Semantic text comparison
Compare two texts using local embedding similarity.
Semantic text comparison
Cara menggunakan Semantic text comparison
Semantic text comparison measures how close two texts are in meaning by embedding each with the all-MiniLM-L6-v2 model in your browser and reporting their cosine similarity to four decimal places. Paste the first text in the main box and the second in Comparison text, each up to 8,000 characters. The score is not a plagiarism, truth or equivalence verdict.
- Paste the source into the input editor, or load the built-in example.
- Set comparison text.
- Run semantic text comparison, review the output, then use the available copy or download controls.
Fungsi alat ini
Downloads an inference model on first run. Chat and generation require WebGPU; smaller recognition models use browser WebAssembly. AI output can omit or invent details. Review it before use.
- Comparison text
- Set this value for your task.
Had dan pemprosesan
Text sources are capped at 8,000 characters; image search accepts up to 20 images and a 500-character query. Models may truncate long text. First downloads can be hundreds of MB and need browser memory.
Contoh input
A useful tool makes everyday work easier. Review the results before sharing.
Soalan lazim
What does the cosine similarity score mean?
It measures how closely the two embedding vectors point in the same direction. Values near 1 mean very similar meaning according to the model, and lower values mean less related content. There is no fixed pass mark, so compare scores across several text pairs.
Can semantic text comparison detect plagiarism?
No. The tool states the score is not a plagiarism, truth or equivalence verdict. A high score only means the model sees related meaning; it does not show copying, agreement or that both texts are correct.
How long can the texts be for semantic comparison?
Each text can be up to 8,000 characters, but the embedding model may truncate long inputs, so later parts of a long text might not influence the score. Comparing paragraph-sized sections gives a more representative result.
Where is my input processed?
Local mode downloads public model/runtime assets and keeps input in this browser. Models may remain cached on this device.
What are the input limits?
Text sources are capped at 8,000 characters; image search accepts up to 20 images and a 500-character query. Models may truncate long text. First downloads can be hundreds of MB and need browser memory.