SeonghoonYu/Masking-KD-Rollouts
Masking-KD Rollouts Training data for Hide to See: Reasoning-prefix Masking for Visual-anchored Thinking in VLM Distillation (NeurIPS 2026). Paper: https://arxiv.org/abs/2605.11651 Code: https://github.com/Seonghoon-Yu/Masking-KD Model: SeonghoonYu/Masking-KD How it was built Questions: ViRL39K, using the processed version PAPOGalaxy/PAPO_ViRL39K_train. Teacher: Qwen3-VL-8B-Thinking, greedy decoding, up to 4096 new tokens. Prompt: the question followed by the… See the full description on the dataset page: https://huggingface.co/datasets/SeonghoonYu/Masking-KD-Rollouts.
| Author / organization | SeonghoonYu |
|---|---|
| License (as declared) | mit |
| Downloads | 175 (as reported by Hugging Face) |
| Likes | 0 |
| Created | 5 October 2026 |
| Last updated at source | 5 October 2026 |
Popularity, last 3 days
175 now (+3 since 2026-10-16). Counts as reported by the source, snapshotted daily by Silytics.
Connected in the knowledge graph
Related papers
More from SeonghoonYu
Silytics shows public metadata from Hugging Face and links to the original. Information may be incomplete or out of date. Last refreshed 18 October 2026. Licenses and terms of the original apply to the work itself. Not financial, legal or professional advice.