Knowledge catalogue
.@Kimi_Moonshot K3 lands on Together on Monday! We ran 452 DeepSWE rollouts against Claude Fable 5: near-flagship coding at ~35% of the price, and K3 pulls ahead at higher pass@k's. Full deep-dive: ht
arXiv:2607.20723v1 Announce Type: cross Abstract: This work presents LeakyLMs, a set of attacks that leak proprietary model, architecture, and deployment information from production language models. L
arXiv:2607.20456v1 Announce Type: cross Abstract: Large language models excel at code generation for mainstream programming languages but struggle with rare, domain-specific languages such as MiniZinc
Learning causality from internet videos in latent space first, and then using RL to teach the foundation model how to act. This approach is 30× cheaper than Gemini 3.1 Flash on pretraining and achieve
arXiv:2607.11029v2 Announce Type: replace-cross Abstract: Recent progress in visual navigation has largely been driven by scale: end-to-end policies with hundreds of millions of parameters trained on
arXiv:2607.20872v1 Announce Type: new Abstract: Long-form legal research reports increasingly rely on LLMs and agentic research systems, but their reliability depends not only on answering the task, b
arXiv:2603.20509v2 Announce Type: replace Abstract: Camera traps are vital for large-scale biodiversity monitoring, yet accurate automated analysis remains challenging due to diverse deployment enviro
arXiv:2607.20924v1 Announce Type: new Abstract: Makeup-transfer applies the reference makeup to the source face while preserving the source identity. Despite advances in full-face editing by diffusion
arXiv:2607.21372v1 Announce Type: cross Abstract: Score Entropy Discrete Diffusion (SEDD) parameterizes discrete reverse processes with unconstrained positive score ratios. While positivity guarantees
arXiv:2607.21570v1 Announce Type: new Abstract: Large Language Models (LLMs) show promise for medical education, but most existing systems focus on localized interactions such as question answering or
arXiv:2607.20769v1 Announce Type: new Abstract: Learning-enabled decision systems often use offline data or computation to reduce online compute cost. Despite the empirical success of such approaches,
arXiv:2607.21404v1 Announce Type: new Abstract: While memory systems are essential for agent architectures, pervasive architectural fragmentation restricts systematic research. Existing implementation
Meta is upgrading its AI chatbot with new productivity features in a bid to compete with rivals like Gemini, ChatGPT, and Claude. The update will allow Meta AI to tap into your calendar to help you pl
arXiv:2509.22047v3 Announce Type: replace Abstract: Group Relative Policy Optimization (GRPO) has been shown to be an effective algorithm when an accurate reward model is available. However, such a hi
arXiv:2607.20433v1 Announce Type: cross Abstract: While language models remain frozen at their training state, the world evolves continuously. Knowledge editing has emerged as a key alternative to ful
arXiv:2607.20908v1 Announce Type: cross Abstract: Reinforcement Learning with Verifiable Rewards (RLVR) has emerged as a powerful technique to enhance the reasoning capacity of LLMs for optimized code
arXiv:2607.20517v1 Announce Type: new Abstract: Retrieval-augmented generation (RAG) over heterogeneous PDF collections remains challenging due to multimodal content, domain-specific terminology, and
arXiv:2607.21384v1 Announce Type: new Abstract: Electroencephalography (EEG) models used for epilepsy are often limited to specific datasets and tasks. This limited approach can make it challenging to
arXiv:2607.21061v1 Announce Type: new Abstract: Affective Image Content Analysis (AICA) aims to recognize and understand emotions elicited by visual content, representing an indispensable step toward
arXiv:2607.20442v1 Announce Type: new Abstract: We release Naver-News-KO, a Korean news summarization dataset of 27,400 (document, summary) pairs collected from Naver News over a ten-day window in Jul
New research with Microsoft's and colleagues on training agents inside the harnesses they actually run in. (bookmark it) Why it matters: Agents today live inside elaborate harnesses like Claude Code,
arXiv:2601.15363v2 Announce Type: replace-cross Abstract: Functional bilevel optimization (FBO) provides a powerful framework for hierarchical learning in function spaces, yet current methods are limi
'The NVIDIA Qwen-Image-Flash model generates images from text prompts using a four-step, DMD2-distilled version of Qwen/Qwen-Image. The distillation used DMD2 from NVIDIA FastGen, NVIDIA Model Optimiz
arXiv:2607.20670v1 Announce Type: new Abstract: Modeling continuous object deformation is important for many computer vision and robotics tasks, such as manipulation and simulation. Existing approache
On FrontierCode 1.1, our benchmark for real-world engineering tasks that grades mergeability and quality, Opus 5 scores 63.6% with a 69.6% pass rate on Extended — approaching Fable 5 at half the cost.
arXiv:2607.21143v1 Announce Type: cross Abstract: Ambiguous user requests make clarification a sequential decision problem for conversational LLM assistants: they must decide whether to ask, what to a
When we introduced the Open Knowledge Format (OKF) in June 2026, we asserted that the context that agents need (table schemas, metric definitions, runbooks) should live in a format, not in a proprieta
This is an open source and free tax engine which scored 96% on TaxCalcBench [highest ever recorded score till date] surpassing fable 5 and sol with just sonnet 5. The only 2 cases where it missed, it
arXiv:2607.21557v1 Announce Type: new Abstract: Modern AI agents rely on elaborate inference harnesses such as Claude Code, Codex, and OpenClaw to drive multi-turn reasoning, tool use, and access to e
I am building a domain-specific AI agent powered by Ollama (using the qwen:2.5 model). For data retrieval, the agent utilizes multiple search APIs: DuckDuckGo Search (DDGS), Tavily, Serper, and Google
David Gewirtz / ZDNET: Opus 5 improves coding, reasoning efficiency, and prompt-cache-friendly tool use, and is priced at 5/1M input tokens and 25/1M output, same as Opus 4.8 — ZDNET's key takeaways —
Opus 5 is a great model for coding, data analysis, design, biology, knowledge work. More than any of these eval scores, what is most exciting to me is something else: Opus 5 is our least prompt inject
Claude Opus 5 is now released in the Hermes Agent, a product of Nous Research and Teknium. Users can access the model through multiple gateways, including the Nous Portal, OpenRouter, and Anthropic Di
arXiv:2607.21268v1 Announce Type: cross Abstract: In many social-science research tasks, such as economics, LLM-based agents must produce outputs for which no cheap, task-complete, machine-readable co
Model quantization has become essential for efficient large language model deployment, yet existing approaches involve clear trade-offs: methods such as GPTQ and AWQ achieve practical compression but
Someone modelled sugarcane farming as an integer program. See, sugarcane only grows next to water. Water costs one tile and can feed at most four cane tiles. The layout therefore becomes a coverage pr
arXiv:2607.20482v1 Announce Type: new Abstract: Recent advances in large language models have enabled web agents to autonomously execute complex tasks. In practice, users frequently provide underspeci
arXiv:2607.20492v1 Announce Type: cross Abstract: Language models in production do not write prose. They fill forms: JSON fields, function arguments, extraction templates. We show that the form itself
arXiv:2603.13026v2 Announce Type: replace Abstract: Prompt injection poses serious security risks to real-world LLM applications, particularly autonomous agents. Although many defenses have been propo
arXiv:2607.20864v1 Announce Type: cross Abstract: Position bias in multiple-choice LLM evaluation is widely cited as a confound in capability comparisons, but published measurements rely on single ans
arXiv:2607.20438v1 Announce Type: cross Abstract: Preference-based post-training is usually understood through endpoint behavior, yet the learned update that produces this behavior remains largely opa
arXiv:2607.21022v1 Announce Type: new Abstract: Improving video captioning quality typically demands retraining large vision-language models, an expensive and often impractical requirement. Existing t
arXiv:2607.20806v1 Announce Type: new Abstract: Lightweight large language models (LLMs) are increasingly being deployed locally on personal computers and are expected to play a growing role in resour
arXiv:2607.20528v1 Announce Type: new Abstract: Online recommendation platforms increasingly use Large Language Models (LLMs) to extract structured features from ad creatives. While deploying a single
arXiv:2607.21063v1 Announce Type: new Abstract: Almost every large language model that reaches a broad audience is quantized: trained in full precision, then compressed for efficiency. This step is as
arXiv:2607.20455v1 Announce Type: cross Abstract: Human-annotated data remains fundamental to training frontier Large Language Models (LLMs). However, crowd-sourced annotations often suffer from quali
arXiv:2607.20791v1 Announce Type: new Abstract: High-temperature sampling is one of the primary mechanisms for increasing diversity in LLMs. Recent advances in truncation-based sampling techniques hav
arXiv:2607.20722v1 Announce Type: new Abstract: Large language models trained and aligned within different linguistic and regional ecosystems may frame the same political, cultural, and geopolitical e
arXiv:2607.21120v1 Announce Type: cross Abstract: In reinforcement learning, critics typically estimate absolute state values V(s), estimating how good a particular situation is in isolation. However,
Replit released several major features this month, including a voice‑enabled Agent that lets users talk directly to the platform. Users can now build from Claude or Slack, and the updated Agent rememb
arXiv:2607.21447v1 Announce Type: cross Abstract: The ability to handle long-term memory in LLMs is becoming increasingly critical, yet existing benchmarks remain English-centric and rely on aggregate
arXiv:2607.20767v1 Announce Type: new Abstract: We introduce Rushes, a dataset and benchmark for studying revealed human engagement preferences in interactive narrative environments. Rushes is collect
arXiv:2607.21518v1 Announce Type: new Abstract: Even a current high-capability LLM can appear safer when shown a dangerous objective directly than when other agents transform and relay its direction.
arXiv:2607.20516v1 Announce Type: cross Abstract: Promising initial results in closed-loop large-language-model-based channel-configuration search demonstrated that neural-network widths can be optimi
arXiv:2607.20652v1 Announce Type: cross Abstract: Language models are thought to exhibit the phenomenon of superposition, representing many more features than dimensions in their residual streams. Spa
arXiv:2607.21562v1 Announce Type: new Abstract: Gradient-based saliency methods reveal which input features most influence a neural network's output, and are a standard tool for model interpretability
arXiv:2607.20926v1 Announce Type: new Abstract: Scientific research involves complex information-seeking and reasoning workflows across heterogeneous sources. However, existing benchmarks primarily em
arXiv:2602.10226v2 Announce Type: replace-cross Abstract: Optimizing large-scale machine learning systems, such as recommendation models for global video platforms, requires navigating a massive hyper
arXiv:2607.20477v1 Announce Type: new Abstract: {em Text-Attributed Graphs} (TAGs) have emerged as an expressive data model for integrating graph topology with rich textual semantics. Existing represe
arXiv:2607.20551v1 Announce Type: cross Abstract: Effective molecular representation learning is crucial for accurate molecular property prediction. Recently, numerous self-supervised learning (SSL) a