Fields

#	Field	Papers	Repos	Stars	Change	Status	Driving it
1	Multimodal	499	—	—	−33.8%	Cooling	REKI: Registering Knowledge in VLAs
2	Agents	497	—	—	−0.6%	Steady	MLExpert: Towards a Generalist Agentic Model for Autonomous Machine Learning Engineering
3	Frontier LLMs	430	—	—	−7.5%	Steady	Data Strategies for Large Language Models: A Unified Perspective on the Training Lifecycle
4	Inference & serving	247	—	—	−18.5%	Cooling	When the KV Cache Is Slower Than the GPU: Scaling Remote KV Tiers for Context-Parallel Prefill
5	Learning verifiable rewards	165	—	—	−17.5%	Cooling	KnowGPO: Knowledge-Grouped Policy Optimization for Search Agents
6	Image segmentation	164	—	—	−21.5%	Cooling	MedPAO-VL: Interpretable and Protocol-Compliant Agent for Structured Radiology Report Generation
7	Graph learning	158	—	—	−30.4%	Cooling	Your Transformer Is Secretly Elastic
8	3D scene	142	—	—	−24.9%	Cooling	Gen4P: Video Generators are 4D Perceivers
9	Evals	108	—	—	new	Steady	Realtime-Bench: Can Frontier Models Solve Real-time Interactive Problems?
10	Voice AI	88	—	—	−25.4%	Cooling	Realtime-Bench: Can Frontier Models Solve Real-time Interactive Problems?
11	Classifier & causal discovery	77	—	—	−28.7%	Cooling	Are You Sure Your LLM Is Sure? Repairing and Auditing Confidence
12	Flow matching	76	—	—	−54.2%	Cooling	On the Stability and Convergence of Flow Map Training
13	Forecasting time	75	—	—	−17.6%	Cooling	PathFM: Fast Generative Forecasting of Multivariate Trajectories
14	Training & fine-tuning	64	—	—	new	Steady	MLExpert: Towards a Generalist Agentic Model for Autonomous Machine Learning Engineering
15	Teacher policy distillation	56	—	—	−24.3%	Cooling	Training-Free On-Policy Distillation
16	Computer vision	51	—	—	new	Steady	Gen4P: Video Generators are 4D Perceivers
17	AI safety	47	—	—	new	Steady	When Reasoning Is Not Enough: Proactive Inquiry for Cross-Modal Safety in Multimodal Large Language Models
18	Stochastic gradient descent	42	—	—	−26.3%	Cooling	Generalization Guarantees of Randomized Block Coordinate Descent for Nonconvex Learning
19	Machine learned interatomic	41	—	—	−46.1%	Cooling	MolLean: A Process-Verifiable Language for Chemical Reasoning
20	Federated learning	40	—	—	+8.1%	Steady	Private Retrieval Augmented Generation via Random Projection
21	Image & video generation	39	—	—	new	Steady	FeaTok: Feature-Loss Tokenization for Image Generation
22	Adversarial training	35	—	—	−47.0%	Cooling	Beyond Generation: Using Multimodal LLMs as Detectors of LLM-Generated Deepfake Facial Images
23	Robotics	32	—	—	new	Steady	REKI: Registering Knowledge in VLAs
24	Code models	24	—	—	new	Steady	WEAVE: Weaving Real-World Code Evolution into Dependency-Structured Tasks for Benchmarking and Training Code Agents
25	Anomaly detectors	23	—	—	−45.2%	Cooling	Retrieval Over Training: Similarity search-based Model Selection for Time Series Anomaly Detection
26	Quantum machine learning	23	—	—	−23.3%	Cooling	Discovering state-of-the-art quantum circuits with large language models
27	Sensing communication	16	—	—	+23.1%	Rising	Scene Reconstruction Using Ubiquitous Wireless Signals via Diffusion Posterior Inference Over physics-Grounded Representations
28	CAD models	15	—	—	−31.8%	Cooling	3DCodeVerse: Exploring 3D Worlds as Code
29	Character recognition OCR	15	—	—	−21.1%	Cooling	Visual Sanad: Auditing Vision-Language Models for False Acceptance of Qur'anic Citations
30	RAG	15	—	—	new	Steady	Latent Compaction with Test-Time Training
31	Small LLMs	12	—	—	new	Steady	How to Train Your Little Dragon: Teaching a Small Language Model to Construct and Refine GPU Optimization Scripts
32	Efficient spiking neural	2	—	—	−85.7%	Cooling	The Dormant Spiking Neuron: A State-Driven Mechanism for Efficient Spiking Neural Networks
33	Vector databases	0	—	—	new	Steady	—

1

Multimodal

499

—

−33.8%

Cooling

REKI: Registering Knowledge in VLAs

2

Agents

497

—

−0.6%