Updated `LocalAIEmbeddings` docstring to better explain why `openai` (#10946)

Fixes my misgivings in
https://github.com/langchain-ai/langchain/issues/10912

libs/langchain/langchain/embeddings/localai.py

@@ -120,16 +120,19 @@ async def async_embed_with_retry(embeddings: LocalAIEmbeddings, **kwargs: Any) -
 class LocalAIEmbeddings(BaseModel, Embeddings):
     """LocalAI embedding models.
 
-    To use, you should have the ``openai`` python package installed, and the
+    Since LocalAI and OpenAI have 1:1 compatibility between APIs, this class
-    environment variable ``OPENAI_API_KEY`` set to a random string. You need to
+    uses the ``openai`` Python package's ``openai.Embedding`` as its client.
-    specify ``OPENAI_API_BASE`` to point to your LocalAI service endpoint.
+    Thus, you should have the ``openai`` python package installed, and defeat
+    the environment variable ``OPENAI_API_KEY`` by setting to a random string.
+    You also need to specify ``OPENAI_API_BASE`` to point to your LocalAI
+    service endpoint.
 
     Example:
         .. code-block:: python
 
             from langchain.embeddings import LocalAIEmbeddings
             openai = LocalAIEmbeddings(
-                openai_api_key="random-key",
+                openai_api_key="random-string",
                 openai_api_base="http://localhost:8080"
             )