new

Get trending papers in your email inbox once a day!

Get trending papers in your email inbox!

Daily Papers

byAK and the research community

Sep 4

Submitted by

yuntian-deng

Compile by Training: Turning Natural-Language Specifications into Local Neural Functions

UWaterloo

University of Waterloo

Submitted by

taesiri

Terminal-Universe: Turning Agent Trajectories into Scalable Terminal Environments

Qwen

Submitted by

HaoxingChen

LLaDA-Image: Building Strong Image Generators with Fully Open Training Recipes

inclusionAI

2

Submitted by

spongebob0715

Knowing When Not to Reuse: Conditional Experience Transfer in Autonomous LLM Post-Training

·
6 authors

1

Submitted by

ChengsongHuang

Random Attention: Rethinking KV Cache Eviction for Efficient Reasoning

Salesforce

Salesforce AI Research

Submitted by

pb09204048

LatentPress: Context Compression Beyond Text and Vision

·
2 authors

Submitted by

dmayboroda

Why Gated DeltaNet Survives 4-Bit Quantization: NVFP4 W4A4 for the Recurrent Half of a Hybrid 27B LLM

minima-ai

1

Submitted by

hbx

Rethinking On-Policy Distillation of Large Language Models II: One Training Example

Thinking-Space

1

Submitted by

KangLiao

Puffin-World: Scaling a Unified Multimodal Model with Native 3D World States

ACERobotics

2

Submitted by

linjohnss

Scal3R: Learning Efficient Multi-Relative Pose Query for Scalable Online 3D Reconstruction

NYCU

National Yang Ming Chiao Tung University

Submitted by

taesiri

Editable Visual Design

Tencent-Hunyuan

Tencent Hunyuan

Submitted by

starry0929

The Missing Temporal Link: Temporal Context Routing for Script-Driven Audio-Video Generation

·
10 authors

1

Submitted by

quhongyu123

Beyond Retrieval: Progressive Latent Memory Evolution for Streaming Video Understanding

·
10 authors

Submitted by

songtingyu

CORE: Improving Compositional Reasoning in MLLM Embedding via Reranker Distillation

Alibaba-NLP

Submitted by

shubhamrgandhi

DRACO: Fine-Grained Credit Assignment with Dynamic Rubrics for Long-Horizon Agent Training

ibm-research

Submitted by

jihyoung

PACE: Towards Surfacing Hidden Conflicts in User Requests

POSTECH

Pohang University of Science and Technology

1

Submitted by

gyuhyeong-k

RealSWE: A Compositional Evaluation of Coding Agents under Realistic User Requests

skku

Sungkyunkwan University

Submitted by

CodeGoat24

WorldReward: Reward Modeling for Camera-Conditioned World Models

·
16 authors

Submitted by

zouhar

Last Translation Benchmark

·
244 authors

Submitted by

jiangnanhugo

Using Grounded Theory for Agent Behavior Analysis at Scale

Purdue

Purdue University

Submitted by

byeongjun-park

FlashRender: Few-Step Generative Rendering via Camera-Controlled Video MeanFlow

everex

Submitted by

taesiri

Environment Evolution for Terminal Agents

Tencent-Hunyuan

Tencent Hunyuan

Submitted by

taesiri

Principia: Relational Physics Tests for Video Models

·
4 authors

Submitted by

Prakh25

Select, Compress, Reinvest: A Controlled Study of Visual-Token Allocation in Long-Video MLLMs

·
1 authors

Submitted by

HYUDHKIM

Let Confidence Change, Not the Prediction: Prediction-Preserving Repair for Post-hoc Calibration

·
3 authors

Submitted by

rsn243

Percolation Dynamics in Optimization : Variance Cascades and Discrete Scale Invariance

·
2 authors

Submitted by

YaroslavPrytula

QCell: Recombining and Aligning Cell Queries for Overlapping Instance Segmentation

·
3 authors

Submitted by

rzdiversity

Locked at the Entrance, Open Inside: Where RLVR Narrows the Solution Space

UniversityofBirmingham

University of Birmingham