How to Use
- Input Four Words: Fill in the words A, B, C, and D. You can use the 'Prepare an example for me' button to auto-fill them.
- Run the Detection: Click the 'Is it an analogy?' button. The model will evaluate if A : B :: C : D forms a valid analogy. For example, 'king : queen :: man : woman'.
Output Information
- Model Decision: The model will indicate whether the analogy is considered 'Valid' or 'Invalid'.
Note
This model is based on a pretrained LLM and operates in a zero-shot setting. Its reasoning is based on language patterns and may occasionally be influenced by biases or lack of context. Computation may take up to 20–40 seconds.
How to Use
- Input Three Words: Provide A, B, and C. You can use the 'Prepare an example for me' button to auto-fill them.
- Run the Resolution: Click 'Get closest solution'. The model will predict a word D such that A : B :: C : D.
Output Information
- Predicted Word: The LLM will suggest the most likely word D to complete the analogy.
Note
Like the classification model, this LLM operates without fine-tuning and may reflect linguistic biases. Expect around 30–40 seconds for a response.