Knowledge catalogue
arXiv:2604.07030v1 Announce Type: new Abstract: Sparse Mixture-of-Experts (MoE) architectures are increasingly popular for frontier large language models (LLM) but they introduce training challenges d
arXiv:2604.07658v1 Announce Type: cross Abstract: Linear recurrent models offer linear-time sequence processing but often suffer from suboptimal long-range memory. We trace this to the decay spectrum:
arXiv:2510.05921v2 Announce Type: replace Abstract: Large language models (LLMs) have achieved remarkable success in a wide range of natural language processing tasks and can be adapted through prompt
arXiv:2603.10512v2 Announce Type: replace Abstract: Artificial intelligence has advanced significantly through the development of intelligent game-playing systems, providing rigorous testbeds for deci
arXiv:2604.06628v1 Announce Type: new Abstract: A prevailing narrative in LLM post-training holds that supervised finetuning (SFT) memorizes while reinforcement learning (RL) generalizes. We revisit t
arXiv:2604.08501v1 Announce Type: cross Abstract: Science currently offers two options for quality assurance, both inadequate. Journal gatekeeping claims to verify both integrity and contribution, but
arXiv:2604.03128v2 Announce Type: replace Abstract: On-policy distillation (OPD) has become a popular training paradigm in the LLM community. This paradigm selects a larger model as the teacher to pro
arXiv:2412.10437v3 Announce Type: replace Abstract: Generating high-quality Scalable Vector Graphics (SVGs) from text remains a significant challenge. Existing LLM-based models that generate SVG code
arXiv:2604.04956v2 Announce Type: replace-cross Abstract: The recent, super-exponential scaling of autonomous Large Language Model (LLM) agents signals a broader, fundamental paradigm shift from machi
arXiv:2604.06182v1 Announce Type: cross Abstract: Existing online benchmarks for mobile GUI agents remain largely app-centric and task-homogeneous, failing to reflect the diversity and instability of
arXiv:2604.06277v1 Announce Type: new Abstract: Existing hallucination detection methods for large language models (LLMs) rely on external verification at inference time, requiring gold answers, retri
arXiv:2604.08510v1 Announce Type: new Abstract: Large language models (LLMs) can perform remarkably complex tasks, yet the fine-grained details of how these capabilities emerge during pretraining rema
llama.cpp release **b8739** is a build of the open-source C/C++ LLM inference engine that introduces HIP backend support for the CDNA4 (gfx950) GPU architecture, enabling hardware acceleration on A...
ConvApparel is a human-AI conversation dataset and comprehensive evaluation framework designed to quantify the 'realism gap' in LLM-based user simulators and improve the training of robust conversa...
In this post, you will learn how to build stateful MCP servers that request user input during execution, invoke LLM sampling for dynamic content generation, and stream progress updates for long-runnin
This paper was accepted at the Workshop on Memory for LLM-Based Agentic Systems at ICLR. Language models have consistently grown to compress more world knowledge into their parameters, but the knowled
Routing every task to your largest model burns tokens, adds latency, and inflates costs. @AI21Labs' Maestro Orchestration Meta Model (OMM) is the layer above your stack that dynamically selects the ri
**v0.20.5-rc2** is a pre-release release candidate for the Ollama open-source local LLM runner (github.com/ollama/ollama), published around April 10, 2026, as part of the v0.20.5 release cycle. It ...
Meta announced Muse Spark today, their first model release since Llama 4 almost exactly a year ago. It's hosted, not open weights, and the API is currently 'a private API preview to select users', but
I ran five coding tasks against nine models on one RX 9060 XT (16 GB). Each run gets a single spec file in an empty directory and has to produce solution.py and test_solution.py, run its own tests, an
arXiv:2610.20630v1 Announce Type: new Abstract: The development of Financial Natural Language Processing (NLP) systems for Brazilian Portuguese is hindered by the scarcity of domain-specific datasets.
arXiv:2610.20809v1 Announce Type: new Abstract: A self-improving agent has to decide which new skills to keep, and in practice it keeps whatever earns a high reward, convinces a judge model, or compre
arXiv:2610.22319v1 Announce Type: new Abstract: We placed four Gemini 4 Argon agents in a sandboxed Minecraft survival world and gave them one goal: defeat the Ender Dragon. Each agent controlled a pl
arXiv:2610.20641v1 Announce Type: new Abstract: Long chain-of-thought decoding is memory-bound: the KV cache of the generated tokens grows linearly with the test-time compute being spent. Evicting cac
arXiv:2507.22612v3 Announce Type: replace-cross Abstract: Speech-to-text alignment is a critical component of neural text to speech (TTS) models. Autoregressive TTS models typically use an attention m
arXiv:2602.16653v4 Announce Type: replace Abstract: Skills are getting increasingly used in recent agentic frameworks, performing robustly with proprietary LLMs. Yet, their effectiveness for small and
arXiv:2610.21938v1 Announce Type: cross Abstract: An agent that learns a user's preferences from their choices faces a cold start problem: before any choices are observed, it must rely solely on a pri
arXiv:2606.21550v2 Announce Type: replace Abstract: Alignment from human feedback uses human judgments about model outputs to steer the behavior of language models after pretraining. When those judgme
arXiv:2610.21147v1 Announce Type: new Abstract: Large language models increasingly rely on external memory to answer queries grounded in long-horizon histories such as multi-session dialogues and leng
arXiv:2610.21882v1 Announce Type: cross Abstract: Embodied spatial intelligence requires more than visual recognition: models must localize entities, reason about their relations, and express this kno
arXiv:2610.21055v1 Announce Type: new Abstract: Recursive self-improvement, in which AI systems build stronger AI systems, is one promising route to superintelligence. We study a weaker form that can
arXiv:2610.21581v1 Announce Type: new Abstract: Mathematical agents are commonly evaluated by answer accuracy or proof acceptance. We introduce ClaimScope, a 72-task benchmark for auditing proposed cl
arXiv:2610.20679v1 Announce Type: new Abstract: Representation-based safety defenses for vision-language models (VLMs) often learn to separate harmful from benign inputs, but separation alone does not
arXiv:2610.22049v1 Announce Type: cross Abstract: Reconstructing what a person perceives from their brain activity is a stringent test of whether neural recordings carry meaningful information. Natura
arXiv:2610.21561v1 Announce Type: cross Abstract: Large language models (LLMs) deployed as conversational agents are increasingly expected to follow procedural logic. While factual hallucination has r
arXiv:2610.20652v1 Announce Type: new Abstract: Always-on virtual assistants increasingly personalize on-device to new user-specific commands via few-shot fine-tuning (e.g., LoRA), but this adaptation
arXiv:2610.22130v1 Announce Type: new Abstract: Key-value (KV) cache management has become a critical bottleneck for scaling Large Language Models (LLMs) to long-context scenarios due to linear memory
arXiv:2610.22290v1 Announce Type: new Abstract: Reinforcement learning has become a common post-training technique to improve the accuracy of LLMs on verifiable tasks. However, such training can narro
arXiv:2610.21368v1 Announce Type: cross Abstract: With the expansion of pre-trained language model usage in recent years, the importance of datasets for performing tasks in specialized domains has sig
arXiv:2606.12674v2 Announce Type: replace Abstract: Small language models (1.5B-4B parameters) offer low-cost, low-latency tool use but often produce plausible plans that fail during execution due to
arXiv:2610.22061v1 Announce Type: new Abstract: Organizations increasingly encode operational procedures as multi-step agent skills, here called Agent Operating Procedures (AOPs), that must be traceab
arXiv:2610.21948v1 Announce Type: cross Abstract: As multimodal AI systems are becoming our everyday thought partners, the shift towards studying how users transfer control to these systems becomes es
arXiv:2509.16998v2 Announce Type: replace Abstract: Design for Robotic Assembly (DfRA) remains largely dependent on manual planning and heuristic simulation, limiting scalability and robustness in com
arXiv:2610.22289v1 Announce Type: new Abstract: Stronger language models may increasingly construct useful procedures for themselves, reducing the advantage of developer-supplied scaffolding. We inves
arXiv:2412.07251v1 Announce Type: cross Abstract: Large language models have exhibited significant enhancements in performance across various tasks. However, the complexity of their evaluation increas
arXiv:2610.22244v1 Announce Type: cross Abstract: Step-level guards decide, before each tool call, whether an agent's next action is safe given the user's request and the trajectory so far. Users writ
arXiv:2610.21427v1 Announce Type: new Abstract: Learning from interaction experience is essential for enabling large language models to continually improve during deployment. Context distillation offe
arXiv:2610.22080v1 Announce Type: new Abstract: A language model can generate several solutions to the same problem, some leading to correct answers and others to incorrect ones. Many post-training me
arXiv:2610.21912v1 Announce Type: cross Abstract: Understanding the objectives that drive agents' behavior is crucial in strategic environments, where multiple agents interact and adapt to one another
arXiv:2610.21025v1 Announce Type: cross Abstract: When does a language model acquire a capability? Developmental studies usually answer this by tracking when models start producing the correct token.
arXiv:2610.22314v1 Announce Type: new Abstract: With people increasingly turning to AI chatbots for mental health support, the need to evaluate these tools is becoming more pressing. Recent benchmarks
arXiv:2610.22344v1 Announce Type: new Abstract: Large language models (LLMs) are expanding from assisting scientific research and engineering to autonomously developing and improving AI systems, advan
arXiv:2610.22086v1 Announce Type: cross Abstract: Supervised fine-tuning datasets remain overwhelmingly English-centric and commercially restricted, as their content is frequently generated by proprie
arXiv:2603.26107v1 Announce Type: cross Abstract: People increasingly use multiple Multimodal Large Language Models (MLLMs) concurrently, selecting each based on its perceived strengths. This cross-pl
arXiv:2610.22019v1 Announce Type: cross Abstract: Frontier models, general-purpose and domain-specific alike, now include a midtraining stage between pretraining and post-training, but little has been
arXiv:2610.21220v1 Announce Type: cross Abstract: Recommender systems often need to balance multiple conflicting objectives. Controllable Multi-Objective Recommendation (CMOR) enables dynamic adjustme
arXiv:2610.20763v1 Announce Type: cross Abstract: Full-duplex spoken language models decide at every frame whether to take the floor, defining a conversational personality -- yet each set of weights e
arXiv:2610.21539v1 Announce Type: new Abstract: Diffusion models (DMs) have steadily demonstrated outstanding performance in high-fidelity, controllable generation across diverse scenarios, following
arXiv:2610.22243v1 Announce Type: new Abstract: AI agents take actions through tools. Some actions are sensitive, so guards that supervise them must themselves be evaluated reliably. Yet the data used
arXiv:2610.21440v1 Announce Type: new Abstract: In automatic speech recognition (ASR), transcription errors with similar lexical or semantic discrepancies can have markedly different consequences for