pith. sign in
Pith Number

pith:DUNYXM5B

pith:2025:DUNYXM5BA4XJRJQEDWYLO2E7N7
not attested not anchored not stored refs resolved

Reward Forcing: Efficient Streaming Video Generation with Rewarded Distribution Matching Distillation

Haobo Li, Hao Ouyang, Hengyuan Cao, Jiapeng Zhu, Ka Leong Cheng, Min Zhang, Qiuyu Wang, Xing Zhu, Yanhong Zeng, Yujun Shen, Yunhong Lu, Zhipeng Zhang

EMA-updated sink tokens and reward-weighted distillation fix copied frames and weak motion in streaming video models.

arxiv:2512.04678 v2 · 2025-12-04 · cs.CV

Add to your LaTeX paper
\usepackage{pith}
\pithnumber{DUNYXM5BA4XJRJQEDWYLO2E7N7}

Prints a linked badge after your title and injects PDF metadata. Compiles on arXiv. Learn more · Embed verified badge

Record completeness

1 Bitcoin timestamp
2 Internet Archive
3 Author claim open · sign in to claim
4 Citations open
5 Replications open
Portable graph bundle live · download bundle · merged state
The bundle contains the canonical record plus signed events. A mirror can host it anywhere and recompute the same current state with the deterministic merge algorithm.

Claims

C1strongest claim

Reward Forcing achieves state-of-the-art performance on standard benchmarks while enabling high-quality streaming video generation at 23.1 FPS on a single H100 GPU.

C2weakest assumption

The vision-language model used to rate motion dynamics provides an accurate and unbiased signal that genuinely improves the distilled model's motion quality without introducing new artifacts or distribution shifts.

C3one line summary

Reward Forcing combines EMA-Sink tokens and Rewarded Distribution Matching Distillation to deliver state-of-the-art streaming video generation at 23.1 FPS without copying initial frames.

References

99 extracted · 99 resolved · 33 Pith anchors

[1] Qwen2.5-VL Technical Report 2025 · arXiv:2502.13923
[2] Philip J. Ball, Jakob Bauer, Frank Belletti, Bethanie Brown- field, Ariel Ephrat, Shlomi Fruchter, Agrim Gupta, Kris- tian Holsheimer, Aleksander Holynski, Jiri Hron, Christos Kaplanis, Marjorie Limon 2025
[3] Stable Video Diffusion: Scaling Latent Video Diffusion Models to Large Datasets 2023 · arXiv:2311.15127
[4] Video generation models as world simulators
[5] Dimension-reduction attack! video generative models are experts on controllable image synthesis.arXiv preprint arXiv:2505.23325, 2025a 2025

Formal links

3 machine-checked theorem links

Cited by

23 papers in Pith

Receipt and verification
First computed 2026-05-17T23:38:47.001172Z
Builder pith-number-builder-2026-05-17-v1
Signature Pith Ed25519 (pith-v1-2026-05) · public key
Schema pith-number/v1.0

Canonical hash

1d1b8bb3a1072e98a6041db0b7689f6ff1667d284b163bd00c2aa568ba216026

Aliases

arxiv: 2512.04678 · arxiv_version: 2512.04678v2 · doi: 10.48550/arxiv.2512.04678 · pith_short_12: DUNYXM5BA4XJ · pith_short_16: DUNYXM5BA4XJRJQE · pith_short_8: DUNYXM5B
Agent API
Verify this Pith Number yourself
curl -sH 'Accept: application/ld+json' https://pith.science/pith/DUNYXM5BA4XJRJQEDWYLO2E7N7 \
  | jq -c '.canonical_record' \
  | python3 -c "import sys,json,hashlib; b=json.dumps(json.loads(sys.stdin.read()), sort_keys=True, separators=(',',':'), ensure_ascii=False).encode(); print(hashlib.sha256(b).hexdigest())"
# expect: 1d1b8bb3a1072e98a6041db0b7689f6ff1667d284b163bd00c2aa568ba216026
Canonical record JSON
{
  "metadata": {
    "abstract_canon_sha256": "ec260685864428546f9219f556523d3969b00abb28dd89bf229c586abfeab891",
    "cross_cats_sorted": [],
    "license": "http://arxiv.org/licenses/nonexclusive-distrib/1.0/",
    "primary_cat": "cs.CV",
    "submitted_at": "2025-12-04T11:12:13Z",
    "title_canon_sha256": "e7ca40a652a7ac5d695737b9deeff6537d1687485008487091d80ac55bf5c989"
  },
  "schema_version": "1.0",
  "source": {
    "id": "2512.04678",
    "kind": "arxiv",
    "version": 2
  }
}