INTACT · SADL shared-BatchNorm

This research variant applies SADL to the INTACT latent world model with shared batch normalization. The one-epoch releases cover joint and separate training on PushT, Cube, Reacher, and TwoRoom. The model predicts future representations and generates goal-conditioned actions with Direct or Guarded A planning.

전체 모델 설명 보기

Simulation research on visual world models and goal-conditioned planning.

논문 및 보고서 (1)

SADL: On the Demand-Supply Geometry of Representation Collapse

논문 있음

arXiv: 2611.21596DOI: 10.48550/arxiv.2611.21596
저자 및 소속 (10)

저자 순서: 원문 순서

  1. Junhan Sun

    State Key Laboratory of CAD&CG, Zhejiang University

  2. Yigui Feng

    RoboParty Lab

  3. Mingzhe Liu

    RoboParty Lab

  4. Haoyan Xu

    RoboParty Lab

  5. Lei Ye

    RoboParty Lab

  6. Ziqi Han

    RoboParty Lab

  7. Hanru Bai

    Alibaba Group

  8. Yanzhao Zhang

    Alibaba Group

  9. Hao ZhaoCorresponding author

    RoboParty Lab · Institute for AI Industry Research (AIR), Tsinghua University

최초 공개
2026-11-16
최신 논문 개정
v1 · 2026-11-16
학회 / 학술지
arXiv

관련 모델 버전

논문 ↗PDF ↗

접근 및 사용 방법

공개 코드 자료
공개 접근
모델 파일
공개 접근
학습 데이터
공개 접근

코드 및 학습 / 미세 조정

접근 방법
공개 접근

출처확인 날짜 · 2026-11-22

코드 라이선스
MIT

모델 파일

모델 파일
E1 · Four tasks · seed3072접근 방법가중치 라이선스상업적 사용
조건부 허용약관

출처확인 날짜 · 2026-11-22

조건 안내

MIT permits commercial use of the released weights. Keep the copyright and license notices in copies or substantial portions. Third-party components remain under their own terms.

E1 · Four tasks · seed3073접근 방법가중치 라이선스상업적 사용
조건부 허용약관

출처확인 날짜 · 2026-11-22

조건 안내

MIT permits commercial use of the released weights. Keep the copyright and license notices in copies or substantial portions. Third-party components remain under their own terms.

E1 · Four tasks · seed3074접근 방법가중치 라이선스상업적 사용
조건부 허용약관

출처확인 날짜 · 2026-11-22

조건 안내

MIT permits commercial use of the released weights. Keep the copyright and license notices in copies or substantial portions. Third-party components remain under their own terms.

E1 · Cube · seed3072접근 방법가중치 라이선스상업적 사용
조건부 허용약관

출처확인 날짜 · 2026-11-22

조건 안내

MIT permits commercial use of the released weights. Keep the copyright and license notices in copies or substantial portions. Third-party components remain under their own terms.

E1 · PushT · seed3072접근 방법가중치 라이선스상업적 사용
조건부 허용약관

출처확인 날짜 · 2026-11-22

조건 안내

MIT permits commercial use of the released weights. Keep the copyright and license notices in copies or substantial portions. Third-party components remain under their own terms.

E1 · Reacher · seed3072접근 방법가중치 라이선스상업적 사용
조건부 허용약관

출처확인 날짜 · 2026-11-22

조건 안내

MIT permits commercial use of the released weights. Keep the copyright and license notices in copies or substantial portions. Third-party components remain under their own terms.

E1 · TwoRoom · seed3072접근 방법가중치 라이선스상업적 사용
조건부 허용약관

출처확인 날짜 · 2026-11-22

조건 안내

MIT permits commercial use of the released weights. Keep the copyright and license notices in copies or substantial portions. Third-party components remain under their own terms.

E1 · Cube · seed3073접근 방법가중치 라이선스상업적 사용
조건부 허용약관

출처확인 날짜 · 2026-11-22

조건 안내

MIT permits commercial use of the released weights. Keep the copyright and license notices in copies or substantial portions. Third-party components remain under their own terms.

E1 · PushT · seed3073접근 방법가중치 라이선스상업적 사용
조건부 허용약관

출처확인 날짜 · 2026-11-22

조건 안내

MIT permits commercial use of the released weights. Keep the copyright and license notices in copies or substantial portions. Third-party components remain under their own terms.

E1 · Reacher · seed3073접근 방법가중치 라이선스상업적 사용
조건부 허용약관

출처확인 날짜 · 2026-11-22

조건 안내

MIT permits commercial use of the released weights. Keep the copyright and license notices in copies or substantial portions. Third-party components remain under their own terms.

E1 · TwoRoom · seed3073접근 방법가중치 라이선스상업적 사용
조건부 허용약관

출처확인 날짜 · 2026-11-22

조건 안내

MIT permits commercial use of the released weights. Keep the copyright and license notices in copies or substantial portions. Third-party components remain under their own terms.

E1 · Cube · seed3074접근 방법가중치 라이선스상업적 사용
조건부 허용약관

출처확인 날짜 · 2026-11-22

조건 안내

MIT permits commercial use of the released weights. Keep the copyright and license notices in copies or substantial portions. Third-party components remain under their own terms.

E1 · PushT · seed3074접근 방법가중치 라이선스상업적 사용
조건부 허용약관

출처확인 날짜 · 2026-11-22

조건 안내

MIT permits commercial use of the released weights. Keep the copyright and license notices in copies or substantial portions. Third-party components remain under their own terms.

E1 · Reacher · seed3074접근 방법가중치 라이선스상업적 사용
조건부 허용약관

출처확인 날짜 · 2026-11-22

조건 안내

MIT permits commercial use of the released weights. Keep the copyright and license notices in copies or substantial portions. Third-party components remain under their own terms.

E1 · TwoRoom · seed3074접근 방법가중치 라이선스상업적 사용
조건부 허용약관

출처확인 날짜 · 2026-11-22

조건 안내

MIT permits commercial use of the released weights. Keep the copyright and license notices in copies or substantial portions. Third-party components remain under their own terms.

출시 정보

출처확인 날짜 · 2026-11-22

배포 전 확인 사항

The reported scores cover four simulation tasks. The release provides no hardware validation for these checkpoints. 가중치나 코드가 공개되어도 로봇에서 즉시 실행할 수 있다는 의미는 아닙니다. 문서에서 구성, 데이터 및 인터페이스 요구 사항을 확인하세요.

배포 및 사용 자료

실행 환경 및 프레임워크
PyTorch

구조 및 입출력

모델 유형
월드 시뮬레이션
입력
목표 이미지 / RGB 이미지
출시일
2026년 11월 19일

적용 범위 및 학습 방식

코드 라이선스

MIT

출력

행동 / 표현

작업 범위

다중 작업

평가 결과

현재 버전 또는 선택한 가중치의 검증된 독립 평가 결과가 없습니다.

발표 기관 평가 · 보고된 결과 · 독립 재현 안 됨

SADL Shared BatchNorm Simulation · Joint training · Guarded A ↗

Macro success rate: 95.1 %

shared-batchnorm-e1

NVIDIA RTX 4090

FP32; solver batch size 1; last five primitive actions supplied. Guarded A: 128 samples × 3 iterations, top 16, initial SD 0.25.

Joint training · Guarded A. Four simulation tasks, averaged across three training seeds (standard deviation 0.3 percentage points).

발표 기관 평가 · 보고된 결과 · 독립 재현 안 됨

SADL Shared BatchNorm Simulation · Separate training · Guarded A ↗

Macro success rate: 95.6 %

shared-batchnorm-e1

NVIDIA RTX 4090

FP32; solver batch size 1; last five primitive actions supplied. Guarded A: 128 samples × 3 iterations, top 16, initial SD 0.25.

Separate training · Guarded A. Four simulation tasks, averaged across three training seeds (standard deviation 0.9 percentage points).

전체 매개변수 및 검증 설명 (모델 카드)

Inputs are 224 × 224 RGB observations, a goal image, and the causal history of the previous five primitive actions. The ViT-Tiny encoder starts without pretrained weights and uses a 192-dimensional latent space. The evaluation uses three training seeds (3072–3074) and three evaluation seeds (0, 1, 42), with 100 episodes per evaluation seed and task on an RTX 4090. Goals are 25 steps ahead, with at most 50 actions, a planning horizon of 5 and replanning every 5 actions. Four-task macro success is 94.8 ± 0.4% (Direct) and 95.1 ± 0.3% (Guarded A). Separately trained models score 94.9 ± 0.5% and 95.6 ± 0.9%, respectively. The values are means ± standard deviations across training seeds. Inference uses FP32 with solver batch size 1. Direct generates the actor’s plan without candidate search. Guarded A evaluates 128 samples per iteration for three iterations, using the top 16 and an initial standard deviation of 0.25. These numbers describe the released Shared BatchNorm checkpoints; the paper’s other ablation protocols report separate results.

출처 또는 수정 제안 보내기