Hugging Face TEI

¶Rerank

The TeiClient provides methods to rerank a list of documents based on their relevance to a given query. This is useful when you want to reorder search results, prioritize the most relevant content, or refine outputs from an initial retrieval phase.

Tip

Reranking helps improve the quality of search or recommendation systems by using models that assess semantic relevance between a query and documents.

¶Example: Simple Reranking with Scores

The rerank() method returns a list of documents with their indexes and relevance scores.

use Partitech\PhpMistral\Clients\Tei\TeiClient;
use Partitech\PhpMistral\MistralClientException;

// Load TEI rerank endpoint and API key
$teiUrl = getenv('TEI_RERANK_URI');
$apiKey = getenv('TEI_API_KEY');

// Instantiate the TEI client
$client = new TeiClient(apiKey: (string) $apiKey, url: $teiUrl);

try {
    $results = $client->rerank(
        query: 'What is the difference between Deep Learning and Machine Learning?',
        texts: [
            'Deep learning is...',
            'cheese is made of',
            'Deep Learning is not...'
        ]
    );
    
    var_dump($results);
} catch (MistralClientException $e) {
    echo $e->getMessage();
}

¶Output

array(3) {
  [0]=>
  array(2) {
    ["index"]=>
    int(0)
    ["score"]=>
    float(0.83828294)
  }
  [1]=>
  array(2) {
    ["index"]=>
    int(2)
    ["score"]=>
    float(0.2656899)
  }
  [2]=>
  array(2) {
    ["index"]=>
    int(1)
    ["score"]=>
    float(3.734357E-5)
  }
}

• index: Refers to the original index of the document in the texts array.
• score: Indicates the relevance of the document to the query (higher is better).

¶Example: Reranking with Top Results and Content

The getRerankedContent() method allows you to:

• Retrieve top N results.
• Include the content of the reranked documents.

try {
    $topResults = $client->getRerankedContent(
        query: 'What is the difference between Deep Learning and Machine Learning?',
        texts: [
            'Deep learning is...',
            'cheese is made of',
            'Deep Learning is not...'
        ],
        top: 2 // Retrieve only top 2 results
    );

    var_dump($topResults);
} catch (MistralClientException $e) {
    echo $e->getMessage();
}

¶Output

array(2) {
  [0]=>
  array(3) {
    ["index"]=>
    int(0)
    ["score"]=>
    float(0.83828294)
    ["content"]=>
    string(19) "Deep learning is..."
  }
  [1]=>
  array(3) {
    ["index"]=>
    int(2)
    ["score"]=>
    float(0.2656899)
    ["content"]=>
    string(23) "Deep Learning is not..."
  }
}

Note

The document with irrelevant content ("cheese is made of") is excluded since it didn't rank in the top 2.

¶Differences Between rerank() and getRerankedContent()

Tip

Use getRerankedContent() when you want both scores and the actual content of the top results.

¶Use Cases

• Search Engines: Improve ranking of search results based on relevance.
• Question Answering: Prioritize the most relevant passages or documents for a question.
• Recommendation Systems: Refine recommendations by reranking candidates based on user queries.

Important

Always ensure the TEI endpoint used is dedicated to reranking tasks to avoid misconfiguration.