Tutorials

TorchRL PPO Action Probing

Set up a TorchRL PPO agent and use tdhook to probe action representations.

notebooks/tutorials/torchrl-ppo.ipynb

Chess Value Saliency

Compute attribution maps for chess model predictions using saliency methods.

notebooks/tutorials/chess-value-saliency.ipynb

Concept Attribution (LRP + RelMax)

Build a striped concept and visualize concept-conditioned relevance on natural images.

notebooks/tutorials/concept-attribution.ipynb

Chess Dimension Estimation

Estimate channel-wise intrinsic dimension from chess activations and compare layers, heads, and opening splits.

notebooks/tutorials/chess-dimension-estimation.ipynb