facebookresearch/llm-transparency-tool
An interactive toolkit for visualizing attention patterns, neuron activations, and contribution graphs inside transformer language models.
Not currently ranked — collecting fresh signals.
star history
The LLM Transparency Tool is an open-source visualization and analysis platform for inspecting how transformer-based language models process information internally. It renders interactive contribution graphs showing how tokens influence each other through attention heads and FFN blocks, lets users trace neuron activation patterns across layers, and projects token representations onto the output vocabulary to reveal which tokens were promoted or suppressed. Built with Streamlit and React, it supports multiple models through TransformerLens integration.
Frequently asked
- What is facebookresearch/llm-transparency-tool?
- An interactive toolkit for visualizing attention patterns, neuron activations, and contribution graphs inside transformer language models.
- Is llm-transparency-tool open source?
- Yes — facebookresearch/llm-transparency-tool is an open-source project tracked on heatdrop.
- What language is llm-transparency-tool written in?
- facebookresearch/llm-transparency-tool is primarily written in Python.
- How popular is llm-transparency-tool?
- facebookresearch/llm-transparency-tool has 1.3k stars on GitHub.
- Where can I find llm-transparency-tool?
- facebookresearch/llm-transparency-tool is on GitHub at https://github.com/facebookresearch/llm-transparency-tool.