llm-token-heatmap is an open-source tool for analyzing and visualizing how large language models (LLMs) select each next token. It provides insights into probabilities, attention, logit-lens, and activations via both CLI and web interfaces, supporting AI researchers and ML engineers in model interpretability.
llm-token-heatmap sits in PulseGate's LLM eval & observability category. It helps researchers and developers interpret and visualize how large language models select tokens and make predictions. llm-token-heatmap is an open-source project aimed at AI researchers and ML engineers. The project is open source (MIT). It runs on the command line and the web.
zangjiucheng builds and maintains llm-token-heatmap, and the product first shipped in 2026. The project is developed in the open on GitHub with 113 commits in the last 90 days. Among its 5 catalogued features are token probability visualization, attention analysis, and logit-lens inspection.
Latest indexed changes and source events
llm-token-heatmap verified by the PulseGate indexer
Other apps tracked under the same category.
