vipan-kumar's picture
Initial commit: Audio Deepfake Detector with 8 detectors trained on jay15k
e6a1f55
Raw
History Blame Contribute Delete
439 Bytes
# Place model checkpoint files here.
# Files named per the spec:
# nes2net.pt - Nes2Net backend weights (state_dict for Nes2NetBackend)
# sonar.pt - SONAR head weights (SRM filters + cross-attention + classifier)
# bicrossmamba_st.pt - BiCrossMamba-ST weights
# When a file is present, the corresponding architecture-only detector switches
# from `architecture_only` status to `live` and runs real inference.