StarForce
Dual system training framework for robotics.
The overall structure borrowed from GR00T, with these modifications:
- Simplifer VLM model, introduced more advanced VLM and larger slow system;
- Connecting fast system without cross-attention, using text encoder instead;
Slow thought system
will goes starforce/model/backbone contains various VLMs. Provides a unified interface connect with fast system (action expert)
Fast action system
Currently support:
- DiT: diffusion transformer
- QwenFlow: flowmatching based action expert
Env install
pip install -e .[base]
Training
training scripts goes to scripts/xxx.sh
training fast system:
sh scripts/v0/sl_0.sh
training slow thinking system:
1 maintainer
Metadata
1 maintainer
Release files for starforce-twin 1.1.0
For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.
Source distribution (sdist)
| File | Size | Uploaded | |
|---|---|---|---|
| starforce_twin-1.1.0.tar.gz | 89.6 kB | Details |
Release files / starforce_twin-1.1.0.tar.gz
| Download URL | starforce_twin-1.1.0.tar.gz |
|---|---|
| Size | 89.6 kB |
| Tags | Source |
|
SHA-256 checksum How to use checksums |
5da33f476f7e02662a8589399ef82f2b9056303ce0d62fb900a4bf964c412255
|
|
BLAKE2b-256 checksum How to use checksums |
075e12d24270dd55f1bd346129ff5c247b024ef1790d5784cb0810ecfe4490e4
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
No |
| Uploaded via |
twine/6.1.0 CPython/3.12.9
|
