ML Newsnew | past | best | rss | submit about
Stories from June 23, 2026 (UTC)
Go back a day, month, or year. Go forward a day.
1. VibeThinker: 3B param model that beats Opus 4.5 on reasoning with novel SFT+GRPO (arxiv.org)
398 points by timhigins 103 days ago | hide | past | pdf | 205 comments
2. Ultralytics YOLO26: Unified Real-Time End-to-End Vision Models (arxiv.org)
55 points by teleforce 103 days ago | hide | past | pdf | 9 comments
3. The Reversal Curse: LLMs trained on "A is B" fail to learn "B is A" (2023) (arxiv.org)
25 points by Anon84 103 days ago | hide | past | pdf | 46 comments
4. Lift4D: Harmonizing Single-View 3D Estimation for 4D Reconstruction In-the-Wild (arxiv.org)
3 points by MediaSquirrel 102 days ago | hide | past | pdf | discuss
5. Confidence estimation is a better metric than agreement for LLM judges (arxiv.org)
3 points by rapiddev 102 days ago | hide | past | pdf | discuss
6. PsychoPass: Geometric Profiling of Multi-Turn Adversarial LLM Conversations (arxiv.org)
3 points by Anon84 103 days ago | hide | past | pdf | discuss
7. Unlimited OCR Works (arxiv.org)
3 points by ilreb 103 days ago | hide | past | pdf | discuss
8. Tapered Language Models (arxiv.org)
3 points by E-Reverance 103 days ago | hide | past | pdf | discuss
9. Inference Compute Shapes Frontier LLM Evaluation (arxiv.org)
2 points by matt_d 102 days ago | hide | past | pdf | discuss
10. Concordia: JIT-Compiled Persistent-Kernel Checkpt for Fault-Tolerant Inference (arxiv.org)
2 points by matt_d 102 days ago | hide | past | pdf | discuss
About | RSS | RSS (all) | HN arXiv