johnmarktaylor91/torchlens

Inspect, visualize and intervene on any operation in any neural network. Capture every activation and gradient, the full computational graph and 800+ metadata fields without changing model code. Supports PyTorch, JAX, TensorFlow, MLX, PaddlePaddle and tinygrad. Validated on 6400+ models. Built for researchers and their AI agents.

2 best points · 1 comments · 1 Hacker News post · 666 GitHub stars · View on GitHub

Hacker News discussions

• TorchLens–package for visualizing nearly any PyTorch model in 1 line of code (2 points, 1 comments, by johnmarktaylor, 2024-07-23)