Instructions to use Synthyra/Boltz2 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
How to use Synthyra/Boltz2 with Transformers:
# Use a pipeline as a high-level helper from transformers import pipeline pipe = pipeline("feature-extraction", model="Synthyra/Boltz2", trust_remote_code=True)# Load model directly from transformers import AutoModel model = AutoModel.from_pretrained("Synthyra/Boltz2", trust_remote_code=True, dtype="auto") - Notebooks
- Google Colab
- Kaggle
Upload README.md with huggingface_hub
Browse files
README.md
CHANGED
|
@@ -87,6 +87,7 @@ The export directory contains:
|
|
| 87 |
## Limitations
|
| 88 |
- Current featurization path is protein-only and minimal.
|
| 89 |
- This implementation is meant for practical inference and export workflows, not full Boltz training parity.
|
|
|
|
| 90 |
|
| 91 |
## Docker-first compliance testing
|
| 92 |
|
|
|
|
| 87 |
## Limitations
|
| 88 |
- Current featurization path is protein-only and minimal.
|
| 89 |
- This implementation is meant for practical inference and export workflows, not full Boltz training parity.
|
| 90 |
+
- Test-time training is not supported for Boltz2 in FastPLMs. TTT is currently limited to sequence PLMs plus ESMFold and ESMFold2 PLM backbones.
|
| 91 |
|
| 92 |
## Docker-first compliance testing
|
| 93 |
|