I

interpretability

LLM token stream viewer: see per-token confidence and perplexity from OpenAI or Anthropic logprobs, and rewrite every other token live. CLI and web UI; mock mode runs locally with no API key. Rust.

Local dashboard that streams a Claude, GPT or Ollama reply chunk by chunk, shows where the model paused, flags claims worth checking and prices the prompt before you send it. Works fully offline with Ollama.

Code release for the paper 'On the Faithfulness of Post-Hoc Concept Bottleneck Models' (ECCV 2026).

A Python library for Secure and Explainable Machine Learning

Documentation available @ https://secml.gitlab.io

Follow us on Twitter @ https://twitter.com/secml_py

A zero-inflated Poisson model is fitted to GSS dataset for estimating the number of mentally unhealthy days of American residents aged 18-24 per month.

SecML models and databases zoo.