Pinned Loading

1. redcell redcell Public

   Async-first, vendor-agnostic locally-run agent, mcp, and AgentGateway developed for the purpose of testing open-source AI-Security tools.

   Python 1

2. ModelDistillation ModelDistillation Public

   A progressive lab course, from the objective function to a reproducible research pipeline, with the real-model track sized for a single workstation with 128 GB of unified CPU–GPU memory at ~273 GB/…

   Jupyter Notebook

3. spark-ablate spark-ablate Public

   Component-ablation toolkit for open-weight causal LMs, built for single-node research on an NVIDIA DGX Spark (and developed/tested on CPU anywhere).

   Python

4. mlx-benchmark mlx-benchmark Public

   MLX LLM Benchmark Suite for Apple Silicon. Comprehensive benchmarking tools to measure LLM inference performance on Apple Silicon Macs, with metrics directly comparable to HuggingFace model cards.

   Python 4

5. APG APG Public

   APG (Agent Policy Gateway) was created as an intent-classification and risk-assessment layer that sits between AgentGateway's transport and Cedar's policy evaluation, turning raw tool calls into se…

   Python 1

6. lora-backdoors lora-backdoors Public

   Accompanying work to the research paper "Token-Level Generalization in LoRA Adapter Backdoors: Attack Characterization and Behavioral Detection" by Travis Lelle

   Jupyter Notebook 3