Daily Papers of 2026-03-30

1. ShotStream: Streaming Multi-Shot Video Generation for Interactive Storytelling 154 upvotes, #1 of 2026-03-30
2. Out of Sight but Not Out of Mind: Hybrid Memory for Dynamic Video World Models 152 upvotes, #2 of 2026-03-30
3. PackForcing: Short Video Training Suffices for Long Video Sampling and Long Context Inference 50 upvotes, #3 of 2026-03-30
4. Trace2Skill: Distill Trajectory-Local Lessons into Transferable Agent Skills 48 upvotes, #4 of 2026-03-30
5. Sommelier: Scalable Open Multi-turn Audio Pre-processing for Full-duplex Speech Language Models 35 upvotes, #5 of 2026-03-30
6. MedOpenClaw: Auditable Medical Imaging Agents Reasoning over Uncurated Full Studies 31 upvotes, #6 of 2026-03-30
7. RealChart2Code: Advancing Chart-to-Code Generation with Real Data and Multi-Task Evaluation 28 upvotes, #7 of 2026-03-30
8. Natural-Language Agent Harnesses 24 upvotes, #8 of 2026-03-30
9. Know3D: Prompting 3D Generation with Knowledge from Vision-Language Models 19 upvotes, #9 of 2026-03-30
10. LongTail Driving Scenarios with Reasoning Traces: The KITScenes LongTail Dataset 17 upvotes, #10 of 2026-03-30
11. Composer 2 Technical Report 15 upvotes, #11 of 2026-03-30
12. GenMask: Adapting DiT for Segmentation via Direct Mask 10 upvotes, #12 of 2026-03-30
13. Diffutron: A Masked Diffusion Language Model for Turkish Language 8 upvotes, #13 of 2026-03-30
14. Learning to Commit: Generating Organic Pull Requests via Online Repository Memory 8 upvotes, #13 of 2026-03-30
15. Lie to Me: How Faithful Is Chain-of-Thought Reasoning in Reasoning Models? 7 upvotes, #15 of 2026-03-30

Data: hysts-bot-data/daily-papers-stats and the Daily Papers API. Open data: tardellirs/paper-pulse-data. Sister project: Model Pulse, the download history of every model on the Hub.