kevinqz's picture
card: fix ACT asset-shape contradiction (single self-contained graph, norm baked in β€” not split-export/denoise_step)
a3f201a verified
|
Raw
History Blame Contribute Delete
5.96 kB
---
license: apache-2.0
base_model: lerobot/act_aloha_sim_insertion_human
pipeline_tag: robotics
library_name: coreai
tags:
- coreai
- core-ai
- coreai-fabric
- aimodel
- coreml
- apple
- apple-silicon
- on-device
- robotics
- vla
- vision-language-action
- action
---
> **Canonical:** [`kevinqz/ACT-Aloha-Insertion-CoreAI`](https://huggingface.co/kevinqz/ACT-Aloha-Insertion-CoreAI) β€” source of truth.
# ACT ALOHA Insertion (fabric)
> ⚠️ **Robot policy β€” needs a matching robot to actuate.** This is
> [lerobot/act_aloha_sim_insertion_human](https://huggingface.co/lerobot/act_aloha_sim_insertion_human) converted to an
> Apple Core AI `.aimodel`. Its output is a **normalized action chunk** for
> **aloha_bimanual_14dof_sim**. Run it on any other robot, or with mismatched calibration /
> normalization stats, and it emits floats that *look* valid but **actuate
> garbage** β€” the most dangerous silent-failure mode. It is a **base checkpoint**:
> fine-tune on your robot's data before expecting useful behavior.
An Apple Core AI conversion of
[lerobot/act_aloha_sim_insertion_human](https://huggingface.co/lerobot/act_aloha_sim_insertion_human) β€” a **robot policy**
that maps images + proprioceptive state (+ a language instruction, when the model
uses one) to a continuous action chunk (act sampler). Produced by
[coreai-fabric](https://github.com/kevinqz/coreai-fabric/blob/main/recipes/act-aloha-insertion.yaml) and indexed by
[coreai-catalog](https://github.com/kevinqz/coreai-catalog).
## Model facts
| Field | Value |
|---|---|
| Parameters | 51.6M |
| Architecture | transformer |
| Capabilities | robotics |
| Embodiment | aloha_bimanual_14dof_sim |
| Sampling | act |
| Quantization / precision | none / float16 |
| On-disk size | 131 MB |
| Asset kind | single-graph policy (self-contained β€” normalization baked in) |
| assetVersion | 2.0 |
## Use it β€” this needs host code you supply
A policy is **not** a chat model: there is no stock high-level Swift runtime for
it. The bundle is a **single self-contained graph** β€” you run it **once per observation** and it returns the whole action chunk directly. Normalization is **baked into the traced graph** (`predict_action_chunk` normalizes the inputs and un-normalizes the outputs), so there is **no `norm_stats.json` sidecar and no sampler loop** (unlike the flow-matching Pi0/Diffusion bundles). **You supply the host wiring** in Swift β€” feed the observation tensors, take the returned action chunk. Recommended integration: keep LeRobot's
Python `RobotClient` for the servos/cameras/calibration, and run inference
on-device β€” see the `io_contract` in the catalog for the exact tensors.
```bash
pip install coreai-catalog && coreai-catalog install act-aloha-insertion
```
## Requirements
- **Deployment: macOS 27.0+ / iOS 27.0+, Xcode 27+.** The asset serializes with `minimum_os v27`,
so the on-device Swift runtime requires macOS/iOS 27+. A Mac on macOS 26 can
convert and inspect it but not run it on-device.
- Apple Silicon. A matching robot to actuate (see the banner).
## Verification (action parity)
- **Gate A (structure): passed** β€” the bundle's layout + metadata were
validated on real hardware; the graphs load.
- **Gate B β€” action_parity: 100.0% min chunk-cosine** vs the fp16 reference over 8 frames of `recorded episodes`, N-step, fixed-noise (measured on Apple Silicon).
- This certifies the export is **numerically faithful to the source policy** β€” it
does **NOT** certify real-world task success, embodiment transfer, or
closed-loop stability. Closed-loop success-rate eval (LIBERO/ManiSkill):
**not_run** (a separate future gate). Reproduce with `coreai-fabric verify`.
## Provenance
| Field | Value |
|---|---|
| Base model | [lerobot/act_aloha_sim_insertion_human](https://huggingface.co/lerobot/act_aloha_sim_insertion_human) @ `33259aa86eb45fdf85350280044a33d9d50e40c3` |
| Converted by | `models/act/export.py` coreai_torch 0.4.1 + coremltools 9.0 |
| Recipe | [act-aloha-insertion](https://github.com/kevinqz/coreai-fabric/blob/main/recipes/act-aloha-insertion.yaml) (recipe_source: fabric) |
| Precision / quantization | float16 / none |
| Conversion date | 2026-07-04 |
Machine-readable, in this repo:
[`parity-report.json`](./parity-report.json) (gate results) Β·
[`reproduce-manifest.json`](./reproduce-manifest.json) Β· [`LICENSE`](./LICENSE)
(upstream terms).
## License and attribution
Weights licensed **apache-2.0** β€” see the bundled `LICENSE`. This artifact is a **converted derivative** of the base policy: its
weights were converted to Apple Core AI format. The conversion itself is
community work.
## Links
- **Base model:** [lerobot/act_aloha_sim_insertion_human](https://huggingface.co/lerobot/act_aloha_sim_insertion_human)
- **Reproduce:** [recipe `act-aloha-insertion`](https://github.com/kevinqz/coreai-fabric/blob/main/recipes/act-aloha-insertion.yaml)
- **Index:** [coreai-catalog](https://github.com/kevinqz/coreai-catalog) β€” the
neutral registry tying upstream ↔ this asset ↔ mirror together
- [HF Collection](https://huggingface.co/collections/kevinqz/coreai-apple-on-device-6a4879f21c7e1a87c99bcf5a)
## The on-device Core AI ecosystem
- [coreai-fabric](https://github.com/kevinqz/coreai-fabric) β€” the reproducible
recipe β†’ `.aimodel` pipeline that produced this asset.
- [coreai-catalog](https://github.com/kevinqz/coreai-catalog) β€” the index of Core
AI models across the community, with provenance and integration snippets.
- [apple/coreai-models](https://github.com/apple/coreai-models) β€” Apple's official
exporters and runtimes.
- [LeRobot](https://huggingface.co/lerobot) β€” the upstream robotics ecosystem.
## Not affiliated with Apple
Community conversion. Not produced, hosted, or endorsed by Apple. Apple and Core
AI are trademarks of Apple Inc., used here only to describe the target
runtime/format. This is an independent community conversion.