model-debugging

Monitoring platform for ML teams building real‑world AI at scale.

XAI-driven augmentation & diagnostics for PyTorch vision - find model failures, fix with saliency-guided augmentation (ICD/AICD), prove with auditable reports.

Codes for the EMNLP 2020 paper -- "FIND: Human-in-the-loop Debugging Deep Text Classifiers"

Custom ML tracking experiment and debugging tools.

ML is a tool for diagnosing model errors: segments, fairness, and calibration

PyTorch NaNs are silent killers. This hook catches them at the exact layer and batch — with ~3 ms overhead vs ~7 ms for set_detect_anomaly.

Autonomous failure investigation, root-cause analysis, and validated interventions for LLMs, VLMs, and agents.

Neural network visual debugger for model graph inspection, activation health, and gradient diagnostics.

Dog Breed Classification using Transfer Learning in AWS Sagemaker

file(1) of the ternary age — balanced-ternary-aware GGUF inspector and debugger in Rust

Git-bisect for neural networks: trace behavioral changes across checkpoints to internal mechanisms, influential training data, and counterfactual causal evidence.

Mechanistic interpretability that ships: MAIR-backed evidence bundles, receipts, and comparison packets.

Easy-to-use UI based tool that visualizes the internal layers and activations of any Pytorch network that takes image as input , built using PyQt

Model Debugging

BRACE - BetteR Accuracy from Concept-based Explanation

Stop guessing why your MLX model outputs garbage. Triage in 30 seconds — no model load required.

ML research on identifying which training change caused a model regression and verifying the diagnosis through controlled retraining.