Add standard call to LiteLLM engine

examples/tool_calling_agent_from_any_llm.py

@@ -6,7 +6,7 @@ from smolagents import tool, HfApiEngine, OpenAIEngine, AnthropicEngine, Transfo
 # llm_engine = AnthropicEngine("claude-3-5-sonnet-20240620")
 # llm_engine = HfApiEngine("meta-llama/Llama-3.3-70B-Instruct")
 # llm_engine = TransformersEngine("meta-llama/Llama-3.2-2B-Instruct")
-llm_engine = LiteLLMEngine()
+llm_engine = LiteLLMEngine("gpt-4o")
 
 @tool
 def get_weather(location: str) -> str:

src/smolagents/llm_engines.py

@@ -19,6 +19,7 @@ from enum import Enum
 from typing import Dict, List, Optional, Tuple
 from transformers import AutoTokenizer, AutoModelForCausalLM, StoppingCriteria, StoppingCriteriaList
 
+import litellm
 import logging
 import os
 import random
@@ -566,11 +567,32 @@ class AnthropicEngine:
 class LiteLLMEngine():
     def __init__(self, model_id="anthropic/claude-3-5-sonnet-20240620"):
         self.model_id = model_id
-        import os, litellm
-
         # IMPORTANT - Set this to TRUE to add the function to the prompt for Non OpenAI LLMs
         litellm.add_function_to_prompt = True
-    
+        self.last_input_token_count = 0
+        self.last_output_token_count = 0
+
+    def __call__(
+        self,
+        messages: List[Dict[str, str]],
+        stop_sequences: Optional[List[str]] = None,
+        grammar: Optional[str] = None,
+        max_tokens: int = 1500,
+    ) -> str:
+        messages = get_clean_message_list(
+            messages, role_conversions=tool_role_conversions
+        )
+
+        response = litellm.completion(
+            model=self.model_id,
+            messages=messages,
+            stop=stop_sequences,
+            max_tokens=max_tokens,
+        )
+        self.last_input_token_count = response.usage.prompt_tokens
+        self.last_output_token_count = response.usage.completion_tokens
+        return response.choices[0].message.content
+
     def get_tool_call(
             self,
             messages: List[Dict[str, str]],
@@ -578,19 +600,20 @@ class LiteLLMEngine():
             stop_sequences: Optional[List[str]] = None,
             max_tokens: int = 1500,
         ):
-        from litellm import completion
         messages = get_clean_message_list(
             messages, role_conversions=tool_role_conversions
         )
-        response = completion(
+        response = litellm.completion(
             model=self.model_id,
             messages=messages,
             tools=[get_json_schema(tool) for tool in available_tools],
             tool_choice="required",
-            max_tokens=max_tokens,
             stop=stop_sequences,
+            max_tokens=max_tokens,
         )
         tool_calls = response.choices[0].message.tool_calls[0]
+        self.last_input_token_count = response.usage.prompt_tokens
+        self.last_output_token_count = response.usage.completion_tokens
         return tool_calls.function.name, tool_calls.function.arguments, tool_calls.id