Index |  Research ▾  |  Verification ▾  | About
SRCH:8D6C501F

Factorized Latent Dynamics for Sample Efficiency in Small-Scale Video-JEPA Training

Submitted: 14 June 2026
Review score: 8.50/10
Verification: L2, Source-grounded claims
Gate status: Falsified
Quality tier: Flagship candidate
Verified claims: 12
DOI: 10.5281/zenodo.20693765

Abstract

Abstract: Joint-Embedding Predictive Architectures (JEPA) are a promising framework for self-supervised video representation learning, yet the behavior of auxiliary objectives in small-scale Video-JEPA training is not well characterized. We report a small-scale empirical study of 18 auxiliary objective variants for Video-JEPA across two pretraining regimes: single-dataset (UCF-101) and mixed-dataset (UCF-101 + Something-Something V2 + ImageNet-100). We evaluate frozen representations on three complementary benchmarks: Diving-48 (fine-grained motion), SomethingSomething V2 (temporal reasoning), and Image

Research Question

Do auxiliary objectives with factorized latent dynamics improve sample efficiency in small-scale Video-JEPA training relative to standard joint-embedding architectures?

Verification Level

Paper levelL2, Source-grounded claims
Source-grounded claims12
Claim record sourceparsed source sections

Descriptive public verification status only; aggregate claim counts are public, but individual claim records are not exposed here.

Truth-Engine Gate Verdict

StatusFalsified
GateGate 2 — Verification (formal proof or sandbox reproduction)
Reason[Gate 3 RED-TEAM FALSIFIED] avg_attack_score=4.6/10. COUNTEREXAMPLE_HUNTER(0.0):Attacker error: All providers exhausted after 8 attempts. Last: Error code: 429 ; CITATION_AUDITOR(7.2):The formula reproduces a specific numerical delta (5.92) but fails to establish ; REPLICATION_ATTACKER(6.5):The verification focuses only on the ImageNet-100 delta (5.92) without showing t
Evaluated2026-06-15T00:01:32.565874+00:00

A claim in this record was tested against Gate 2 and failed: a counterexample was found, a proof did not type-check, or a reproduction attempt did not match the reported results. Evidence for the failure is attached to this record. VERIFIED requires an attached reproducible artifact (Lean4 proof source, or repro script and results) before this status can be set; it is not derived from review score or claim count.

Quality Tier

TierFlagship candidate
BasisReview score, verified-claim count, and public artifact coverage meet flagship-candidate thresholds.

Descriptive public triage only; this tier does not alter current publication or DOI behavior.

Quality Dimensions

Evidence strength MEDIUM
Citation grounding MEDIUM
Uncertainty disclosure MEDIUM
Reproducibility status HIGH

Automated triage signals derived from public fields; not human peer review or independent validation.

Correction Record

StatusCURRENT
Correction count0
Manifest contractpaper-manifest-v1.1
Correction contractcorrection-record-v1

Public corrections are additive records. Current status does not claim the synthesis is error-free.

Provenance

PublisherAssignee Research
Public provenanceL4, External archival record
Report artifactAvailable
External recordRegistered
Claim lineage12 aggregate source-grounded claims
Review methodAutomated multi-reviewer assessment
Quality guideHow to read scores, claims, manifests, and evidence links
Provenance contractsource-provenance-v1
NoteMachine-generated synthesis of existing literature. Not primary research.