model releases

Learning-Augmented and Randomized Algorithms for Line Aggregation with Delays

arXiv:2607.27807v1 Announce Type: new Abstract: This paper studies learning-augmented and randomized online aggregation with delays on a line metric. We consider advice given as online suggested servi

Model Releases

Learning Color Grading, No Photo Sharing: Federated Aesthetic Preference Learning for Personalized Image Enhancement

DGX agent

arXiv:2607.27659v1 Announce Type: new Abstract: Personalized image enhancement should reflect individual aesthetic taste, yet learning such preferences commonly depends on private photos and ratings t

model-releasesarxiv-cs-cv

31 Jul 2026

Learning features from Newton's algorithm: a way to accelerate nonlinear parametrized PDE solvers

arXiv:2607.28036v1 Announce Type: new Abstract: It is well known that Newton's method converges faster when the initial guess is closer to a root of a system of nonlinear equations. In this paper, a t

Learning path to fully understand the Kimi K3 technical report?

Hi everyone, Can anyone suggest a learning path to fully understand the technical report for Kimi K3? My background: • I've taken a graduate-level deep learning course. • I understand the Transformer

model-releasesr-ollama

Learning to Trace Seiberg Dualities

arXiv:2607.28628v1 Announce Type: cross Abstract: Dualities play an important role in establishing both microscopic and emergent phenomena in a wide range of physical systems. In practice, though, it

Lightning OPD 2.0: Mitigating Style Bias in Cross-Teacher On-Policy Distillation for Large Reasoning Models

arXiv:2607.28449v1 Announce Type: new Abstract: On-policy distillation (OPD) provides dense token-level supervision from a teacher, but its effectiveness can depend on teacher consistency, meaning tha

LLM-Guided Initialization for Accelerated Hybrid Quantum-Classical Medical Image Classification

arXiv:2607.27262v1 Announce Type: cross Abstract: Variational quantum algorithms often encounter barren plateaus, where cost gradients decay rapidly with increasing circuit depth, undermining the trai

LLM Self-Correction with DeCRIM: Decompose, Critique, and Refine for Enhanced Following of Instructions with Multiple Constraints

arXiv:2410.06458v2 Announce Type: replace Abstract: Instruction following is a key capability for LLMs. However, recent studies have shown that LLMs often struggle with instructions containing multipl

LLM2Vec-Gen: Generative Embeddings from Large Language Models

arXiv:2603.10913v3 Announce Type: replace Abstract: Fine-tuning LLM-based text embedders via contrastive learning maps inputs and outputs into a new representational space, discarding the LLM's output

LLMs struggle to simulate human belief updates in controlled environments

arXiv:2607.28347v1 Announce Type: new Abstract: LLMs are increasingly deployed as proxies for human study participants in social science experiments, yet the fidelity of this practice has rarely been

LM-GRASP: Instance-Specific Language Models for Combinatorial Construction via Online Imitation Learning

arXiv:2607.28135v1 Announce Type: new Abstract: Machine learning for combinatorial optimization typically relies on neural constructors trained via reinforcement learning on large offline datasets for

LoMeVQA: A Comprehensive Benchmark for Longitudinal Medical VQA

arXiv:2607.27806v1 Announce Type: new Abstract: In clinical practice, patients often undergo multiple imaging examinations over successive visits, yielding longitudinal data. Modeling such temporal in

Looped Transformers with Source-Centered State Evolution

arXiv:2607.27656v1 Announce Type: cross Abstract: Looped Transformers create a useful train- and test-time compute axis by reusing the same Transformer block over recurrent depth, increasing effective

LoRA Scaffolded Policy Optimization (LSPO): A Sampling-Time Low-Rank Scaffold for Recovering Reinforcement-Learning Gradient on Zero-Reward Cliff Prompts

arXiv:2607.27787v1 Announce Type: new Abstract: Reinforcement learning from verifiable rewards (RLVR) for mathematical reasoning suffers from a structural blind spot: on 'cliff' prompts-those on which

MagicSelector: Joint Optimization for Agent Tool Selection via Counterfactual Decomposition and Progressive Reranking

arXiv:2607.17751v2 Announce Type: cross Abstract: We present MagicSelector, a joint optimization framework integrating Counterfactual task decomposition, Progressive reranking, and Dynamic Top-K, desi

MatCreatioNN: Machine learning-guided computational discovery of photocatalysts for environmental applications

arXiv:2607.27295v1 Announce Type: cross Abstract: The rational design of photocatalysts for environmental remediation and CO2 conversion remains limited by the high computational cost and sparse exper

Measuring Alignment With Reader Highlights Net of Position and Length

arXiv:2607.27739v1 Announce Type: cross Abstract: Context compression discards most of a document before a language model reads it, and is normally evaluated by downstream task accuracy - which makes

MedHallTune: An Instruction-Tuning Benchmark for Mitigating Medical Hallucination in Vision-Language Models

arXiv:2502.20780v2 Announce Type: replace-cross Abstract: The increasing use of vision-language models (VLMs) in healthcare applications presents great challenges related to hallucinations, in which t

Meituan just dropped LongCat-Flash-Lite-Sparse

It’s an MoE with ~3B active params and a 30B n-gram lookup table offloaded to RAM for fast 256k context on a 24GB GPU. Reminds me of Gemma 4’s PLE trick. Initial analysis suggest it wont be replacing

Memory Decoder at Scale: A Pretrained, Parametric Long-Term Memory

arXiv:2607.27919v1 Announce Type: new Abstract: Decoder-only language models entangle long-term memory and reasoning in a single parameter set, making it difficult to scale memory capacity independent

MemSecBench: Tracking Agent Memory Poisoning from Persistence to Consequence and Repair

model-releasesarxiv-cs-ai

arXiv:2607.27080v1 Announce Type: cross Abstract: Memory systems allow agents to retain and reuse information from past interactions, but they can also let malicious content persist. A malicious instr

MiniMax H3 discussion

model-releasesr-stablediffusion

https://x.com/MiniMax_AI/status/2083008095488516262 (says it's coming in a few days) it can do text-to-image, and image editing everyone seems to be mostly hyped about the video generation part (I am

MiniMax H3: Open-weight multimodel video model

model-releasesr-stablediffusion

Just saw this posted by Fal.ai and then by Hailuo themselves, the next video model will be open weight released! Here's the blurb and link to to the feature post: Today, we're launching MiniMax H3, a

Minimax-H3 video model released, open weights coming in the next few days

https://x.com/MiniMax_AI/status/2083006198828417501?s=20 Quote from their article: Today, we're launching MiniMax H3, a general-purpose multimodal generation model. H3 understands unified context acro

Minimum VRAM GPU to run DeepSeek-V4-Flash-0731 Q4_K_XL at around 30 t/s ?

Hello guys, I'm curious about running DeepSeek-V4-Flash-0731 locally. Since it’s a Mixture of Experts (MoE) model with only 13B active parameters, I was hoping the VRAM requirements might be manageabl

MMAC: A Massive Multi-dimensional Benchmark for Audio Captioning

model-releasesarxiv-cs-ai

arXiv:2607.27109v2 Announce Type: cross Abstract: With the development of audio large language models (AudioLLMs), audio captioning needs to move from brief descriptions toward open-ended and fine-gra

MMHBench: A Multi-Perspective Benchmark for Mental Health Understanding in Long-Form Videos

arXiv:2607.27895v1 Announce Type: cross Abstract: Mental health understanding in long-form videos requires nuanced reasoning over observable behavior, interpersonal context, and latent psychological s

MMOOC: A Comprehensive Benchmark for Out-of-Context Evaluation in Multimodal Large Language Models

arXiv:2607.27637v1 Announce Type: new Abstract: Multimodal Large Language Models (MLLMs) have achieved strong performance on a wide range of vision-language tasks, but often fail under imperfect or sh

Models for minimalist RAG: B1ade 335M Embedding and 1B Parameter Small Language Models

arXiv:2607.27506v1 Announce Type: new Abstract: Language and embedding models used in RAG systems are conventionally assumed to require large-scale pretraining and explicit grounding supervision. We p

MOON2.0: Dynamic Modality-balanced Multimodal Representation Learning for E-commerce Product Understanding

arXiv:2511.12449v3 Announce Type: replace Abstract: Recent Multimodal Large Language Models (MLLMs) have significantly advanced e-commerce product understanding. However, they still face three challen

MORFES: A Benchmark for Productive Inflectional Competence in Modern Greek

arXiv:2607.28274v1 Announce Type: new Abstract: Modern Greek is a richly inflected language, yet the language models built for it are evaluated mainly on factual knowledge, and no benchmark is dedicat

MPIE-Bench: Benchmarking Anatomically Plausible Multi-Person Interaction Editing

arXiv:2607.27616v1 Announce Type: new Abstract: Text-to-image and personalized editing models now synthesize high-fidelity single-subject images with ease. Yet placing multiple named people into share

MSCM-net: A hyperspectral image classiffcation method based on multi-scale convolution and Mamba

arXiv:2607.28277v1 Announce Type: new Abstract: Hyperspectral imaging is widely used in remote sensing and engineering. Therefore, research on its classification methods is crucial. While CNN and Tran

MultivationBench: A Benchmark for Multimodal Sequential Motivation Reasoning

model-releasesarxiv-cs-ai

arXiv:2607.26465v1 Announce Type: new Abstract: Multimodal Large Language Models have sparked significant interest due to their potential for social intelligence; however, their ability to perform seq

Neat work on long-horizon agents. Splitting a hard task across agents is typically how standard multi-agent work. The usual design lets them…

Neat work on long-horizon agents. Splitting a hard task across agents is typically how standard multi-agent work. The usual design lets them exchange findings only at phase boundaries, through staged

model-releasesdair-ai--x

Neural Network-Assisted CLEAN for Channel Modeling in Low-SNR Regimes

arXiv:2607.27450v1 Announce Type: new Abstract: Accurate multipath parameter estimation is critical for modern wireless communication systems, particularly in challenging low-SNR environments. Traditi

New research from Microsoft. This one is on training computer-use agents at scale. Recent pipelines generate synthetic environments in bulk,…

New research from Microsoft. This one is on training computer-use agents at scale. Recent pipelines generate synthetic environments in bulk, which moved the bottleneck from how many exist to what is i

model-releasesdair-ai--x

Now Suddenly too many choices for DGX Spark with Qwen 3.5 122B . What would be the next upgrade?

Laguna 2.1 at NVFP4 Deepseek v4 at Q2 Inkling-Small at IQ3 Which models you guys running now ? How it compares to 122b? Upcoming in few days : Ling 3.0 124B (Could be new king) LongCat 69B A3B ( very

Objective-Aligned Direct Answer SFT for Robust Multi-Frame Medical VQA

arXiv:2607.27566v1 Announce Type: new Abstract: Multi-frame medical VQA appears to reward increasingly complex adaptation: controller-style inference, localization-aware reranking, static hard-negativ

OK, GPT-5.6 Luna is a bit of a beast. Given the 80% price drop today I decided to try it in Datasette Agent, and it's furiously quick and ge…

model-releasessimon-willison--x

OK, GPT-5.6 Luna is a bit of a beast. Given the 80% price drop today I decided to try it in Datasette Agent, and it's furiously quick and generates all the SQL, HTML and JavaScript (for Datasette Apps

On a joint simultaneous learning of relevant feature subsets and subspaces in regression-like problems

arXiv:2607.28080v1 Announce Type: cross Abstract: We extend a recently introduced Entropy-Optimal Manifold Clustering (EOMC) to allow for a joint simultaneous identification of subsets and subspaces o

One Patch Is Enough: Reinforcement-Optimized Visual Token Grounding for MLLM-Based Scene Text Spotting

arXiv:2607.27902v1 Announce Type: new Abstract: Scene text spotting requires high-precision alignment between textual recognition and spatial localization. While visual-token grounding has emerged as

Open Source Ternary LLM Engine in Rust/CUDA for Quantization, Serving, and Training of models on consumer GPUs, called Tritium (Apache 2.0)

This post was not written by a clanker. Hey guys, I'm a comp sci major who wanted to introduce a cool project I built for quantizing models to ternary (1.58 bit) with as minimal of loss as possible, a

OpenAI’s entire growth loop: new models, price cuts, and Tibo’s token resets

model-releasesjerry-liu--x

OpenAI’s entire growth loop: new models, price cuts, and Tibo’s token resets major price cuts today: *80% drop for GPT-5.6 Luna, now 0.20 per million input tokens and 1.20 per million output *20% drop

Optimal Realistic Local AI for Most

So you’ve got a 3090 or maybe even a 5090? Or more likely a 4060 8GB Ti. You wanna try local AI, you don’t know what it can/can’t do. 1) Install the best model you can. If you have a 3090 or a 5090, t

Oracle-Budgeted Molecular Optimization with Short-Term Graph Memory

arXiv:2607.28437v1 Announce Type: new Abstract: Molecular optimization is commonly performed under a limited oracle budget, which makes deciding what to evaluate as important as deciding what to gener

ORCA-bench: How Ready Are Language Model Agents for Oncall?

arXiv:2607.28545v1 Announce Type: new Abstract: Large language models can write, patch, and search code, but oncall root cause analysis (RCA) demands something different: reasoning over noisy metrics,

OSReward: Instituting Standardized Evaluation for Cross-Platform Computer-Use Reward Models

arXiv:2607.28609v1 Announce Type: cross Abstract: Computer-using agents (CUAs) are advancing rapidly across the digital world. A CUA trajectory records the agent's actions, states, and reasoning. Veri