (Reasoning Trajectory Detection, Voight-Kampff Generative AI Detection, and
50K likes on this!? Most of what Karpthy has shared is things we have been doing for over a year. No surprises, as many replies are highlighting. But I still think this raises an interesting discussio
I've been wanting to test out LLM inference on FPGAs for a while now, but didn't have a big reason too since there were no frontier class models at 9B-27B scale (3.6 27B was great still but it didn't
we're seeing our coding agent costs decrease significantly for second month in row longer blog on this later, but three quick steps we took 1/ cost visibility. you cant control what you cant measure.
I've come up with a meta-prompt for generating YuE2 prompts that seems to work pretty well. Copy-paste into an LLM (it should have internet searching capabilities, I mostly just use ChatGPT). Just to
There are quite often posts here complaining that Krea 2 is too predictable / different seeds give almost the same output / you need to type in extremely long prompts to get good images / there is no
Benchmarking an LLM here with a NVIDIA RTX 5070 12 GB VRAM here I had been working on a llama.cpp based expert streaming setup for Qwen3.8-Flash-Next 177B (UD-IQ3_XXS) on Windows. Benchmark is about 1
arXiv:2610.00651v1 Announce Type: new Abstract: Agent evaluations are increasingly used to compare LLMs and inform deployment decisions, yet ranks can reflect not only the model but also the effects o
arXiv:2610.01821v1 Announce Type: new Abstract: Understanding information processing in large language models (LLMs) requires dissecting the geometric organization of their internal token representati
arXiv:2606.15420v2 Announce Type: replace-cross Abstract: Safety evaluations test a policy on prompts that omit the incentive information deployment supplies: a commission, a performance score, a dash
Fine-tuning teaches a small search agent your tools and environment, giving it the reliability of a frontier model at lower latency and cost. In this post, we fine-tune an LLM-powered search agent wit
arXiv:2610.02048v1 Announce Type: new Abstract: When a SCADA alarm is raised in a water distribution network, operators must decide quickly whether it reflects a cyberattack, a physical fault, a norma
arXiv:2610.00437v1 Announce Type: new Abstract: LLM agents generate intermediate reasoning and actions token by token, making extended interactions slow and computationally expensive. Jev-style models