Paper Tracking

An automated weekly digest. Every Monday an agent scans arxiv and HuggingFace Daily Papers, downloads the PDFs, and summarizes the top 3 picks per topic.

Week of August 17, 2026 · View archive →

Graph × LLM

Graph-structured reasoning, graph foundation models, and LLM-augmented GNNs.

R4DSG: Relative 4D Scene Graph Memory for Object-Centric Question Answering in Long Egocentric Video

↑ 0 📚 45123 ★ 0 Aug 11

Problem: Object-centric question answering in long egocentric video requires preserving persistent object identity and structured spatial change, which caption-based methods fail to capture.
Model: R4DSG: Relative 4D scene graph memory that indexes video by time, place, persistent objects, and anchor-relative transitions without requiring global world coordinates.
Code: https://dualtransparency.github.io/R4DSG/

ProTAGAD: A Foundation Model for TAG Anomaly Detection with Decoupled Topological and Textual Prototypes

↑ 0 📚 15967 ★ 0 Aug 11

Problem: Text-Attributed Graphs (TAGs), endowed with abundant textual content along with topological structures, have emerged as a versatile backbone for real-world anom
Model:
Code: not released

ConnectionMind: Leveraging Social Networks and Large Language Models for Personalized Recommendation at Meta

↑ 0 ★ 0 Aug 10

Problem: Recommendation systems lack reasoning capability to integrate multi-relational social context for fine-grained personalization at scale.
Model: "ConnectionMind: LLM-guided reasoning policy over heterogeneous social graphs, trained with SFT then end-to-end RL for path-based recommendation discovery"
Code: not released

Vision-Language Models

Multimodal VLMs, vision encoders, and image-text foundation models.

Spatial Memory Agent: Experience-Grounded Procedure Memory for Spatial Intelligence

↑ 41 📚 116435 ★ 0 Aug 12

Problem: How can frozen VLMs improve spatial reasoning through parameter-update-free self-evolution without external spatial tools at inference time?
Model: "Spatial Memory Agent (SMA): experience-grounded runtime framework converting verified spatial experience into reusable transferable lessons with Transfer Reliability Score calibration"
Code: not released

Intern-S2-Preview: Scientific Agentic Foundation Model

↑ 56 ★ 0 Aug 12

Problem: AI systems need to reason over heterogeneous scientific modalities, interact with tools, and sustain progress over long task horizons for scientific discovery.
Model: "Intern-S2-Preview-397B": multimodal scientific foundation model with time series forecasting, agentic RL training, and modular Memory Decoder for domain specialization.
Code: InternLM/xtuner

AdvFD: Boosting Visual Generation via Adversarial Fr'echet Distance Loss

↑ 26 📚 9561 ★ 0 Aug 10

Problem: Static pretrained feature spaces used in Fréchet distance losses enable Fréchet hacking, where optimized metrics improve while visual quality degrades.
Model: "AdvFD: Adversarial Fréchet Distance" — combines static pretrained encoders with an adversarially learned adaptive representation and real-feature whitening to expose residual distribution mismatches during generator post-training.
Code: not released

World Models

Predictive world models, JEPA-style learning, video generation, and embodied simulators.

Alaya-EVOKE: From Linear-Scaling Supervision to Endless World

↑ 119 📚 14838 ★ 0 Aug 13

Problem: Interactive world models struggle to balance persistent memory, low-latency response, and long-horizon generation due to conflicting computational demands.
Model: "Evoke": world model with externalized scene geometry bank and linear-attention teacher for long-horizon supervision enabling few-step student generation.
Code: not released
⚠ Interested, but agent could not fetch the PDF — summary based on abstract only.

DreamX-Phi 1.0: Action-Conditioned Video World Model for Robotic Manipulation

↑ 93 📚 16421 ★ 0 Aug 13

Problem: Action-conditioned video world models struggle to faithfully predict robot manipulation while preserving arm identity, object state, and scene geometry.
Model: DreamX-Phi 1.0: action-conditioned video world model using PRoPE-style SE(3) geometric encoding, depth supervision, SAM3 masks, and V-JEPA teacher for faithful robotic manipulation prediction.
Code: AMAP-ML/DreamX-Phi

PlayWorld: Benchmarking World Models with Agent Players over Long-Horizon Objectives

↑ 43 📚 37560 ★ 0 Aug 13

Problem: Existing world model benchmarks use fixed action sequences that fail to enable fair comparison across models with different action granularities and response dynamics.
Model: PlayWorld: benchmark with multi-modal Agent Player that adaptively executes long-horizon objectives to evaluate world models on geometry consistency, interaction fidelity, and state evolution.
Code: kxding/PlayWorld

Spatial Single-Cell Study

Single-cell + spatial omics foundation models, virtual cells, biological world models, and biomedical digital twins.

Disentangled Shared Representations Improve Morpho-Transcriptomic Integration

↑ 0 ★ 0 Aug 14

Problem: Standard multimodal models for spatial transcriptomics compress H&E and gene expression into shared latent space without explicitly separating shared versus modality-specific variation.
Model: approach: Disentangled multimodal representation learning comparing VAE-based (MMVAE+, MMVAE+sg) and contrastive (disSSL) models that factorize shared and private latent components for H&E and spatial transcriptomics integration.
Code: not released

Program-space Diffusion for Morphology-to-Transcriptomics Prediction