scale-1-lora
Portable Speculators static_lora inference bundle.
Load with speculators.generator.serving.bundle.bundle_state. Use the downloaded folder as a bundle in the generator benchmark configuration. This bundle requires the pinned target and drafter models recorded in bundle.json.
resolved_experiment.json records architecture, training data identity, and optimization settings. Training commands, patches, metrics, and available extraction provenance are included.
Source checkpoint SHA256: b145a18fea0a70344c33c801f63a400f449b890a278bf9165e30a4e5773bb3eb.
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support