feat: LiteLLM backend

mellea/backends/litellm.py

@@ -0,0 +1,354 @@
+"""A generic LiteLLM compatible backend that wraps around the openai python sdk."""
+
+import datetime
+import json
+from collections.abc import Callable
+from typing import Any
+
+import litellm
+import litellm.litellm_core_utils
+import litellm.litellm_core_utils.get_supported_openai_params
+
+import mellea.backends.model_ids as model_ids
+from mellea.backends import BaseModelSubclass
+from mellea.backends.formatter import Formatter, FormatterBackend, TemplateFormatter
+from mellea.backends.tools import (
+    add_tools_from_context_actions,
+    add_tools_from_model_options,
+    convert_tools_to_json,
+)
+from mellea.backends.types import ModelOption
+from mellea.helpers.fancy_logger import FancyLogger
+from mellea.stdlib.base import (
+    CBlock,
+    Component,
+    Context,
+    GenerateLog,
+    ModelOutputThunk,
+    ModelToolCall,
+)
+from mellea.stdlib.chat import Message
+from mellea.stdlib.requirement import ALoraRequirement
+
+
+class LiteLLMBackend(FormatterBackend):
+    """A generic LiteLLM compatible backend."""
+
+    def __init__(
+        self,
+        model_id: str = "ollama/" + str(model_ids.IBM_GRANITE_3_3_8B.ollama_name),
+        formatter: Formatter | None = None,
+        base_url: str | None = "http://localhost:11434",
+        model_options: dict | None = None,
+    ):
+        """Initialize and OpenAI compatible backend. For any additional kwargs that you need to pass the the client, pass them as a part of **kwargs.
+
+        Args:
+            model_id : The LiteLLM model identifier. Make sure that all necessary credentials are in OS environment variables.
+            formatter: A custom formatter based on backend.If None, defaults to TemplateFormatter
+            base_url : Base url for LLM API. Defaults to None.
+            model_options : Generation options to pass to the LLM. Defaults to None.
+        """
+        super().__init__(
+            model_id=model_id,
+            formatter=(
+                formatter
+                if formatter is not None
+                else TemplateFormatter(model_id=model_id)
+            ),
+            model_options=model_options,
+        )
+
+        assert isinstance(model_id, str), "Model ID must be a string."
+        self._model_id = model_id
+
+        if base_url is None:
+            self._base_url = "http://localhost:11434/v1"  # ollama
+        else:
+            self._base_url = base_url
+
+        # A mapping of common options for this backend mapped to their Mellea ModelOptions equivalent.
+        # These are usually values that must be extracted before hand or that are common among backend providers.
+        # OpenAI has some deprecated parameters. Those map to the same mellea parameter, but
+        # users should only be specifying a single one in their request.
+        self.to_mellea_model_opts_map = {
+            "system": ModelOption.SYSTEM_PROMPT,
+            "reasoning_effort": ModelOption.THINKING,  # TODO: JAL; see which of these are actually extracted...
+            "seed": ModelOption.SEED,
+            "max_completion_tokens": ModelOption.MAX_NEW_TOKENS,
+            "max_tokens": ModelOption.MAX_NEW_TOKENS,
+            "tools": ModelOption.TOOLS,
+            "functions": ModelOption.TOOLS,
+        }
+
+        # A mapping of Mellea specific ModelOptions to the specific names for this backend.
+        # These options should almost always be a subset of those specified in the `to_mellea_model_opts_map`.
+        # Usually, values that are intentionally extracted while prepping for the backend generate call
+        # will be omitted here so that they will be removed when model_options are processed
+        # for the call to the model.
+        self.from_mellea_model_opts_map = {
+            ModelOption.SEED: "seed",
+            ModelOption.MAX_NEW_TOKENS: "max_completion_tokens",
+        }
+
+    def generate_from_context(
+        self,
+        action: Component | CBlock,
+        ctx: Context,
+        *,
+        format: type[BaseModelSubclass] | None = None,
+        model_options: dict | None = None,
+        generate_logs: list[GenerateLog] | None = None,
+        tool_calls: bool = False,
+    ):
+        """See `generate_from_chat_context`."""
+        assert ctx.is_chat_context, NotImplementedError(
+            "The Openai backend only supports chat-like contexts."
+        )
+        return self._generate_from_chat_context_standard(
+            action,
+            ctx,
+            format=format,
+            model_options=model_options,
+            generate_logs=generate_logs,
+            tool_calls=tool_calls,
+        )
+
+    def _simplify_and_merge(
+        self, model_options: dict[str, Any] | None
+    ) -> dict[str, Any]:
+        """Simplifies model_options to use the Mellea specific ModelOption.Option and merges the backend's model_options with those passed into this call.
+
+        Rules:
+        - Within a model_options dict, existing keys take precedence. This means remapping to mellea specific keys will maintain the value of the mellea specific key if one already exists.
+        - When merging, the keys/values from the dictionary passed into this function take precedence.
+
+        Because this function simplifies and then merges, non-Mellea keys from the passed in model_options will replace
+        Mellea specific keys from the backend's model_options.
+
+        Args:
+            model_options: the model_options for this call
+
+        Returns:
+            a new dict
+        """
+        backend_model_opts = ModelOption.replace_keys(
+            self.model_options, self.to_mellea_model_opts_map
+        )
+
+        if model_options is None:
+            return backend_model_opts
+
+        generate_call_model_opts = ModelOption.replace_keys(
+            model_options, self.to_mellea_model_opts_map
+        )
+        return ModelOption.merge_model_options(
+            backend_model_opts, generate_call_model_opts
+        )
+
+    def _make_backend_specific_and_remove(
+        self, model_options: dict[str, Any]
+    ) -> dict[str, Any]:
+        """Maps specified Mellea specific keys to their backend specific version and removes any remaining Mellea keys.
+
+        Additionally, logs any params unknown to litellm and any params that are openai specific but not supported by this model/provider.
+
+        Args:
+            model_options: the model_options for this call
+
+        Returns:
+            a new dict
+        """
+        backend_specific = ModelOption.replace_keys(
+            model_options, self.from_mellea_model_opts_map
+        )
+        backend_specific = ModelOption.remove_special_keys(backend_specific)
+
+        # We set `drop_params=True` which will drop non-supported openai params; check for non-openai
+        # params that might cause errors and log which openai params aren't supported here.
+        # See https://docs.litellm.ai/docs/completion/input.
+        # standard_openai_subset = litellm.get_standard_openai_params(backend_specific)
+        supported_params_list = litellm.litellm_core_utils.get_supported_openai_params.get_supported_openai_params(
+            self._model_id
+        )
+        supported_params = (
+            set(supported_params_list) if supported_params_list is not None else set()
+        )
+
+        # unknown_keys = []  # keys that are unknown to litellm
+        unsupported_openai_params = []  # openai params that are known to litellm but not supported for this model/provider
+        for key in backend_specific.keys():
+            if key not in supported_params:
+                unsupported_openai_params.append(key)
+
+        # if len(unknown_keys) > 0:
+        #     FancyLogger.get_logger().warning(
+        #         f"litellm allows for unknown / non-openai input params; mellea won't validate the following params that may cause issues: {', '.join(unknown_keys)}"
+        #     )
+
+        if len(unsupported_openai_params) > 0:
+            FancyLogger.get_logger().warning(
+                f"litellm will automatically drop the following openai keys that aren't supported by the current model/provider: {', '.join(unsupported_openai_params)}"
+            )
+            for key in unsupported_openai_params:
+                del backend_specific[key]
+
+        return backend_specific
+
+    def _generate_from_chat_context_standard(
+        self,
+        action: Component | CBlock,
+        ctx: Context,
+        *,
+        format: type[BaseModelSubclass]
+        | None = None,  # Type[BaseModelSubclass] is a class object of a subclass of BaseModel
+        model_options: dict | None = None,
+        generate_logs: list[GenerateLog] | None = None,
+        tool_calls: bool = False,
+    ) -> ModelOutputThunk:
+        model_opts = self._simplify_and_merge(model_options)
+        linearized_context = ctx.render_for_generation()
+        assert linearized_context is not None, (
+            "Cannot generate from a non-linear context in a FormatterBackend."
+        )
+        # Convert our linearized context into a sequence of chat messages. Template formatters have a standard way of doing this.
+        messages: list[Message] = self.formatter.to_chat_messages(linearized_context)
+        # Add the final message.
+        match action:
+            case ALoraRequirement():
+                raise Exception("The LiteLLM backend does not support activated LoRAs.")
+            case _:
+                messages.extend(self.formatter.to_chat_messages([action]))
+
+        conversation: list[dict] = []
+        system_prompt = model_opts.get(ModelOption.SYSTEM_PROMPT, "")
+        if system_prompt != "":
+            conversation.append({"role": "system", "content": system_prompt})
+        conversation.extend([{"role": m.role, "content": m.content} for m in messages])
+
+        if format is not None:
+            response_format = {
+                "type": "json_schema",
+                "json_schema": {
+                    "name": format.__name__,
+                    "schema": format.model_json_schema(),
+                    "strict": True,
+                },
+            }
+        else:
+            response_format = {"type": "text"}
+
+        thinking = model_opts.get(ModelOption.THINKING, None)
+        if type(thinking) is bool and thinking:
+            # OpenAI uses strings for its reasoning levels.
+            thinking = "medium"
+
+        # Append tool call information if applicable.
+        tools = self._extract_tools(action, format, model_opts, tool_calls, ctx)
+        formatted_tools = convert_tools_to_json(tools) if len(tools) > 0 else None
+
+        model_specific_options = self._make_backend_specific_and_remove(model_opts)
+
+        chat_response: litellm.ModelResponse = litellm.completion(
+            model=self._model_id,
+            messages=conversation,
+            tools=formatted_tools,
+            response_format=response_format,
+            reasoning_effort=thinking,  # type: ignore
+            drop_params=True,  # See note in `_make_backend_specific_and_remove`.
+            **model_specific_options,
+        )
+
+        choice_0 = chat_response.choices[0]
+        assert isinstance(choice_0, litellm.utils.Choices), (
+            "Only works for non-streaming response for now"
+        )
+        result = ModelOutputThunk(
+            value=choice_0.message.content,
+            meta={
+                "litellm_chat_response": chat_response.choices[0].model_dump()
+            },  # NOTE: Using model dump here to comply with `TemplateFormatter`
+            tool_calls=self._extract_model_tool_requests(tools, chat_response),
+        )
+
+        parsed_result = self.formatter.parse(source_component=action, result=result)
+
+        if generate_logs is not None:
+            assert isinstance(generate_logs, list)
+            generate_log = GenerateLog()
+            generate_log.prompt = conversation
+            generate_log.backend = f"litellm::{self.model_id!s}"
+            generate_log.model_options = model_specific_options
+            generate_log.date = datetime.datetime.now()
+            generate_log.model_output = chat_response
+            generate_log.extra = {
+                "format": format,
+                "tools_available": tools,
+                "tools_called": result.tool_calls,
+                "seed": model_opts.get("seed", None),
+            }
+            generate_log.action = action
+            generate_log.result = parsed_result
+            generate_logs.append(generate_log)
+
+        return parsed_result
+
+    @staticmethod
+    def _extract_tools(
+        action, format, model_opts, tool_calls, ctx
+    ) -> dict[str, Callable]:
+        tools: dict[str, Callable] = dict()
+        if tool_calls:
+            if format:
+                FancyLogger.get_logger().warning(
+                    f"Tool calling typically uses constrained generation, but you have specified a `format` in your generate call. NB: tool calling is superseded by format; we will NOT call tools for your request: {action}"
+                )
+            else:
+                add_tools_from_model_options(tools, model_opts)
+                add_tools_from_context_actions(tools, ctx.actions_for_available_tools())
+
+                # Add the tools from the action for this generation last so that
+                # they overwrite conflicting names.
+                add_tools_from_context_actions(tools, [action])
+            FancyLogger.get_logger().info(f"Tools for call: {tools.keys()}")
+        return tools
+
+    def _generate_from_raw(
+        self,
+        actions: list[Component | CBlock],
+        *,
+        format: type[BaseModelSubclass] | None = None,
+        model_options: dict | None = None,
+        generate_logs: list[GenerateLog] | None = None,
+    ) -> list[ModelOutputThunk]:
+        """Generate using the completions api. Gives the input provided to the model without templating."""
+        raise NotImplementedError("This method is not implemented yet.")
+
+    def _extract_model_tool_requests(
+        self, tools: dict[str, Callable], chat_response: litellm.ModelResponse
+    ) -> dict[str, ModelToolCall] | None:
+        model_tool_calls: dict[str, ModelToolCall] = {}
+        choice_0 = chat_response.choices[0]
+        assert isinstance(choice_0, litellm.utils.Choices), (
+            "Only works for non-streaming response for now"
+        )
+        calls = choice_0.message.tool_calls
+        if calls:
+            for tool_call in calls:
+                tool_name = str(tool_call.function.name)
+                tool_args = tool_call.function.arguments
+
+                func = tools.get(tool_name)
+                if func is None:
+                    FancyLogger.get_logger().warning(
+                        f"model attempted to call a non-existing function: {tool_name}"
+                    )
+                    continue  # skip this function if we can't find it.
+
+                # Returns the args as a string. Parse it here.
+                args = json.loads(tool_args)
+                model_tool_calls[tool_name] = ModelToolCall(tool_name, func, args)
+
+        if len(model_tool_calls) > 0:
+            return model_tool_calls
+        return None

pyproject.toml

@@ -62,14 +62,18 @@ hf = [
     "trl>=0.19.0",
 ]
 
+litellm = [
+    "litellm>=1.76"
+]
+
 watsonx = [
     "ibm-watsonx-ai>=1.3.31",
 ]
 docling = [
     "docling>=2.45.0",
 ]
 
-all = ["mellea[watsonx,docling,hf]"]
+all = ["mellea[watsonx,docling,hf,litellm]"]
 
 [dependency-groups]
 # Use these like:
@@ -140,7 +144,7 @@ ignore = [
 #   "UP006",  # List vs list, etc
 #   "UP007",  # Option and Union
 #   "UP035",  # `typing.Set` is deprecated, use `set` instead"
-  "PD901",  # Avoid using the generic variable name `df` for DataFrames 
+  "PD901",  # Avoid using the generic variable name `df` for DataFrames
 ]
 
 [tool.ruff.lint.pydocstyle]