[ENH] XAI embedding function

chromadb/utils/embedding_functions/xai_embedding_function.py

@@ -0,0 +1,63 @@
+import logging
+from typing import List, cast, TypedDict
+
+import httpx
+
+from chromadb.api.types import Documents, EmbeddingFunction, Embeddings, Embedding
+
+logger = logging.getLogger(__name__)
+
+class XAIEmbedding(TypedDict):
+    Float: List[float]
+
+class XAIResponseItem(TypedDict):
+    embedding: Embedding
+    index: int
+    object: str
+
+
+class XAIEmbeddingFunction(EmbeddingFunction[Documents]):
+    """
+    This class is used to get embeddings for a list of texts using the XAI API.
+    It requires an API key and a model name. You can use the "list embedding models" endpoint
+    to verify what embeddings models are available for your API key.
+    """
+
+    def __init__(self, api_key: str, model_name: str):
+        """
+        Initialize the XAIEmbeddingFunction.
+
+        Args:
+            api_key (str): Your API key for the XAI API.
+            model_name (str, optional): The name of the model to use for embeddings.
+        """
+        self._model_name = model_name
+        self._api_url = "https://api.x.ai/v1/embeddings"
+        self._session = httpx.Client()
+        self._session.headers.update(
+            {"Authorization": f"Bearer {api_key}", "Accept-Encoding": "identity"}
+        )
+
+    def __call__(self, input: Documents) -> Embeddings:
+        """
+        Get the embeddings for a list of texts.
+
+        Args:
+            texts (Documents): A list of texts to get embeddings for.
+
+        Returns:
+            Embeddings: The embeddings for the texts.
+        """
+        resp = self._session.post(
+            self._api_url, json={"input": input, "model": self._model_name, "encoding_format": "float"}
+        ).json()
+        if "data" not in resp:
+            raise RuntimeError(resp["error"])
+
+        embeddings: List[XAIResponseItem] = resp["data"]
+
+        # Sort resulting embeddings by index
+        sorted_embeddings = sorted(embeddings, key=lambda e: e["index"])
+
+        # Return just the embeddings
+        return cast(Embeddings, [result["embedding"]["Float"] for result in sorted_embeddings])