Semantic Similarity Table - GitHub

Drag and drop your Excel or CSV file to calculate batch semantic similarity for text queries using latest embedding models. Using transformers.js with a small model from model2vec fully privately in your browser. Your data stays on your computer and is not sent anywhere!
Need sample data? Use legal_texts_EU.xlsx, legal_texts_EU.csv or SDG_Targets_2023.xlsx. Note that the first run is slower as it loads and caches the model once (smallest is only 32Mb). Conecutive runs are pretty much instant. Created by Dominik Weckmüller (Blog).
Some of my other projects that inspired semantic similarity table: SemanticFinder, JS Text Chunker, SDG Analyzer, Geospatial Semantic Search, Qdrant Frontend. I'm available as freelancer for anything related to AI, LLMs, NLP and particularly semantic search and RAG - let's connect on LinkedIn!

Drop your Excel or CSV file here
or click to select

Select model to be used (more parameters = better quality but larger file size)

Text column to be used for semantic similarity (first row is used for headers)

Query text (theoretically unlimited in size but for best results keep < 300 words)

Select columns to add:

Max similarity

Mean similarity

Max similarity chunk

Chunks count

Include embeddings (causes large file size; use only with csv as xlsx has a limit of ~32k chars/cell)