-
Add Run 8 test results (FAR<=1% unmet at v2)
-
Add Run 9 exact-threshold test results (FAR 0.97%)
-
Card: Run 8/9 results — FAR <= 1% crossed (0.97% @ thr 0.9906); add neg checkpoints + test artifacts
-
Upload o1sound_neg_v2.metrics.json with huggingface_hub
-
Upload o1sound_neg_v1.metrics.json with huggingface_hub
-
Upload o1sound_neg_v2.pt with huggingface_hub
-
Upload o1sound_neg_v1.pt with huggingface_hub
-
Upload o1sound_multiclass_v2.pt with huggingface_hub
-
Add paper/demo links to model card
-
O1-Sound research release: binary + multiclass checkpoints, streaming ONNX, per-language metrics
-
initial commit