[CVPR 2021] Official PyTorch implementation for Transformer Interpretability Beyond Attention Visualization, a novel method to visualize classifications by Transformer based networks.

Detect & Filter korean curse text using huggingface transformer, KcBERT, Transformer-Interpret

Probing quality-evaluative geometry in transformer hidden states. GPT-2 encodes quality better than BERT, with a negativity bias that mirrors human cognition.

Configurable character-level transformer training suite with built-in mechanistic interpretability toolkit — scale to 150M+ parameters and beyond, no ceilings, only hardware limits. Inspect attention weights, hidden states, and head specialisation across all layers. Documented circuit findings included.

Fourier, graph, Hodge, and signed-circulation probes for transformer hidden-state trajectories.

A diagnostic control paradigm for activation measurements in transformer language models. Cross-replay separates text-bound from architecture-bound components by replaying generated sequences through intact and perturbed model variants.

The Spectral Gap-Statement: when the negative subspace of attention transport is a well-posed invariant