chore: remove embeddings request

lib/evagg/llm/aoai.py

@@ -7,7 +7,6 @@
 
 import openai
 from openai import AsyncAzureOpenAI, AsyncOpenAI
-from openai.types import CreateEmbeddingResponse
 from openai.types.chat import (
     ChatCompletionMessageParam,
     ChatCompletionSystemMessageParam,
@@ -204,41 +203,6 @@ async def prompt_file(
         user_prompt = self._load_prompt_file(user_prompt_file)
         return await self.prompt(user_prompt, system_prompt, params, prompt_settings)
 
-    async def embeddings(
-        self, inputs: List[str], embedding_settings: Optional[Dict[str, Any]] = None
-    ) -> Dict[str, List[float]]:
-        settings = {"model": "text-embedding-ada-002-v2", **(embedding_settings or {})}
-
-        embeddings = {}
-
-        async def _run_single_embedding(input: str) -> int:
-            connection_errors = 0
-            while True:
-                try:
-                    result: CreateEmbeddingResponse = await self._client.embeddings.create(
-                        input=[input], encoding_format="float", **settings
-                    )
-                    embeddings[input] = result.data[0].embedding
-                    return result.usage.prompt_tokens
-                except (openai.RateLimitError, openai.InternalServerError) as e:
-                    logger.warning(f"Rate limit error on embeddings: {e}")
-                    await asyncio.sleep(1)
-                except (openai.APIConnectionError, openai.APITimeoutError):
-                    if connection_errors > 2:
-                        if hasattr(self._config, "endpoint") and self._config.endpoint.startswith("http://localhost"):
-                            logger.error("Azure OpenAI API unreachable - have failed to start a local proxy?")
-                        raise
-                    logger.warning("Connectivity error on embeddings, retrying...")
-                    connection_errors += 1
-                    await asyncio.sleep(1)
-
-        start_overall = time.time()
-        tokens = await asyncio.gather(*[_run_single_embedding(input) for input in inputs])
-        elapsed = time.time() - start_overall
-
-        logger.info(f"{len(inputs)} embeddings produced in {elapsed:.1f} seconds using {sum(tokens)} tokens.")
-        return embeddings
-
 
 class OpenAICacheClient(OpenAIClient):
     def __init__(self, client_class: str, config: Dict[str, Any]) -> None:

lib/evagg/llm/interfaces.py

@@ -1,4 +1,4 @@
-from typing import Any, Dict, List, Optional, Protocol
+from typing import Any, Dict, Optional, Protocol
 
 
 class IPromptClient(Protocol):
@@ -21,9 +21,3 @@ async def prompt_file(
     ) -> str:
         """Get the response from a prompt with an input file."""
         ...  # pragma: no cover
-
-    async def embeddings(
-        self, inputs: List[str], embedding_settings: Optional[Dict[str, Any]] = None
-    ) -> Dict[str, List[float]]:
-        """Get embeddings for the given inputs."""
-        ...  # pragma: no cover

test/evagg/test_llm.py

@@ -44,27 +44,3 @@ async def test_openai_client_prompt(mock_openai, test_file_contents) -> None:
         temperature=1.5,
         model="gpt-8",
     )
-
-
-@patch("lib.evagg.llm.aoai.AsyncAzureOpenAI", return_value=AsyncMock())
-async def test_openai_client_embeddings(mock_openai) -> None:
-    embedding = MagicMock(data=[MagicMock(embedding=[0.4, 0.5, 0.6])], usage=MagicMock(prompt_tokens=10))
-    mock_openai.return_value.embeddings.create.return_value = embedding
-
-    inputs = [f"input_{i}" for i in range(1)]
-    client = OpenAIClient(
-        "AsyncAzureOpenAI",
-        {
-            "deployment": "gpt-8",
-            "endpoint": "https://ai",
-            "api_key": "test",
-            "api_version": "test",
-            "timeout": 60,
-        },
-    )
-    response = await client.embeddings(inputs)
-    mock_openai.assert_called_once_with(azure_endpoint="https://ai", api_key="test", api_version="test", timeout=60)
-    mock_openai.return_value.embeddings.create.assert_has_calls(
-        [call(input=[input], encoding_format="float", model="text-embedding-ada-002-v2") for input in inputs]
-    )
-    assert response == {input: [0.4, 0.5, 0.6] for input in inputs}