model releases

Operating Multi-Node Full Fine-Tuning on NVIDIA B300: A Field Report on Telemetry-Based Triage, Negative Results, and Operational Hardening

arXiv:2608.05944v1 Announce Type: cross Abstract: We report operational experience full-fine-tuning a 32.76B-parameter dense model (Qwen3-32B) on 16 x NVIDIA B300 (two nodes, FSDP / ZeRO-3) -- among t

Content type

All Blog X Post Paper YouTube Reddit GitHub

Clear filters

OrchestraBench: Evaluating Multi-Agent Orchestration Failure Modes, Recovery, and Decomposition Quality

arXiv:2608.05263v1 Announce Type: new Abstract: Multi-agent orchestration frameworks are moving from demos to production, yet benchmarks typically report task accuracy without diagnosing why a pipelin

Otter: A Time-Aware, History-Conditioned Human Chess AI

arXiv:2608.05206v1 Announce Type: new Abstract: Otter is a 15.3M-parameter human chess AI that predicts human move selection by modeling play as a time-aware, sequential process rather than treating e

Outstanding cost-to-performance from DeepSeek GPT-5.6 Luna (Max) performance for a 1/4th of the cost on ARC-AGI

model-releasesfrancois-chollet--x

Outstanding cost-to-performance from DeepSeek GPT-5.6 Luna (Max) performance for a 1/4th of the cost on ARC-AGI DeepSeek V4 Flash from @deepseek_ai on ARC-AGI (Verified): - ARC-AGI-2: 61.4%, 0.04/task

Parameter-Efficient Semantic Augmentation for Enhancing Open-Vocabulary Object Detection

arXiv:2604.04444v2 Announce Type: replace Abstract: Open-vocabulary object detection (OVOD) enables models to detect any object category, including unseen ones. Benefiting from large-scale pre-trainin

Perfect reconstruction of sparse signals using nonconvexity control and one-step RSB message passing

arXiv:2512.17426v2 Announce Type: replace-cross Abstract: We consider sparse signal reconstruction via minimization of the smoothly clipped absolute deviation (SCAD) penalty, and develop one-step repl

PhaseCoder: Microphone Geometry-Agnostic Spatial Audio Understanding for Multimodal LLMs

arXiv:2601.21124v2 Announce Type: replace-cross Abstract: Current multimodal LLMs process audio as a mono stream, ignoring the rich spatial information essential for embodied AI. Existing spatial audi

Plausible Patients, Impossible Populations: Auditing Epidemiological Fidelity in Large Language Model Mental Health Simulations

arXiv:2604.17359v2 Announce Type: replace-cross Abstract: Language models asked to simulate psychiatric patients produce cases that survive inspection one at a time and populations that match no real

PoolBench: A Benchmark for Pooling Strategies in Concept Representation Evaluation for Decoder-Only LLMs

arXiv:2608.05162v1 Announce Type: new Abstract: Pooling is a consequential but under-examined design choice in decoder-only concept representation work: practitioners must collapse token-level hidden

Position: It's Time to Optimize LLMs for Self-Consistency

arXiv:2608.05188v1 Announce Type: cross Abstract: Despite ever-increasing sophistication in language model (LM) pre- and post-training pipelines, many important failures persist: models overcondition

Project2Task: Graph-Guided Project-Level Planning for Autonomous Research

arXiv:2608.05225v1 Announce Type: new Abstract: Research agents can increasingly search literature, propose hypotheses, generate code, run experiments, and draft manuscripts from a single topic. Howev

Qwen 3.6 27B flags/settings in llama.cpp

model-releasesr-localllama

I run the following on a 5090 and have been okay with its performance, it does most things somewhere 80-100 t/s, though that can slow down at full 262k context - more like 40 t/s at times. I use it pr

Rectifying Geometric Misalignment: Online Source-Free Adaptation for Class-Imbalanced EEG

arXiv:2608.05315v1 Announce Type: new Abstract: Electroencephalography (EEG) based Brain-Computer Interfaces (BCIs) often require unsupervised domain adaptation (UDA) to generalize across subjects and

Recursive Synthesis for Long-Horizon Terminal Tasks

arXiv:2608.05466v1 Announce Type: new Abstract: High-quality long-horizon training data for terminal agents is expensive to produce, often costing hundreds to thousands of dollars per task, because ea

Retailers are updating their websites to rank highly in chatbot results, while making sure purchases are done on their own sites to collect customer data (Arriana McLymore/Reuters)

Arriana McLymore / Reuters: Retailers are updating their websites to rank highly in chatbot results, while making sure purchases are done on their own sites to collect customer data — As shoppers incr

model-releasestechmeme

Robot Learning from Human Demonstrations: Handwritten Alphabet Trajectories and Human-Likeness Evaluation

arXiv:2608.06221v1 Announce Type: cross Abstract: Learning from demonstration (LfD) provides a developmental framework through which robots can develop motor skills by observing and imitating human dy

Robust Native Language Identification through Agentic Decomposition

arXiv:2509.16666v2 Announce Type: replace Abstract: Large language models (LLMs) often achieve high performance in native language identification (NLI) benchmarks by leveraging superficial contextual

Runtime Observability for Heterogeneous Attention Memory

arXiv:2608.05863v1 Announce Type: new Abstract: Modern models no longer keep a plain KV cache: latent caches, learned sparse selectors and recurrent states each carry the model's memory in a different

SafeDivertor: Faithful Divertor Heat Flux Reconstruction from Macroscopic Plasma State Signals via Time-Frequency Prior Exploitation

arXiv:2608.05669v1 Announce Type: cross Abstract: Divertor heat-flux analysis is essential for understanding plasma-wall interactions and protecting plasma-facing components in magnetic-confinement fu

Scaffold-Mediated Post-Training: Co-Evolving Model Parameters and Procedural Scaffold Graphs

arXiv:2608.05156v1 Announce Type: new Abstract: Post-training of large language models optimizes only parameters, while inference-time procedural scaffolds are typically designed independently of para

Schema-Guided Hierarchical Information Extraction and Semantic Evaluation Using Generative AI

arXiv:2608.06167v1 Announce Type: new Abstract: We present a schema-based framework for extracting complex, structured information from unstructured text documents using generative AI, followed by aut

Scientific Machine Learning of Chaotic Systems Learns Reduced-Order Equations for Neural Populations

arXiv:2507.03631v4 Announce Type: replace Abstract: Extracting interpretable mathematical models from complex dynamical systems is difficult, especially for chaotic dynamics observed with noisy experi

SEAM: Global consistency beyond local accuracy in scientific machine learning

arXiv:2608.05702v1 Announce Type: new Abstract: Scientific machine learning commonly validates models at the level of a subdomain, a benchmark split, or an explanation for one prediction. Yet such loc

SearchAuditor: Auditing and Attributing Failures in Long-Horizon Search Agents

arXiv:2608.05212v1 Announce Type: new Abstract: Deep search agents tackle challenging questions through long-horizon web interactions, a process that is both complex and fragile: small reasoning error

Seeing Is Not Deciding: Can Multimodal LLMs Act as Effective CEOs?

arXiv:2608.05864v1 Announce Type: new Abstract: Large language models are increasingly applied as autonomous decision-making agents. However, in executive business decisions, existing benchmarks are l

SemiAdapt-Instruct: Extensible Instruction Tuning via Latent Domain-Specialised Adapters

arXiv:2608.05161v1 Announce Type: new Abstract: Instruction-tuned LLMs are deployed into environments where domains evolve, yet extending a fine-tuned model's capabilities without full retraining rema

Serving Deepseek v4 Flash 0731 on 2x DGX Spark — 5-7 GB OS headroom, what would you do to lower VRAM usage and increase OS available RAM?

model-releasesr-localllama

Hey all, I'm serving DSv4Flash 0731 on a cluster of 2x DGX Sparks but am running into constant issues with having almost no RAM (unified memory) left for the OS/cache and I'd love to hear the communit

Shrinking the Generation-Verification Gap with Weak Verifiers

arXiv:2506.18203v3 Announce Type: replace Abstract: Verifiers can improve language model capabilities by scoring and ranking responses from generated candidates. Currently, high-quality verifiers are

Simulator-Grounded Large Language Models for Industrial Causal Reasoning: Tool-Use, Structured Injection, and Plant-Portable Retrieval for Wastewater Treatment Decision Support

arXiv:2608.05151v1 Announce Type: cross Abstract: Wastewater operators need answers grounded in how their plant's variables interact and how fast effects propagate, not in generic pretraining text, wh

SkillHEX: Improving Agent Skills via Hypothesis-Driven Autonomous Exploration and Exploitation

arXiv:2608.05628v1 Announce Type: new Abstract: Although agent skills equip LLMs with reusable procedural knowledge, manual maintenance suffers from high costs, unscalability, and misalignment. Real-w

SkillMemo: Expert-guided Skill Memory Framework for Compositional Embodied Manipulation

arXiv:2608.05970v1 Announce Type: cross Abstract: Embodied visuomotor models, including Diffusion Policy (DP) and Vision-Language-Action (VLA) models, have demonstrated promising performance on roboti

SkillTFM: Gated Skill Evolution for Training-Free Adaptation of Tabular Foundation Models

arXiv:2608.06137v1 Announce Type: new Abstract: Tabular data are ubiquitous in real-world applications and are crucial for data-driven prediction and decision-making across science, industry, finance,

SkillTV-Bench: Benchmarking How Well Judges Perform on Skill-Augmented Agentic Execution

arXiv:2608.05573v1 Announce Type: new Abstract: LLM agents increasingly execute long-horizon tasks through tool use and environment interaction, shifting evaluation from final-response scoring to veri

StepReflect: Structured UI Transition Reflection for Mobile GUI Agents

arXiv:2608.05587v1 Announce Type: new Abstract: Autonomous mobile GUI agents require accurate action reflection for reliable long-horizon execution. Existing approaches rely on open-ended multimodal r

Stochastic Parrots or Singing in Harmony? Testing Five Leading LLMs for their Ability to Replicate a Human Survey with Synthetic Data

arXiv:2603.00059v3 Announce Type: replace-cross Abstract: How well can AI-derived synthetic research data replicate the responses of human participants? An emerging literature has begun to engage with

StreamArena: Toward Continuous, Interactive, and Long-Horizon Agentic Streaming Video Understanding

arXiv:2608.05703v1 Announce Type: new Abstract: Deploying autonomous multimodal agents in continuous, real-world environments requires them to ingest unbounded audio-visual streams and maintain hour-s

Subliminal Learning is Non-Semantic Distillation

arXiv:2608.05734v1 Announce Type: new Abstract: Subliminal Learning (SL) is a surprising type of generalization displayed by modern language models. It allows the transfer of a bias or behavior from a

Task-Conditional Flow Matching for Balanced Multilingual Text Embedding Adaptation

arXiv:2608.05785v1 Announce Type: cross Abstract: Multilingual text embedding models are commonly adapted using a single training objective across diverse tasks, despite different tasks requiring fund

TAU-Bench: From Anomaly Instance Tracking to Fine-Grained Video Anomaly Understanding

arXiv:2608.05699v1 Announce Type: new Abstract: Humans understand anomalous events through a coherent perceptual process in which they identify the focal instance, follow its behavior as the event unf

Text-Guided Refinement of Multi-sequence Glioma Subregion Segmentation with a Vision-Language Foundation Model

arXiv:2608.05389v1 Announce Type: new Abstract: Background: Accurate glioma subregion delineation is important for radiotherapy planning and longitudinal monitoring, but manual contour correction is t

Thanks for the thorough testing! With Qwen3.8-Max, everyone can observe the world in detail. 👀

Alibaba’s Qwen team announced the new Qwen‑3.8‑Max model and thanked users for extensive testing, noting its ability to provide highly detailed world observations. A community member highlighted that

model-releasesqwen--x

THBKG: A Temporal Biomedical Knowledge Graph for Decision-Aligned Clinical Advancement Prediction

arXiv:2608.05982v1 Announce Type: new Abstract: Inadequate target--disease linkage accounts for 40--50% of Phase~II efficacy failures, so anticipating which programmes will advance would let sponsors

The backgrounds were generated with MiniMax M3. Everything else, including the full game design, was done with DeepSeek Flash. Most importan…

model-releasesnous-research--x

The backgrounds were generated with MiniMax M3. Everything else, including the full game design, was done with DeepSeek Flash. Most importantly, all of this was done inside Hermes Agent. You won't bel

The Bitter Lesson of Tool Calling

arXiv:2608.06370v1 Announce Type: new Abstract: Tool use transforms LLMs into agents that act beyond their training data, and for code-capable models, programmatic tool calling extends this further by

The cloud was always a cat. Qwen3.8-Max just saw it first. 😼☁️ Try it yourself!

The cloud was always a cat. Qwen3.8-Max just saw it first. 😼☁️ Try it yourself! Qwen 3.8 Max is actually impressive Sent a sky pic to it along with Claude Opus 5, Kimi K3, and GPT 5.6 Sol, and asked t

model-releasesqwen--x

The Ignition Index: Measuring Global Workspace Dynamics in Language Models

arXiv:2608.05160v1 Announce Type: new Abstract: We introduce the Ignition Index (I), a validated scalar metric that operationalizes Global Workspace Theory's (GWT) all-or-none ignition prediction in t

The Low Frequency Trap: Video Language Models Fail at Simple Event Bookkeeping

arXiv:2608.06361v1 Announce Type: new Abstract: Real-world video benchmarks provide broad coverage, but their fixed clips entangle event count, rate, duration, and visual complexity, making failure mo

The Next Screenshot Knows: Gated Hindsight Distillation for Mobile GUI Agents

arXiv:2608.06065v1 Announce Type: new Abstract: GUI agents are commonly trained offline from successful interaction trajectories. Standard training decomposes each trajectory into prefix-action pairs: