VINAY-UMRETHE/SigMamba-V1-Small
Video Classification • 0.4B • Updated • 265
Vinay Umrethe
VINAY-UMRETHE
AI & ML interests
Machine Learning, Multimodal LLMs, Data Engineering, Data Science
Recent Activity
updated a bucket about 17 hours ago
heretic-org/Heretic-Grimoire-Storage updated a Space 2 days ago
heretic-org/Heretic-Grimoire posted an update 4 days ago
Presenting SigMamba-V1
https://huggingface.co/collections/VINAY-UMRETHE/sigmamba-inventory
A unified architecture that couples SigLIP2 vision encoder with a trainable Mamba state space model for temporal reasoning with O(N) complexity.
Trained under the Multiple Instance Learning (MIL) paradigm with the Temporal Feature Magnitude (RTFM) loss, SigMamba achieves 89.82% frame-level AUC on the UCF-Crime benchmark while processing over 1000 frames per second on a single GPU.
Released two model variants https://huggingface.co/VINAY-UMRETHE/SigMamba-V1-Large and https://huggingface.co/VINAY-UMRETHE/SigMamba-V1-Small along with all training code and datasets under an open-source license.
GitHub: https://github.com/Vinay-Umrethe/SigMamba-V1Organizations
