2 days ago
-
Model card: training modalities
-
Model card: simpler inputs wording
-
Model card: state the inputs plainly (RGB + DINO always, depth optional)
-
Model card: DINO is always used; training depth per corpus
-
Card: plain wording for the action head
-
Card: renamed to flexpi-pretrain-preview
-
Card: lean preview card; 32-wide fine-tune stats
-
Stats: anonymized corpus keys and a public fit note; card wording
8 days ago
-
Model card: record the offline GPU checks
-
Add files using upload-large-folder tool
