Pith Number

pith:C2JAM3NK

pith:2026:C2JAM3NK5YGL7RX6FLMB73WJOI

not attested not anchored not stored refs pending

Salt: Self-Consistent Distribution Matching with Cache-Aware Training for Fast Video Generation

Bingqi Ma, Dailan He, Guanglu Song, Jun Zhang, Xiahong Wang, Xingtong Ge, Yi Zhang, Yu Liu, Yushi Huang

Salt distills video models to 2-4 steps by regularizing the endpoint consistency of consecutive denoising updates and conditioning on KV cache states.

arxiv:2604.03118 v2 · 2026-04-03 · cs.CV · eess.IV

Open paper page JSON Open Graph Bundle Merged state Verified badge What is a Pith Number?

Add to your LaTeX paper

\usepackage{pith}
\pithnumber{C2JAM3NK5YGL7RX6FLMB73WJOI}

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

Across extensive experiments on both non-autoregressive backbones (e.g., Wan 2.1) and autoregressive real-time paradigms (e.g., Self Forcing), our method, dubbed Salt, consistently improves low-NFE video generation quality while remaining compatible with diverse KV-cache memory mechanisms.

C2weakest assumption

That explicitly regularizing the endpoint-consistent composition of consecutive denoising updates will prevent drift in composed rollouts, and that the cache-conditioned feature alignment will reliably steer low-quality outputs toward high-quality references without introducing new inconsistencies.

C3one line summary

Salt improves low-step video generation quality by adding endpoint-consistent regularization to distribution matching distillation and using cache-conditioned feature alignment for autoregressive models.

Formal links

2 machine-checked theorem links

Cited by

2 papers in Pith

PARE: Pruning and Adaptive Routing for Efficient Video Generation

One-Forcing: Towards Stable One-Step Autoregressive Video Generation

Receipt and verification

First computed	2026-07-02T01:17:30.448942Z
Builder	pith-number-builder-2026-05-17-v1
Signature	Pith Ed25519 (`pith-v1-2026-05`) · public key
Schema	pith-number/v1.0

Canonical hash

1692066daaee0cbfc6fe2ad81feec9723902bb4d9e73a88769cb1904398fc75b

Aliases

arxiv: 2604.03118 · arxiv_version: 2604.03118v2 · doi: 10.48550/arxiv.2604.03118 · pith_short_12: C2JAM3NK5YGL · pith_short_16: C2JAM3NK5YGL7RX6 · pith_short_8: C2JAM3NK

Agent API

Resolver JSON Graph JSON Events JSON Schema Signing key

Verify this Pith Number yourself

curl -sH 'Accept: application/ld+json' https://pith.science/pith/C2JAM3NK5YGL7RX6FLMB73WJOI \
  | 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: 1692066daaee0cbfc6fe2ad81feec9723902bb4d9e73a88769cb1904398fc75b

Canonical record JSON

{
  "metadata": {
    "abstract_canon_sha256": "2e5b42066cdece165e571d2c17fce0e5582fb712271a3342f2469b7e8aa7f87c",
    "cross_cats_sorted": [
      "eess.IV"
    ],
    "license": "http://arxiv.org/licenses/nonexclusive-distrib/1.0/",
    "primary_cat": "cs.CV",
    "submitted_at": "2026-04-03T15:43:21Z",
    "title_canon_sha256": "1f81f90fc8f84eee1cad6c816ad45b6b5266f180f735a92289f264f474f1ece6"
  },
  "schema_version": "1.0",
  "source": {
    "id": "2604.03118",
    "kind": "arxiv",
    "version": 2
  }
}