Pinned Loading

1. FlexSplat FlexSplat Public

   [BMVC 2026] Feed-forward 3D Gaussian splatting for uncalibrated object-centric novel view synthesis, using jointly trained VGGT geometry and query-based multi-view Gaussian decoding.

   Python 8

2. Label-Agreement-Does-Not-Measure-Authorization Label-Agreement-Does-Not-Measure-Authorization Public

   [NeurIPS 2026 (AIM Workshop)] Evidence-grounded audit framework for agentic neuroimaging metadata harmonization, covering authorization, contract completeness, evidence sensitivity, and downstream …

   Python 1

3. confident-unreliable-vlm-brain-mri-audit confident-unreliable-vlm-brain-mri-audit Public

   [ACM AI 2026] Safety-audit framework for vision-language models on brain MRI, measuring calibration, confident errors, hallucination, abstention, and slice-level accuracy.

   Python 1

4. machine-unlearning-retain-forget-experiments-in-pytorch machine-unlearning-retain-forget-experiments-in-pytorch Public

   PyTorch machine-unlearning lab comparing exact retraining with cheap retain-finetuning, negative-gradient, and class-reset forgetting methods on a small CPU-friendly dataset

   Python

5. Enterprise-Knowledge-Assistant Enterprise-Knowledge-Assistant Public

   Production-style RAG/LLM system with hybrid retrieval, reranking, citation verification, evaluation metrics, latency/cost tracking, caching, FastAPI, frontend, and Docker.

   Python

6. teacher-student-LLM-distillation teacher-student-LLM-distillation Public

   PyTorch pipeline for distilling a larger causal language model into a compact student using teacher top-k logits, temperature-scaled KL loss, hard-label CE, and generation evaluation

   Python