Migrate vllm

functionary/inference_utils.py

@@ -1,3 +1,6 @@
+from copy import deepcopy
+
+import jsonref
 import torch
 from transformers import StoppingCriteria, StoppingCriteriaList
 
@@ -35,3 +38,20 @@ def analyze_tools_and_tool_choice(request):
         tool_func_choice = "none"
 
     return tools_or_functions, tool_func_choice
+
+
+def resolve_json_refs(tools_or_functions):
+    tools = deepcopy(tools_or_functions)
+    if tools:
+        for i in range(len(tools)):
+            if "type" in tools[i]:
+                if tools[i]["type"] == "function":
+                    tools[i]["function"]["parameters"] = deepcopy(
+                        jsonref.JsonRef.replace_refs(tools[i]["function"]["parameters"])
+                    )
+            else:
+                tools[i]["parameters"] = deepcopy(
+                    jsonref.JsonRef.replace_refs(tools[i]["parameters"])
+                )
+
+    return tools

functionary/openai_types.py

@@ -97,11 +97,13 @@ class StreamChoice(BaseModel):
     finish_reason: Optional[str] = "stop"
     index: int = 0
 
+
 class UsageInfo(BaseModel):
     prompt_tokens: int = 0
     total_tokens: int = 0
     completion_tokens: Optional[int] = 0
 
+
 class ChatCompletionChunk(BaseModel):
     id: str
     object: str = "chat.completion.chunk"
@@ -132,7 +134,6 @@ class ChatCompletionRequest(BaseModel):
     best_of: Optional[int] = None
     top_k: Optional[int] = -1
     ignore_eos: Optional[bool] = False
-    use_beam_search: Optional[bool] = False
 
     # @validator("tool_choice", always=True)
     # def validate_tool_choice(cls, value, values):

functionary/prompt_template/base_template.py

@@ -3,10 +3,12 @@
 import json
 import re
 from abc import abstractmethod
+from copy import deepcopy
 from typing import Any, Dict, List, Literal, Optional, Tuple, Union
 
 import jinja2
 
+from functionary.inference_utils import resolve_json_refs
 from functionary.openai_types import Function, Tool
 from functionary.prompt_template import prompt_utils
 
@@ -125,9 +127,11 @@ def get_prompt_from_messages(
             str: the prompt for inference/training
         """
 
+        tools = resolve_json_refs(tools_or_functions=tools_or_functions)
+
         prompt = self._jinja_template.render(
             messages=messages,
-            tools=tools_or_functions,
+            tools=tools,
             bos_token=bos_token,
             add_generation_prompt=add_generation_prompt,
         )

functionary/vllm_inference.py

@@ -198,7 +198,6 @@ async def process_chat_completion(
             best_of=request.best_of,
             top_k=request.top_k,
             ignore_eos=request.ignore_eos,
-            use_beam_search=request.use_beam_search,
             skip_special_tokens=False,
             logprobs=logprobs,
         )
@@ -207,7 +206,7 @@ async def process_chat_completion(
 
     if enable_grammar_sampling:
         result_generator = engine.generate(
-            inputs=TokensPrompt(prompt_token_ids=prompt_token_ids),
+            prompt=TokensPrompt(prompt_token_ids=prompt_token_ids),
             sampling_params=sampling_params,
             request_id=request_id,
             tools_or_functions=tools_or_functions,
@@ -216,7 +215,7 @@ async def process_chat_completion(
         )
     else:
         result_generator = engine.generate(
-            inputs=TokensPrompt(prompt_token_ids=prompt_token_ids),
+            prompt=TokensPrompt(prompt_token_ids=prompt_token_ids),
             sampling_params=sampling_params,
             request_id=request_id,
         )