AI radar · Pulse

Radar

The AI stack

Hugging Face models, datasets, and spaces ranked by a likes + downloads heat score. arXiv cs.AI is the research tape. Counts come from public APIs — if a field is missing, it is omitted.

As of Nov 4, 8:38 AM EST · revalidates daily

Sources live · hf-models 24 · hf-datasets 24 · hf-spaces 24 · arxiv 8

Models

Hottest weights

Data

Datasets

Research

arXiv cs.AI

2026-11-03 · Jisoo Kim, Chaehyun Kim, Junsu Kim
DreamX: Video Diffusion Model is a Universal Latent Action Learner
Learning a universal action representation from video requires modeling dynamics that are consistent across different executions of the same action yet invariant to changes in embodiment and appearance. Existing latent action models infer actions from visual transitions, but thei
2026-11-03 · Hao Fei
From Autonomous Agents to Cognitive Symbionts: Toward AI-Augmented Human Intelligence
For a decade, progress in artificial intelligence has been measured by how much a machine can do on its own. Large foundation models now write code and run multistep tasks, and personal agents act on a person's behalf. Yet people still forget what they once knew, miss what they a
2026-11-03 · Austin Tsai, Saurav Nanda, Ganapathy Parthasarathy
Aligning Language Models with General Preferences: An Empirical Study of Direct Nash Optimization
Direct Nash Optimization (DNO) aligns a language model to a general preference function by approximating the Nash equilibrium of a two-player preference game. Its practical variant, DNO-Prct, iterates on-policy sampling, LLM-based rating of the samples together with a teacher's g
2026-11-03 · Luxi He, Pengcheng Jiang, Jifan Zhang
Suppressing Misaligned Behaviors in Language Models, Durably

Repos

Language mix on the GitHub tape

HuggingFaceFW/fineweb-edu

dataset · 1.4k likes · 438k dl

85

07

nvidia/PhysicalAI-Autonomous-Vehicles

dataset · 1.2k likes · 236k dl

82

08

XiaomiMiMo/MiMo-V2.6-RL-oss

dataset · 1.0k likes · 285k dl

82

09

Anthropic/hh-rlhf

dataset · 2.5k likes · 42k dl

81

10

roneneldan/TinyStories

dataset · 1.2k likes · 169k dl

81

11

tatsu-lab/alpaca

dataset · 1.4k likes · 82k dl

79

12

HuggingFaceH4/ultrachat_200k

dataset · 1.2k likes · 90k dl

79

13

OpenAssistant/oasst1

dataset · 1.6k likes · 34k dl

78

14

databricks/databricks-dolly-15k

dataset · 1.4k likes · 49k dl

78

15

ILSVRC/imagenet-1k

dataset · 1.1k likes · 81k dl

78

16

Open-Orca/OpenOrca

dataset · 1.6k likes · 14k dl

75

17

cais/hle

dataset · 1.1k likes · 30k dl

75

18

FreedomIntelligence/medical-o1-reasoning-SFT

dataset · 1.2k likes · 19k dl

74

19

gsdf/EasyNegative

dataset · 1.2k likes · 7.9k dl

71

20

bigcode/the-stack

dataset · 1.1k likes · 6.8k dl

70

21

lmsys/lmsys-chat-1m

dataset · 1.1k likes · 4.3k dl

69

22

togethercomputer/RedPajama-Data-1T

dataset · 1.2k likes · 3.2k dl

68

23

Nerfgun3/bad_prompt

dataset · 992 likes · 4.8k dl

68

24

allenai/dolma

dataset · 1.1k likes · 2.6k dl

67

Apps

Spaces

Undesired model behaviors can emerge or re-emerge at any stage of training. Once observed, how effectively can we suppress them? To study this question, we first construct two model organisms--language models deliberately trained to exhibit a specific misalignment--that differ in

2026-11-03 · Hongyin Luo, Wei Fang, Lijie Yang

Adversarial Speculative Reasoning Improves Long Horizon Agents

Speculative decoding has been treated purely as a systems optimization that speeds up large language model decoding without changing the model behavior. Under this view, the growing effort to train better draft models has been focused solely on acceptance rate. We argue that this

2026-11-03 · Humza Nusrat, Omar Nusrat

Weight-space programming for auditing clinical decision networks

Machine learning models in radiotherapy are commissioned by testing them on cases with known answers, which may not reveal a shortcut that the test cases share with the training data. Frontier language models have recently been reported to read and write the raw weights of small

2026-11-03 · Omar Nusrat

MedPhysBench: A Benchmark for Evaluating Language Models on Medical Physics Calculations and Error Detection

Medical physicists turn a prescription into a delivered dose and check one another's calculations, because one error in a beam calibration or a monitor-unit (MU) calculation can affect every patient treated with it. MedPhysBench has ten seeded generators of calculation items in r

2026-11-03 · Bowen Li, Danqing Zhou, Chuan He

A simple adaptive accelerated optimistic dual averaging method for unconstrained convex optimization in $p$-norm geometry

Efficient preconditioning plays an important role in large-scale first-order optimization. In this paper, we propose an adaptive accelerated optimistic dual averaging ($\text{A}^2$ODA) method for unconstrained convex optimization that adopts low-cost preconditioning based on $p$-

DreamX: Video Diffusion Model is a Universal Latent Action Learner

From Autonomous Agents to Cognitive Symbionts: Toward AI-Augmented Human Intelligence

Aligning Language Models with General Preferences: An Empirical Study of Direct Nash Optimization

Suppressing Misaligned Behaviors in Language Models, Durably

Adversarial Speculative Reasoning Improves Long Horizon Agents

Weight-space programming for auditing clinical decision networks

MedPhysBench: A Benchmark for Evaluating Language Models on Medical Physics Calculations and Error Detection

A simple adaptive accelerated optimistic dual averaging method for unconstrained convex optimization in $p$-norm geometry