Skip to content
TheChatArchiveThe Chat Archive
Databricksdolly-v2-12b· DatabricksHugging Face Arena Human Preference 55K

please provide a framework to perform text similarity specifically for human names

From a Chatbot Arena battleThe same person sent this prompt to two anonymous models at once. Compare it with Claudeclaude-instant-1’s answer.
  1. User

    please provide a framework to perform text similarity specifically for human names

  2. Databricks Assistant

    The most common framework to perform text similarity is the cosine similarity. Given a vector representation for each text, the cosine similarity measures the similarity as the absolute value of the dot product of the two vectors divided by the product of their lengths.

    There are known issues with this metric when the texts have different pre-processing applied such as lowercasing, stemming, or tokenization. In those cases, the L2 normalization can be applied to the vectors before computing the cosine similarity.

    There are other similarity metrics such as BLEU, METEOR, and ROUGE that can be used in name similarity tasks.

Responses were produced in different, uncontrolled contexts (system prompts, settings, prior turns). This is not a benchmark.