Search: “research”

the inference platforms are doing hardcore work, post-training open weight models to solve customer problems: @FireworksAI_HQ ember-1 @fal H…

the inference platforms are doing hardcore work, post-training open weight models to solve customer problems: @FireworksAI_HQ ember-1 @fal H3 max more to come;) makes total sense considering their pos

industrysonya-huang--x

27 Sept 2026

Content type

All Blog X Post Paper YouTube Reddit GitHub

Clear filters

Safety

There’s a SIGNIFICANT token cost associated with their hacking rampage. They’ve spent millions of dollars attacking other orgs. They are lia…

DGX agent

There’s a SIGNIFICANT token cost associated with their hacking rampage. They’ve spent millions of dollars attacking other orgs. They are liable and negligent. Attacker economics has always been a part

safetygary-marcus--x

27 Sept 2026

Safety

This cover certainly didn’t age well.

This cover certainly didn’t age well. TIME’s new cover: In 2026, OpenAI has seen key departures, rogue AI agents, major lawsuits, and has seen increased competition in the AI race. “We clearly had som

safetygary-marcus--x

27 Sept 2026

We released VeriLoop E2 (27B, Apache-2.0). The design question behind it: should an LLM be allowed to commit its own state?

Disclosure: I’m one of the authors of VeriLoop E2, a 27B model post-trained from Qwen3.8-27B. The weights are Apache-2.0; the harness we evaluate it in is not open source (details at the bottom). This

27 Sept 2026

85 GB DeepSeek-V4-Flash at ~3 tok/s on a 12 GB RTX 3060 + 64 GB DDR5 RAM - Overspill for FreeToken, inspired by Colibri

I've been experimenting with ways to run MoE models that don't fit comfortably in RAM, and I ended up making Overspill, a disk tier for FreeToken. The basic idea came from looking at how Colibri handl

At what point do LLMs start bootstrapping themselves and generating the next LLM

I suppose technically if that happens it will signal the true start of a singularity because from that point on progress will be exponential and not dependant on humans. right now you still need huge

local-air-localllama

Google Threat Intelligence Group finds dark web marketplaces selling access to AI models, including from Anthropic, Google, and OpenAI, at up to 97% discounts (Tom Wilson/Financial Times)

Tom Wilson / Financial Times: Google Threat Intelligence Group finds dark web marketplaces selling access to AI models, including from Anthropic, Google, and OpenAI, at up to 97% discounts — Security

industrytechmeme

How accessible is local AI actually, and what happens if affordable access to frontier models doesn’t last?

Sometimes it’s easy to forget that this sub and others like it are probably the extreme minority when it comes to this hobby. Most people, I would think, don’t use or can’t afford one good GPU, let al

incredibly based that Xiaomi MiMo released their RL dataset link: https://huggingface.co/datasets/XiaomiMiMo/MiMo-V2.6-RL-oss

model-releasesclem-delangue--x

Xiaomi’s MiMo project has released a reinforcement‑learning dataset called “MiMo‑V2.6‑RL‑oss.” The dataset is hosted on Hugging Face and can be accessed at https://huggingface.co/datasets/XiaomiMiMo/M

This is huge. Xiaomi just open-sourced 7k+ reinforcement learning task environments used to train MiMo, covering code, cybersecurity, genera…

model-releasesclem-delangue--x

This is huge. Xiaomi just open-sourced 7k+ reinforcement learning task environments used to train MiMo, covering code, cybersecurity, general tool use, visual web development, and music-related tasks.

xiaomi MiMo oss RL environments + training code https://github.com/XiaomiMiMo/verl https://huggingface.co/datasets/XiaomiMiMo/MiMo-V2.6-RL-o…

Xiaomi has released an open‑source reinforcement‑learning framework called MiMo (version V2.6) that bundles a suite of simulated environments together with training scripts. The code is hosted on GitH

industryclem-delangue--x

4-5 days replacing Claude w Qwen 3.8 Next

Hi, human here with rambling thoughts to share. Feel free to skip Overall, I don’t feel like I’m missing much; if anything. On my hardware(M1 ultra w 128Gb) it’s probably not as fast as Claude but I’v

Applications

A Procedure for Classifying Attachments and Affective Social Bonds in Human-Robot Dyads

arXiv:2609.29063v1 Announce Type: cross Abstract: Human-robot interaction (HRI) claims that people form attachments and social bonds with artificial agents, yet the terms are often applied without the

applicationsarxiv-cs-ro

A Rapid Pipeline for Training and Deploying ML Models on WeBe Band

arXiv:2609.29084v1 Announce Type: new Abstract: Developing optimized machine-learning algorithms for edge devices with limited computational and memory resources is challenging, time-consuming, and hi

local-aiarxiv-cs-ai

Banger paper from Microsoft and colleagues. We talk about human taste being important in this AI era. But for recursive self-improvement, an…

Banger paper from Microsoft and colleagues. We talk about human taste being important in this AI era. But for recursive self-improvement, an agent's taste also matters. The big question is: How often

model-releasesdair-ai--x

Blockchain-Enabled Artificial Intelligence and AI Agents for Secure Data Sharing and Cybersecurity Applications

arXiv:2609.28843v1 Announce Type: cross Abstract: Blockchain and artificial intelligence (AI) are converging into a single infrastructural layer for securing data sharing, model integrity, and autonom

agentsarxiv-cs-ai

CodeGraph: Open-Taxonomy Knowledge Graph for Source Code with Wikidata Grounding

arXiv:2609.29474v1 Announce Type: cross Abstract: Public software repositories, like GitHub and Software Heritage Archive, store billions of files, yet extracting their implicit engineering knowledge

agentsarxiv-cs-cl

Delay-of-Gratification as a Multi-Agent Survival Micro-benchmark for Long-Horizon LLMs: Social Exposure, Personas, and Tool Use Budgets

arXiv:2609.29509v1 Announce Type: new Abstract: Large language models (LLMs) are increasingly deployed as multi-turn agents that must sustain goals, use tools, and adapt to other agents over extended

Tutorials

Functional dynamic mode decomposition: Learning infinite-dimensional systems from data

arXiv:2609.29159v1 Announce Type: cross Abstract: Dynamic mode decomposition (DMD) is a data-driven method that computes the best linear approximation of the underlying dynamical system and decomposes

tutorialsarxiv-cs-lg

GPT-6-Astra Lights Up Embodied Navigation: Evaluation in Zero-Shot Vision-and-Language Navigation in Continuous Environments

model-releasesarxiv-cs-ro

arXiv:2609.29861v1 Announce Type: new Abstract: We investigate whether GPT-6-Astra, a general-purpose foundation model, can navigate unfamiliar environments using its own perception, reasoning, and de

iCoder-27B: Recursive AI-Led Development of Frontier Industrial Coding Model

arXiv:2609.29626v1 Announce Type: new Abstract: Recursive AI, the prospect of AI taking an increasingly complete role in building and improving AI, is a crown jewel of AI for AI. Although recursive se

Interpretable Similarity of Synthetic Image Utility

model-releasesarxiv-cs-cv

arXiv:2512.17080v2 Announce Type: replace Abstract: Synthetic medical image data can unlock the potential of deep learning (DL)-based clinical decision support (CDS) systems through the creation of la

Hardware

Less is More: Encoder-only Audio-Visual Segmentation

arXiv:2609.29121v1 Announce Type: cross Abstract: Audio-Visual Semantic Segmentation (AVSS) aims to identify, segment, and classify sound-emitting objects in video frames. Previous Transformer-based A

hardwarearxiv-cs-ai

Tutorials

Neuro-symbolic AI for Industrial Configuration

arXiv:2609.29947v1 Announce Type: new Abstract: Large Language Models (LLMs) have shown impressive performance on a wide range of generative tasks. Yet their probabilistic nature makes them, in isolat

tutorialsarxiv-cs-ai

Qwengram-0.8B: I transferred Qwen3.8 Flash-Next’s n-gram memory into Qwen3.5-0.8B — 5.05% lower validation perplexity

I’ve been experimenting with whether Qwen3.8-Flash-Next’s pretrained PLE n-gram memory can improve a much smaller Qwen3.5-0.8B model. I trained the 0.8B setup with limited resources, mostly using free

Recommended benchmark. I expect voice to become one of the main ways people interact with robots and physical AI. That makes speech recognit…

Recommended benchmark. I expect voice to become one of the main ways people interact with robots and physical AI. That makes speech recognition in real conversations much more important than it seems

model-releasesdair-ai--x

Applications

Spatio-temporally complementary feature propagation on graphs for longitudinal AADT estimation

arXiv:2609.29906v1 Announce Type: new Abstract: The estimation of Annual Average Daily Traffic (AADT) is vital for transportation planning and infrastructure maintenance, yet obtaining accurate values

applicationsarxiv-cs-lg

Stream Recursion Model (SRM)

model-releasesarxiv-cs-lg

arXiv:2609.28809v1 Announce Type: new Abstract: Mechanistic interpretability seeks to make verifiable statements about the internal behavior of large language models (LLMs). Many interpretability tech

Tired of RAM prices, so I have been pooling spare RAM across old devices to run bigger local models

Been dealing with the same thing a lot of people here probably are - RAM prices keep climbing and buying a single machine with enough memory for the models I actually want to run just isn't realistic

local-air-localllama

UCON: Uncertainty-aware Navigation with Historical Re-association in Dynamic Environments

arXiv:2609.29419v1 Announce Type: new Abstract: Autonomous navigation in dynamic environments is hindered by two fundamental challenges: perception instability and uncertainty-optimization mismatch. T

agentsarxiv-cs-ro

Vanderbilt University extends identity governance to AI agents

Identity governance must accommodate changing relationships among people, institutions and technology. Vanderbilt University sees that complexity deepening as artificial intelligence agents enter inst

industrysiliconangle

WebArxiv: A Reproducible Benchmark for Evaluating Multimodal Web Agents on arXiv Tasks

arXiv:2507.00938v4 Announce Type: replace-cross Abstract: Foundation models now enable autonomous agents to interact with real-world websites, but existing benchmarks emphasize general-purpose browsin

Safety

When Agents Act Unwatched: The Reduced-Supervision Paradox in Agentic AI

arXiv:2609.29547v1 Announce Type: cross Abstract: Agentic AI is sold on a simple promise: the system keeps acting when the user stops watching. That promise creates an accountability inversion. As ste

safetyarxiv-cs-ai

When Fancy Eviction Fails: Rethinking Cache Replacement For LLM Prefix Reuse

arXiv:2609.28870v1 Announce Type: cross Abstract: Long-running LLM applications repeatedly send growing context, making prefix caching critical for reducing prefill cost. Yet prefix-cache behavior und

agentsarxiv-cs-lg

Working with Agentic `Teammates': When a New Organizational Actor Collides with the Human Ecosystem of Work

arXiv:2609.29901v1 Announce Type: cross Abstract: Enterprise AI is transitioning from single-user, reactive tools toward proactive, multi-user 'teammates,' but our empirical understanding of this tran

agentsarxiv-cs-ai

A Systematic Evaluation of Infrastructure-Based Radar System for Highway Traffic Monitoring

model-releasesarxiv-cs-cv

arXiv:2609.27143v1 Announce Type: new Abstract: Infrastructure-based radar systems offer robust and long-range solutions for traffic monitoring, yet their detection and tracking performance under real

Accenture leverages Oracle’s Deep Data Security for database-driven trust boundaries

Oracle Corp.’s Deep Data Security offers a basic premise on enterprise security: When the application layer is an AI agent, the trust boundary must move somewhere else. The trust boundary can be reloc

agentssiliconangle

Tutorials

Active Learning for Biodiversity Monitoring: From Label Efficiency to Reliable Ecological Inference

arXiv:2609.27409v1 Announce Type: new Abstract: Limited expert annotation capacity is a pervasive constraint in biodiversity monitoring. Passive acoustic recorders and camera traps generate data faste

tutorialsarxiv-cs-lg

After months of heads-down grinding, I’m proud to share that the @GPTZeroAI team has released the world’s most accurate AI detection model, …

model-releasesemad-mostaque--x

After months of heads-down grinding, I’m proud to share that the @GPTZeroAI team has released the world’s most accurate AI detection model, GPTZero 4o. It’s called 4o because there are four zeros: we

AgroBench: A Reproducible Multimodal Benchmark for Weakly Supervised Crop Yield Learning from County Statistics and Pixel Observations

model-releasesarxiv-cs-cv

arXiv:2609.26809v1 Announce Type: new Abstract: Reliable agricultural yield statistics are typically reported at coarse administrative scales, whereas modern geospatial machine learning methods requir

ChronoSteer: Bridging Large Language Model and Time Series Foundation Model via Synthetic Cross-Modal Alignment Dataset

model-releasesarxiv-cs-lg

arXiv:2505.10083v2 Announce Type: replace Abstract: Conventional forecasting methods are trained end-to-end on unimodal time series, which limits their ability to exploit textual information and under

Compliant with Local Controls, Collectively Discriminatory. A Governance Architecture for Multi-Agent AI in Regulated Finance

arXiv:2609.27994v1 Announce Type: cross Abstract: Financial institutions are beginning to deploy agentic workflows in credit, fraud, collections, compliance, and operational control. Governance remain

local-aiarxiv-cs-ai

Hardware

Does Step Law Transfer to Small-Scale Language Models? An Empirical Recalibration Below 59M Parameters

arXiv:2609.27581v1 Announce Type: cross Abstract: Step Law gives power-law formulas for the optimal peak learning rate eta* and batch size B* when pre-training language models. It was calibrated on mo

hardwarearxiv-cs-cl

Evaluating Feedback Focus and Pedagogical Adaptivity in LLM-Generated Feedback on Student Writing

arXiv:2609.28026v1 Announce Type: cross Abstract: We investigate whether state-of-the-art large language models (LLMs) generate feedback that reflects the pedagogical practices of expert teachers in t

Hardware

Fentanyl Grade Compute: Why a GB300 Rack Will Out-Price Blow by Weight in 2030 A GB300 NVL72 weighs roughly 1,580 kg fully populated and rec…

Fentanyl Grade Compute: Why a GB300 Rack Will Out-Price Blow by Weight in 2030 A GB300 NVL72 weighs roughly 1,580 kg fully populated and recent purchase orders put it at 5M per rack. That is 3,165/kg.

hardwaredylan-patel--x

Safety

From Agent Output to Authorized Transition

arXiv:2609.28216v1 Announce Type: cross Abstract: Agentic engineering systems can edit repositories, run tools and tests, build firmware, synthesize schematics, and prepare deployable or manufacturabl

safetyarxiv-cs-ai

How growing Latin American midsize businesses are building in the AI era

model-releasesgoogle-cloud-ai

Latin America’s small and medium-sized businesses are the heartbeat of the region's economy — accounting for more than 60% of total employment in the region, according to United Nations estimates. And

Hunyuan-A13B Technical Report

arXiv:2609.27284v1 Announce Type: new Abstract: We present Hunyuan-A13B, an open-source large language model based on a Mixture-of-Experts architecture. It contains 80 billion total parameters but act

agentsarxiv-cs-ai