Trained Deep Learning Models

Fine-tuned model checkpoints, weights, and specialized neural network pipelines.

Exploitability Prediction Transformers

Fine-tuned checkpoints of DistilBERT and GPT-2 architectures trained on the NVD Exploit Likelihood Corpus to classify raw vulnerability descriptions into exploit probability percentiles.

PD Gait Classifier GCN

Weights for a Graph Convolutional Network (GCN) designed to identify Parkinson's Disease gait anomalies and FOG precursors from 3D sensor graph topology streams.

Wan 2.2/2.6 Lip-Sync Adapter

Optimized configurations and ComfyUI pipeline definitions for avatar lip-sync inference using LatentSync and F5 voice cloning adapters.