KAIST / University of Washington et al. (Latent Action Pretraining)
LAPA
7B Commercial use OK 8/8 verified
Facts
- Released
- 2024-11 1 Source
- Parameters
- 7 B 0 Source
- Architecture
- Monolithic autoregressive 0 Source
- License
- MIT (tagged on both code and model card) 2 Source
- Commercial
- Commercial use OK 0 Source
- Weights
- huggingface.co ↗ 1 Source
- Code
- github.com ↗ 1 Source
- Paper
- arxiv.org ↗ 0 Source
- Input modalities
- rgb · language 0 Source
- Action space
- Latent actions → mapped to real actions after fine-tuning 0 Source
- Cross-embodiment
- Open X-Embodiment pretraining; SIMPLER evaluation 1 Source
- Pretraining data (hours)
- Unverified 0 Source
- Target forms
- Single arm 0 Source
Architecture · Learns latent actions from action-free video, then pretrains the VLA; based on LWM-Chat-1M (Large World Model); ICLR 2025
Sources · 2 sources
release_date,license,cross_embodiment,code_url https://github.com/LatentActionPretraining/LAPA Official page · Fetched 2026-09-05 · No snapshot stored yet README：weights released 2024-11-22；MIT License；Open-X Embodiment；ICLR 2025
weights_url,license https://huggingface.co/api/models/latent-action-pretraining/LAPA-7B-openx Model hub page · Fetched 2026-09-05 · No snapshot stored yet Model card license: mit; base_model LargeWorldModel/LWM-Chat-1M-Jax
Activity signals
Public counters fetched daily from GitHub, Hugging Face and ModelScope. Numbers only; no judgement. · fetched 2026-10-31
GitHub stars
586
+8 / 30 d
Commits · 90 d
0
last push 648 d ago
HF downloads · 30 d
0
all-time 0 · 16 likes
Weights last modified
2024-11-22
709 d ago
ModelScopeno same-name mirror found (searched by repo name)
Hugging Facelatent-action-pretraining/LAPA-7B-openx
stars · last 30 d
Public benchmark scores
As printed in the source; protocols differ between papers. Full table with notes on the benchmarks page.
| Benchmark | Subset | Score | Source |
|---|---|---|---|
| Other benchmarks | In-domain / Seen | 62 | paper · Table 1, row 'LAPA', In-domain (1k) Seen |
| Other benchmarks | In-domain / Unseen | 49.6 | paper · Table 1, row 'LAPA', In-domain (1k) Unseen |
| Real-robot evaluation (self-reported)LAPA (Bridge) | avg | 36.8 | paper · Table 2, row 'LAPA (Bridge)', AVG |
| Real-robot evaluation (self-reported)LAPA (Human Videos) | avg | 34 | paper · Table 2, row 'LAPA (Human Videos)', AVG |
| Real-robot evaluation (self-reported)LAPA (Open-X) | avg | 50.1 | paper · Table 2, row 'LAPA (Open-X)', AVG |
| SimplerEnv · WidowX (Bridge)LAPA (BridgeV2 pretrain) | overall | 57.3 | paper · Table 11, 'Success Rate' block, row 'AVG', column 'LAPA' |
| SimplerEnv · WidowX (Bridge)LAPA (Sthv2 human video pretrain) | overall | 52.1 | paper · Table 12, 'Success Rate' block, row 'AVG', column 'LAPA' |
Deployment recipe
No recipe yet. We publish only steps that actually ran on a rented GPU with a measured latency; this model has not been through that yet.
Record history
2026-09-05
Sources fetched
release_date,license,cross_embodiment,code_url · weights_url,license
Latency & cost
cost = p50 ms ÷ 1000 × rental ÷ 3600 × 1000
| Hardware | p50 ms | p95 ms | Peak VRAM | Cost per 1k inferences ¥ | samples |
|---|---|---|---|---|---|
NVIDIA RTX 4090 ¥1.68/h · GongJi Compute (2026) | — | — | — | — no latency data | No data |
NVIDIA RTX 5090 ¥2.4/h · Runyun 2.4 / GongJi 2.5 / Paratera Cloud 2.98 (2026) | — | — | — | — no latency data | No data |
NVIDIA A100 80GB no rental price | — | — | — | — no latency data | No data |
NVIDIA Jetson AGX Thor no rental price | — | — | — | — no latency data | No data |
NVIDIA H100 80GB no rental price | — | — | — | — no latency data | No data |
NVIDIA A800 80GB no rental price | — | — | — | — no latency data | No data |
No latency measurements for this model yet · Data comes from literature, maintainer runs, or crowd CLI uploads. None exist so far.
Compatibility
Colored only where we hold evidence. An amber dot means the vendor declares this form as a target; that is a claim, not a verified compatibility.
| Unitree G1 | Unitree H1 | AgiBot Genie G1 | AgileX Piper arm | SO-101 | Franka FR3 / Panda | Fourier GR-2 | WidowX 250 S | Google Robot（Everyday Robots） | ALOHA 2 / Cobot Magic (dual-arm) | Galaxea R1 Pro | Franka Emika Panda | |
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| LAPA |
Legend Officially supportedCommunity reproducedFine-tunable in theoryUnsupportedUnknown Vendor-declared target form 2 / 12 cells with a definite status
Reproduction records
Did this model run on your robot? Report it with a link (log, video or repo). Records are reviewed by the maintainers; a verified “works” or “works after fine-tuning” upgrades the matrix cell to “community verified” with your link as evidence. Anonymous is fine; nothing but the fields below is stored.
No verified record yet for this model.
Clear direction, grounded judgement.