MiniLM · English
Model-based extractive summary
Select central sentences from an English text using model embeddings.
MiniLM is an English embedding model, not a chatbot. First use loads about 47 MB of model/runtime data from this site before HTTP compression; transferred bytes depend on hosting and cache. Up to 40 source sentences, 800 characters per sentence and 256 model tokens per input. Short text comparison and topic matching support 800 characters. Model scores are not probabilities. Loading has a five-minute limit; inference has a two-minute limit.
Source text
Result
Choose an input, then run the tool.
Method, limits & licensing
MiniLM is an English embedding model, not a chatbot. First use loads about 47 MB of model/runtime data from this site before HTTP compression; transferred bytes depend on hosting and cache. Up to 40 source sentences, 800 characters per sentence and 256 model tokens per input. Short text comparison and topic matching support 800 characters. Model scores are not probabilities. Loading has a five-minute limit; inference has a two-minute limit.
Model sources, versions and checksumsA guide to this tool
How to use Model-based extractive summary
- Enter English text or choose the sample.
- Set the second input and result count where available.
- Run the local model and review both the result and its score limitations.
A sample you can try
Sample input
The dog is running through the park. A puppy plays outside on the grass. The stock market closed higher today. Investors are watching company earnings. The cat sleeps on a warm sofa. A kitten rests indoors.
What to expect
Selected original sentences in source order, without newly generated prose.