model releases

Lit Review on Benchmarking LLMs Running in your phone!: MobileWorld: Benchmarking Autonomous Mobile Agents in Agent-User Interactive and MCP-Augmented Environments

Back to reading about LLMs as agents on your phone doing GUI tasks! This time I read about MobileWorld: Benchmarking Autonomous Mobile Agents in Agent-User Interactive and MCP-Augmented Environments a

Content type

All Blog X Post Paper YouTube Reddit GitHub

Clear filters

LLM-Driven Algorithm Design for Quantum Circuit Synthesis based on Binary Decision Diagrams

arXiv:2609.05327v1 Announce Type: new Abstract: Quantum circuits are central to implementing quantum algorithms on quantum devices, where quantum gates must be reversible. Many quantum algorithms rely

LLM-Guided Program Evolution for Circle Packing: Breaking 10 Packomania Records for $28

arXiv:2609.05093v1 Announce Type: new Abstract: We present Discovery Loop, a lightweight system that uses a large language model (LLM) to iteratively evolve optimization algorithms. Starting from a si

LLM-guided program evolution improves 10 best-known circle-packing solutions (Packomania csqv, N=101-114) [R]

model-releasesr-machinelearning

I used an LLM to iteratively evolve an optimization algorithm rather than solve the packing directly. Starting from a simple seed solver, the LLM proposes algorithmic changes guided by a scoreboard of

MABPD: Multi-Agent Bias Probing & Detection via Structured Argument Debate

arXiv:2609.04841v1 Announce Type: cross Abstract: Media bias in news articles operates through subtle linguistic cues---loaded language, selective framing, and strategic omission---that resist single-

MageTrail - Small Scale Full-Finetune To Show MageFlow 4B Potential As Foundational Architecture

model-releasesr-stablediffusion

Hi! Today I'm releasing MageTrail, a Danbooru/E621 proof of concept Full-Finetune of Microsoft's MageFlow 4B T2I model, using a diversity maximized condensed 41k images dataset (originally made by Lod

Mapping Dark-Matter Clusters via Physics-Guided Diffusion Models

arXiv:2603.14503v2 Announce Type: replace Abstract: Galaxy clusters are powerful probes of astrophysics and cosmology through gravitational lensing: the clusters' mass, dominated by 85% dark matter, d

MaxKernel: Agentic Kernel Generation for TPUs

arXiv:2609.04523v1 Announce Type: new Abstract: Designing and authoring high-performance custom kernels for accelerators is a complex task that requires deep hardware-level expertise. Large Language M

Measuring LLM performance drift: observations and methodology from 31,352 repeated benchmark measurements [D]

model-releasesr-machinelearning

One thing that has bothered me about LLM benchmarks for a while is that most of them are essentially snapshots. A model is evaluated, a score is published, and we tend to talk about that score as if i

Memory as transformation: LETHE, a self-referential gan-inspired architecture

model-releasesarxiv-cs-cl

arXiv:2609.04289v1 Announce Type: new Abstract: LETHE (Latent-parameter Evolution with Temporal Hierarchical quasi-Equilibrium) is a self-referential sonic-oblivion system implemented in SuperCollider

MEOX: Compact Multimodal Mixture-of-Experts for Earth Observation

arXiv:2609.05351v1 Announce Type: new Abstract: Recent advances in Earth Observation representation learning accommodate heterogeneous sensors and missing observations, often through larger architectu

Mitra-v2 Technical Report

arXiv:2609.04540v1 Announce Type: new Abstract: We introduce Mitra-v2, a tabular foundation model that delivers state-of-the-art performance on real-world classification and regression problems, from

MM-IFEval-Pro: A Multilingual and Attack-Resistant Benchmark for Instruction-Following in Vision-Language Models

arXiv:2609.04859v1 Announce Type: new Abstract: As vision-language models (VLMs) rapidly advance in image understanding, cross-modal reasoning, and complex instruction execution, instruction-following

MMTClinic: Multimodal, Multilingual Time Series Question Answering and Reasoning Benchmark for Clinical Domain

arXiv:2609.04842v1 Announce Type: cross Abstract: Time-series data in clinical settings is crucial for capturing dynamic changes in a patient's health over time, enabling timely diagnosis, personalize

Molecular Deja Vu: Digit-Level Retrieval of Published Values in Frontier Language Models

arXiv:2609.05381v1 Announce Type: new Abstract: Large language models (LLMs) are increasingly evaluated on molecular property benchmarks, but accuracy cannot distinguish a model that predicts a proper

Multi-scale Image Representation Compression

arXiv:2609.04274v1 Announce Type: cross Abstract: Overfitted codecs have demonstrated promising performance for image and video compression. In particular, for image compression, the Cool-chic family

Multi-Step Tool-Calling over Korean Open Public APIs: A Benchmark and a Data-Synthesis Recipe

arXiv:2609.05395v1 Announce Type: new Abstract: Data-sovereignty regulations increasingly require public institutions to deploy open-source, on-premise LLM agents that chain multiple tool-calls across

MultiAttenGastro: Multi-Dimensional Attention Augmentation for Gastrointestinal Endoscopy Classification

arXiv:2609.05070v1 Announce Type: new Abstract: Automated gastrointestinal (GI) endoscopy classification requires models that generalize across diverse modalities and class distributions, often far fr

MultihopSpatial: Multi-hop Compositional Spatial Reasoning Benchmark for Vision-Language Model

arXiv:2603.18892v2 Announce Type: replace-cross Abstract: Spatial reasoning is foundational for Vision-Language Models (VLMs), particularly when deployed as Vision-Language-Action (VLA) agents in phys

My LLM framework. (people asked) (SAI)

DISCLAIMER I ramble a lot and am not good at focusing on documentation and or coherent thoughts. TLDR I did things I made a thing and it works for me. I will try to answer any questions. A few people

My Qwen3.8-27B task-aware quant reaches 99% of BF16 reasoning performance at 15% of the size.

TL;DR My TAK quant of Qwen 3.8 27b scored 82.81% on reasoning, comparted with 77.34% for the byte matched Unsloth UD IQ2_S and 83.59% for BF16. Edit: Some of you have tried coding with this reasoning-

NavArena: Automated Construction of Goal-Oriented Navigation Benchmarks from 3D Gaussian Splatting Reconstructions

model-releasesarxiv-cs-ro

arXiv:2609.04602v1 Announce Type: new Abstract: Fixed 3D Gaussian Splatting (3DGS) reconstructions provide realistic novel views but lack the traversability constraints, valid goals, and closed-loop p

Necessary or Sufficient? Evaluating LLM Explanations With Behavioural Evidence

arXiv:2609.05385v1 Announce Type: new Abstract: LLM decision components that can operate within agent workflows often produce action-relevant recommendations or judgements together with explanations.

New Benchmark: The Struggle Bench

How it works. The model being tested is given a server capable of running it's weights and full context. That server is placed in a median priced apartment. The AI is given a bank account with for ren

NS-ST-GraphRAG: Neuro-Symbolic Spatio-Temporal GraphRAG for Literary Knowledge Processing

model-releasesarxiv-cs-cl

arXiv:2609.05139v1 Announce Type: new Abstract: Long-form literary narratives pose a distinctive information-processing challenge for retrieval-augmented generation: relevant evidence is distributed a

One Diffusion Model, Two Roles: Guided Trajectory Planning and Safety-Critical Scenario Generation in Closed-Loop Simulation

arXiv:2609.04921v1 Announce Type: cross Abstract: Diffusion probabilistic models can capture the multi-modal, interaction-rich distribution of joint future trajectories in driving scenes. We show that

One Word, Different Action: A Real-Robot Benchmark for Language-Conditioned Embodied Reasoning

model-releasesarxiv-cs-ro

arXiv:2609.05260v1 Announce Type: new Abstract: Natural-language instruction changes can directly alter robot behavior. A reliable embodied system should preserve its action when the task is unchanged

Optimal Rates for Agentic Networked Information Aggregation

arXiv:2609.05318v1 Announce Type: new Abstract: Building on the pioneering paper of Kearns, Roth, and Ryu (SODA'26), we study information aggregation in a networked learning model. The model captures

ParaRNN: Parallel Training for Non-Linear RNNs (sLSTM, RWKV-7, CfC, Titans) via Triton Newton Scans [P]

model-releasesr-machinelearning

I've built and open-sourced ParaRNN, a PyTorch/Triton library that parallelizes the training of non-linear recurrent architectures across time using Newton-Raphson iterations mapped to associative pre

PerfReasoning: How Well Do LLMs Reason on Hardware Performance?

arXiv:2609.04476v1 Announce Type: new Abstract: Performance modeling is central to hardware design and software optimization, yet constructing these models requires structured reasoning about computat

PetQA: Benchmarking Veterinary Knowledge and Clinical Reasoning

arXiv:2609.04598v1 Announce Type: cross Abstract: We introduce PetQA, a Korean long-form question-answering (QA) benchmark for evaluating veterinary knowledge and clinical reasoning in large language

Phase Transition Frequency as a Training Time Predictor of Test Accuracy in ResNets

arXiv:2609.05194v1 Announce Type: cross Abstract: The number of discrete class-separability jumps observed during ResNet finetuning is examined empirically as a predictor of final test accuracy. Acros

Physics-Aware Random Walk Fingerprints for Scalable Power Grid Graph Classification

arXiv:2609.04943v1 Announce Type: new Abstract: Recent benchmarks such as PowerGraph provide large collections of power-grid graphs for cascading-failure classification. Graph neural networks (GNNs) a

PLUME: Parameter-Efficient Personalization of Large Language Models via Low-Rank User Modulation in Shared Subspaces

arXiv:2609.04715v1 Announce Type: new Abstract: Personalizing large language models (LLMs) is essential for delivering AI assistance that aligns with individual users' styles, intents, and preferences

Post Fusion Bird's Eye View Feature Stabilization for Robust Multimodal 3D Detection

arXiv:2603.05623v2 Announce Type: replace-cross Abstract: Camera-LiDAR fusion is widely used in autonomous driving to enable accurate 3D object detection. However, bird's-eye view (BEV) fusion detecto

PRICE: A Systematic Study of LLM Adaptation Choices for Bitcoin Price Forecasting

arXiv:2609.05235v1 Announce Type: cross Abstract: Cryptocurrency markets exhibit extreme volatility and non-stationary dynamics that challenge conventional forecasting methods. Although Large Language

PRISM-Bench: An Audio-Centric Diagnostic Benchmark for Text-to-Audio-Video Generation

arXiv:2609.04867v1 Announce Type: cross Abstract: Text-to-audio-video (T2AV) generation has advanced rapidly, but its evaluation still underestimates the audio modality. Existing benchmarks either tre

ProtLingo: Efficient Protein Language Modeling via Conditional Memory and Expert Routing

arXiv:2609.04793v1 Announce Type: new Abstract: Proteins perform diverse cellular functions, and even single amino-acid substitutions can alter stability, activity, or molecular interactions. Protein

PuTR-CouT: Counting-by-Tracking in Camera-Trap Image Sequences

arXiv:2609.05038v1 Announce Type: new Abstract: Species identification in camera trap images has been widely studied, but key ecological modeling tasks such as species abundance or density estimation

Quality Recovery for Quantized KV Caches via Low-Rank Attention Adaptation

model-releasesarxiv-cs-cl

arXiv:2609.04263v1 Announce Type: cross Abstract: Low-bit key--value (KV) caches reduce the memory required for autoregressive decoding, but the resulting quality loss depends on the model and quantiz

Quantum-Assisted Memory-Efficient Training for Parameter-Intensive Wi-Fi-Based Human Activity Recognition

arXiv:2609.04271v1 Announce Type: new Abstract: Wi-Fi-based human activity recognition (HAR) has become an important part of integrated sensing and communications, paving the way for a range of contex

Qwen 3.8 27B run just 0.7tok/s on 5060ti 16GB

I just tried to run Qwen 3.8 27B 4bit qwant on 5060ti 16GB and got terrible results. With Qwen 3.6 I got something around 15-20 tok/s. Are there any combination of parameters or tweeks how to push the

model-releasesr-ollama

Qwen 3.8 Next Flash is really really REALLY verbose..

Long time user of 3.6 27b, switched over to Next Flash since it's a logical step up even from 3.8 27b. It's soooo verbose, i'm talking 13 minutes of thinking time on single turn coding requests at app

Qwen3.5 0.8B on CPU

Since the Qwen3.5 0.8B model is an interesting one for small specialized fine tunes, I was curious how fast it can run on CPUs. Why CPUs? Mainly because I want to use it as a local dictation cleanup m

Qwen3.8-27B on M1 Max 32GB: MLX 15.8 tok/s vs llama.cpp 9.7 tok/s - but llama.cpp prefill is faster

I’ve been setting up an M1 Max Mac Studio (24-core GPU, 32GB unified memory) as a local LLM server and wanted to compare MLX vs llama.cpp on Qwen3.8-27B. I tried to keep the model footprint and benchm

model-releasesr-ollama

Radeon RX 7900 GRE 16GB and RX 480 8GB Vulkan benchmarks llama.cpp

My XFX Radeon RX 7900 GRE 16GB Vram GPU struggles with models over 20B size. I added my Radeon RX 480 8GB Vram GPU to the system and ran a few benchmarks using llama.cpp Ubuntu Vulkan build 10453. htt

Real-World Multi-Modal and Longitudinal Lung Cancer Dataset

arXiv:2609.05202v1 Announce Type: cross Abstract: Multi-modal learning has demonstrated strong potential in medical applications by integrating heterogeneous data sources such as medical imaging, clin

Really strong benchmark paper on coding agents. Claude Opus 5 running under Claude Code passes 23.9% of the evaluations. An expert human ref…

Really strong benchmark paper on coding agents. Claude Opus 5 running under Claude Code passes 23.9% of the evaluations. An expert human reference scores 82.2%. Here is what the task actually is. The

model-releasesdair-ai--x