Add provider config and update spice with new models

benchmarks/benchmark_runner.py

@@ -59,7 +59,7 @@ async def grade(to_grade, prompt, model="gpt-4-1106-preview"):
             messages[1]["content"] = messages[1]["content"][:-chars_to_remove]
 
         llm_api_handler = SESSION_CONTEXT.get().llm_api_handler
-        llm_grade = await llm_api_handler.call_llm_api(messages, model, False, ResponseFormat(type="json_object"))
+        llm_grade = await llm_api_handler.call_llm_api(messages, model, None, False, ResponseFormat(type="json_object"))
         content = llm_grade.text
         return json.loads(content)
     except Exception as e:

benchmarks/exercism_practice.py

@@ -65,7 +65,7 @@ async def failure_analysis(exercise_runner, language):
     response = ""
     try:
         llm_api_handler = SESSION_CONTEXT.get().llm_api_handler
-        llm_grade = await llm_api_handler.call_llm_api(messages, model, False)
+        llm_grade = await llm_api_handler.call_llm_api(messages, model, None, False)
         response = llm_grade.text
     except BadRequestError:
         response = "Unable to analyze test case\nreason: too many tokens to analyze"

mentat/agent_handler.py

@@ -47,7 +47,7 @@ async def enable_agent_mode(self):
             ),
         ]
         model = ctx.config.model
-        response = await ctx.llm_api_handler.call_llm_api(messages, model, False)
+        response = await ctx.llm_api_handler.call_llm_api(messages, model, ctx.config.provider, False)
         content = response.text
 
         paths = [Path(path) for path in content.strip().split("\n") if Path(path).exists()]
@@ -81,7 +81,7 @@ async def _determine_commands(self) -> List[str]:
 
         try:
             # TODO: Should this even be a separate call or should we collect commands in the edit call?
-            response = await ctx.llm_api_handler.call_llm_api(messages, model, False)
+            response = await ctx.llm_api_handler.call_llm_api(messages, model, ctx.config.provider, False)
             ctx.cost_tracker.display_last_api_call()
         except BadRequestError as e:
             ctx.stream.send(f"Error accessing OpenAI API: {e.message}", style="error")

mentat/config.py

@@ -4,6 +4,7 @@
 from argparse import ArgumentParser, Namespace
 from json import JSONDecodeError
 from pathlib import Path
+from typing import Optional
 
 import attr
 from attr import converters, validators
@@ -38,14 +39,21 @@ class Config:
         default="gpt-4-0125-preview",
         metadata={"auto_completions": list(known_models.keys())},
     )
+    provider: Optional[str] = attr.field(default=None, metadata={"auto_completions": ["openai", "anthropic", "azure"]})
     feature_selection_model: str = attr.field(
         default="gpt-4-1106-preview",
         metadata={"auto_completions": list(known_models.keys())},
     )
+    feature_selection_provider: Optional[str] = attr.field(
+        default=None, metadata={"auto_completions": ["openai", "anthropic", "azure"]}
+    )
     embedding_model: str = attr.field(
         default="text-embedding-ada-002",
         metadata={"auto_completions": [model.name for model in known_models.values() if model.embedding_model]},
     )
+    embedding_provider: Optional[str] = attr.field(
+        default=None, metadata={"auto_completions": ["openai", "anthropic", "azure"]}
+    )
     temperature: float = attr.field(default=0.2, converter=float, validator=[validators.le(1), validators.ge(0)])
 
     maximum_context: int | None = attr.field(

mentat/conversation.py

@@ -177,6 +177,7 @@ async def _stream_model_response(
         response = await llm_api_handler.call_llm_api(
             messages,
             config.model,
+            config.provider,
             stream=True,
             response_format=parser.response_format(),
         )

mentat/feature_filters/llm_feature_filter.py

@@ -90,6 +90,7 @@ async def filter(
         llm_response = await llm_api_handler.call_llm_api(
             messages=messages,
             model=model,
+            provider=config.feature_selection_provider,
             stream=False,
             response_format=ResponseFormat(type="json_object"),
         )

mentat/llm_api_handler.py

@@ -349,6 +349,7 @@ async def call_llm_api(
         self,
         messages: List[SpiceMessage],
         model: str,
+        provider: Optional[str],
         stream: Literal[False],
         response_format: ResponseFormat = ResponseFormat(type="text"),
     ) -> SpiceResponse:
@@ -359,6 +360,7 @@ async def call_llm_api(
         self,
         messages: List[SpiceMessage],
         model: str,
+        provider: Optional[str],
         stream: Literal[True],
         response_format: ResponseFormat = ResponseFormat(type="text"),
     ) -> StreamingSpiceResponse:
@@ -369,6 +371,7 @@ async def call_llm_api(
         self,
         messages: List[SpiceMessage],
         model: str,
+        provider: Optional[str],
         stream: bool,
         response_format: ResponseFormat = ResponseFormat(type="text"),
     ) -> SpiceResponse | StreamingSpiceResponse:
@@ -386,6 +389,7 @@ async def call_llm_api(
             if not stream:
                 response = await self.spice.get_response(
                     model=model,
+                    provider=provider,
                     messages=messages,
                     temperature=config.temperature,
                     response_format=response_format,  # pyright: ignore
@@ -394,6 +398,7 @@ async def call_llm_api(
             else:
                 response = await self.spice.stream_response(
                     model=model,
+                    provider=provider,
                     messages=messages,
                     temperature=config.temperature,
                     response_format=response_format,  # pyright: ignore
@@ -403,7 +408,8 @@ async def call_llm_api(
 
     @api_guard
     def call_embedding_api(self, input_texts: list[str], model: str = "text-embedding-ada-002") -> Embeddings:
-        return self.spice.get_embeddings_sync(input_texts, model)  # pyright: ignore
+        ctx = SESSION_CONTEXT.get()
+        return self.spice.get_embeddings_sync(input_texts, model, provider=ctx.config.embedding_provider)  # pyright: ignore
 
     @api_guard
     async def call_whisper_api(self, audio_path: Path) -> str:

mentat/revisor/revisor.py

@@ -66,7 +66,9 @@ async def revise_edit(file_edit: FileEdit):
         "\nRevising edits for file" f" {get_relative_path(file_edit.file_path, ctx.cwd)}...",
         style="info",
     )
-    response = await ctx.llm_api_handler.call_llm_api(messages, model=ctx.config.model, stream=False)
+    response = await ctx.llm_api_handler.call_llm_api(
+        messages, model=ctx.config.model, provider=ctx.config.provider, stream=False
+    )
     message = response.text
     messages.append(ChatCompletionAssistantMessageParam(content=message, role="assistant"))
     ctx.conversation.add_transcript_message(

requirements.txt

@@ -21,7 +21,7 @@ selenium==4.15.2
 sentry-sdk==1.34.0
 sounddevice==0.4.6
 soundfile==0.12.1
-spiceai==0.1.8
+spiceai==0.1.9
 termcolor==2.3.0
 textual==0.47.1
 textual-autocomplete==2.1.0b0

tests/conftest.py

@@ -98,7 +98,7 @@ def mock_call_llm_api(mocker):
     completion_mock = mocker.patch.object(LlmApiHandler, "call_llm_api")
 
     def wrap_unstreamed_string(value):
-        return SpiceResponse(SpiceCallArgs("gpt-3.5-turbo", [], False), value, 1, 0, 0, True)
+        return SpiceResponse(SpiceCallArgs("gpt-3.5-turbo", [], False), value, 1, 0, 0, True, 1)
 
     def wrap_streamed_strings(values):
         class MockStreamingSpiceResponse:
@@ -115,7 +115,7 @@ async def __anext__(self):
                 return values[self.cur_value - 1]
 
             def current_response(self):
-                return SpiceResponse(SpiceCallArgs("gpt-3.5-turbo", [], True), "".join(values), 1, 0, 0, True)
+                return SpiceResponse(SpiceCallArgs("gpt-3.5-turbo", [], True), "".join(values), 1, 0, 0, True, 1)
 
         mock_spice_response = MockStreamingSpiceResponse()
         return mock_spice_response
@@ -131,7 +131,7 @@ def set_unstreamed_values(value):
     completion_mock.set_unstreamed_values = set_unstreamed_values
 
     def set_return_values(values):
-        async def call_llm_api_mock(messages, model, stream, response_format="unused"):
+        async def call_llm_api_mock(messages, model, provider, stream, response_format="unused"):
             value = call_llm_api_mock.values.pop()
             if stream:
                 return wrap_streamed_strings([value])