Fixes #429 (ignore chunks with empty choices array) (#430)

README.md

@@ -70,6 +70,8 @@ You'll need to have API access to GPT-4 to run Mentat. There are a few options t
 
 Mentat also works with the Azure OpenAI API. To use the Azure API, provide the `AZURE_OPENAI_ENDPOINT` (`https://<your-instance-name>.openai.azure.com/`) and `AZURE_OPENAI_KEY` environment variables instead of `OPENAI_API_KEY`.
 
+In addition, Mentat uses the `gpt-4-1106-preview` by default. On Azure, this model is available under a different name: `gpt-4-1106-Preview` (with a capital P). To use it, override the default model as described in [configuration.md](docs/configuration.md).
+
 > **_Important:_** Due to changes in the OpenAI Python SDK, you can no longer use `OPENAI_API_BASE` to access the Azure API with Mentat.
 
 ## Configuration

mentat/cost_tracker.py

@@ -77,6 +77,9 @@ async def response_logger_wrapper(
         full_response = ""
         start_time = default_timer()
         async for chunk in response:
+            # On Azure OpenAI, the first chunk streamed may contain only metadata relating to content filtering.
+            if len(chunk.choices) == 0:
+                continue
             full_response += chunk.choices[0].delta.content or ""
             yield chunk
         time_elapsed = default_timer() - start_time

mentat/llm_api_handler.py

@@ -80,7 +80,9 @@ async def wrapper(*args: Any, **kwargs: Any) -> Any:
 
 # Ensures that each chunk will have at most one newline character
 def chunk_to_lines(chunk: ChatCompletionChunk) -> list[str]:
-    content = chunk.choices[0].delta.content
+    content = None
+    if len(chunk.choices) > 0:
+        content = chunk.choices[0].delta.content
     return ("" if content is None else content).splitlines(keepends=True)