Repository navigation
v0.8.500
- Atom-JEPA serving: GPU inference image, async GPU defaults with a one-instance cap, endpoints.atomjepa_utils split, threaded conformers, weight and compressed-feature refusals
- Atom-JEPA docs: LogD result and framework MPPB numbers
- Atom-JEPA docs: when to use it, measured results, independent evaluation, references
- bumping version of workbench for all images