Merge branch 'main' into add-watsonx-inference-adapter

llama_stack/models/llama/datatypes.py

@@ -47,7 +47,14 @@ RecursiveType = Union[Primitive, List[Primitive], Dict[str, Primitive]]
 class ToolCall(BaseModel):
     call_id: str
     tool_name: Union[BuiltinTool, str]
-    arguments: Dict[str, RecursiveType]
+    # Plan is to deprecate the Dict in favor of a JSON string
+    # that is parsed on the client side instead of trying to manage
+    # the recursive type here.
+    # Making this a union so that client side can start prepping for this change.
+    # Eventually, we will remove both the Dict and arguments_json field,
+    # and arguments will just be a str
+    arguments: Union[str, Dict[str, RecursiveType]]
+    arguments_json: Optional[str] = None
 
     @field_validator("tool_name", mode="before")
     @classmethod
@@ -179,13 +186,11 @@ class TopKSamplingStrategy(BaseModel):
     top_k: int = Field(..., ge=1)
 
 
-SamplingStrategy = register_schema(
-    Annotated[
-        Union[GreedySamplingStrategy, TopPSamplingStrategy, TopKSamplingStrategy],
-        Field(discriminator="type"),
-    ],
-    name="SamplingStrategy",
-)
+SamplingStrategy = Annotated[
+    Union[GreedySamplingStrategy, TopPSamplingStrategy, TopKSamplingStrategy],
+    Field(discriminator="type"),
+]
+register_schema(SamplingStrategy, name="SamplingStrategy")
 
 
 @json_schema_type

llama_stack/models/llama/llama3/chat_format.py

@@ -12,6 +12,7 @@
 # the top-level of this source tree.
 
 import io
+import json
 import uuid
 from dataclasses import dataclass
 from typing import Dict, List, Optional, Tuple
@@ -203,9 +204,10 @@ class ChatFormat:
             # This code tries to handle that case
             if tool_name in BuiltinTool.__members__:
                 tool_name = BuiltinTool[tool_name]
-                tool_arguments = {
-                    "query": list(tool_arguments.values())[0],
-                }
+                if isinstance(tool_arguments, dict):
+                    tool_arguments = {
+                        "query": list(tool_arguments.values())[0],
+                    }
         else:
             builtin_tool_info = ToolUtils.maybe_extract_builtin_tool_call(content)
             if builtin_tool_info is not None:
@@ -229,6 +231,7 @@ class ChatFormat:
                     call_id=call_id,
                     tool_name=tool_name,
                     arguments=tool_arguments,
+                    arguments_json=json.dumps(tool_arguments),
                 )
             )
             content = ""

llama_stack/models/llama/llama3/prompt_templates/system_prompts.py

@@ -34,7 +34,9 @@ class SystemDefaultGenerator(PromptTemplateGeneratorBase):
         )
         return PromptTemplate(
             template_str.lstrip("\n"),
-            {"today": datetime.now().strftime("%d %B %Y")},
+            {
+                "today": datetime.now().strftime("%d %B %Y")  # noqa: DTZ005 - we don't care about timezones here since we are displaying the date
+            },
         )
 
     def data_examples(self) -> List[Any]:

llama_stack/models/llama/llama3/template_data.py

@@ -11,11 +11,8 @@
 # top-level folder for each specific model found within the models/ directory at
 # the top-level of this source tree.
 
-from llama_stack.models.llama.datatypes import (
-    BuiltinTool,
-    StopReason,
-    ToolCall,
-)
+
+from llama_stack.models.llama.datatypes import BuiltinTool, StopReason, ToolCall
 
 from .prompt_templates import (
     BuiltinToolGenerator,

llama_stack/models/llama/llama3/tool_utils.py

@@ -15,8 +15,11 @@ import json
 import re
 from typing import Optional, Tuple
 
+from llama_stack.log import get_logger
 from llama_stack.models.llama.datatypes import BuiltinTool, RecursiveType, ToolCall, ToolPromptFormat
 
+logger = get_logger(name=__name__, category="inference")
+
 BUILTIN_TOOL_PATTERN = r'\b(?P<tool_name>\w+)\.call\(query="(?P<query>[^"]*)"\)'
 CUSTOM_TOOL_CALL_PATTERN = re.compile(r"<function=(?P<function_name>[^}]+)>(?P<args>{.*?})")
 
@@ -92,7 +95,15 @@ def parse_python_list_for_function_calls(input_string):
 
             # Extract keyword arguments
             for keyword in node.keywords:
-                function_args[keyword.arg] = ast.literal_eval(keyword.value)
+                try:
+                    function_args[keyword.arg] = ast.literal_eval(keyword.value)
+                except ValueError as e:
+                    logger.error(
+                        f"Error parsing tool call argument '{keyword.arg}': {e}, full input string: '{input_string}'"
+                    )
+                    raise ValueError(
+                        f"Error parsing tool call argument '{keyword.arg}', full input string: '{input_string}'"
+                    ) from e
 
             result.append((function_name, function_args))