Entity dossier

Model model--lda-1b

LDA-1B

Latent world-action foundation model trained on the EI-30k mixed embodied corpus.

Reviewed relationships 6

  1. Verified

    WAM-TTT Model adapts LDA-1B

    Scope test-time training research lineage

    Boundary WAM-TTT steers a frozen world-action model in the LDA line; this is not proof of a product checkpoint.

    First observed · reviewed · [R5] · Show in graph

  2. Verified

    Galbot Organization coauthors LDA-1B

    Scope paper affiliation

    Boundary Research authorship does not by itself prove a shipping product integration.

    First observed · reviewed · [R9] · Show in graph

  3. Verified

    PKU EPIC Lab Organization coauthors LDA-1B

    Scope paper affiliation

    Boundary Academic research lineage, not a product deployment claim.

    First observed · reviewed · [R9] · Show in graph

  4. Verified

    LDA-1B evaluated on Galbot G1 Robot

    Scope real-world paper figures

    Boundary The paper uses Galbot G1 with gripper and Sharpa-hand variants.

    First observed · reviewed · [R9] · Show in graph

  5. Boundary

    LDA-1B not proven on Galbot ET1 Robot

    Scope public evidence boundary

    Boundary The paper names Galbot G1 as the eval body; that does not prove an exact ET1 checkpoint.

    First observed · reviewed · [R9] · Show in graph

  6. Verified

    LDA-1B trained on EI-30k Dataset

    Scope world-action pretraining corpus

    Boundary EI-30k is the mixed human and robot corpus used with LDA-1B.

    First observed · reviewed · [R9] · Show in graph

Canonical evidence claims 5

Claims are projected from authored tables. A dash remains an explicit missing value; it is never filled from a neighboring product or name match.

Research — recent papers Paper Body Galbot G1 5 rows · 1 declared gap

MetricValue and evidence
LDA-1BarXiv:2602.12215 [R9]
LDA-1Bpaper eval — real-world figures in the paper, not a store SKU [R9]
LDA-1Bparallel-jaw gripper; Sharpa 22-DoF hand [R9]
LDA-1BGalbot G1 [R9]
LDA-1B Declared gap

Source ledger 2

  1. [R5]ResearchWAM-TTT — “Steering World-Action Models by Watching Human Play at Test Time”. Type: research paper. ID: arXiv:2607.06988. Retrieved: 2026-08-26.Retrieved
  2. [R9]ResearchLDA-1B — “Scaling Latent Dynamics Action Model via Universal Embodied Data Ingestion”. Type: research paper. ID: arXiv:2602.12215. Retrieved: 2026-08-26.Retrieved

What is not inferred

No fuzzy entity matching, adjacent-column borrowing, universal compatibility, commercial maturity, or checkpoint-to-product identity. Absence of a relation means unreviewed or unsupported here—not proven incompatibility.