|
| 1 | +# Copyright 2025 Google LLC |
| 2 | +# |
| 3 | +# Licensed under the Apache License, Version 2.0 (the "License"); |
| 4 | +# you may not use this file except in compliance with the License. |
| 5 | +# You may obtain a copy of the License at |
| 6 | +# |
| 7 | +# http://www.apache.org/licenses/LICENSE-2.0 |
| 8 | +# |
| 9 | +# Unless required by applicable law or agreed to in writing, software |
| 10 | +# distributed under the License is distributed on an "AS IS" BASIS, |
| 11 | +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. |
| 12 | +# See the License for the specific language governing permissions and |
| 13 | +# limitations under the License. |
| 14 | + |
| 15 | + |
| 16 | +from typing import Any, Optional |
| 17 | + |
| 18 | +from google.genai import types as genai_types |
| 19 | +from pydantic import BaseModel |
| 20 | +from pydantic import Field |
| 21 | + |
| 22 | + |
| 23 | +class IntermediateData(BaseModel): |
| 24 | + """Container for intermediate data that an agent would generate as it responds with a final answer.""" |
| 25 | + |
| 26 | + tool_uses: list[genai_types.FunctionCall] |
| 27 | + """Tool use trajectory in chronological order.""" |
| 28 | + |
| 29 | + intermediate_responses: list[genai_types.Part] |
| 30 | + """Intermediate responses generated by sub-agents to convey progress or status |
| 31 | + in a multi-agent system, distinct from the final response.""" |
| 32 | + |
| 33 | + |
| 34 | +class Invocation(BaseModel): |
| 35 | + """Represents a single invocation.""" |
| 36 | + |
| 37 | + invocation_id: str = '' |
| 38 | + """Unique identifier for the invocation.""" |
| 39 | + |
| 40 | + user_content: genai_types.Content |
| 41 | + """Content provided by the user in this invocation.""" |
| 42 | + |
| 43 | + final_response: Optional[genai_types.Content] |
| 44 | + """Final response from the agent that acts a reference or benchmark.""" |
| 45 | + |
| 46 | + intermediate_data: IntermediateData |
| 47 | + """Reference intermediate steps generated as a part of Agent execution. |
| 48 | +
|
| 49 | + For a multi-agent system, it is also helpful to inspect the route that |
| 50 | + the agent took to generate final response. |
| 51 | + """ |
| 52 | + |
| 53 | + creation_timestamp: float = 0.0 |
| 54 | + """Timestamp for the current invocation, primarily intended for debugging purposes.""" |
| 55 | + |
| 56 | + |
| 57 | +class SessionInput(BaseModel): |
| 58 | + """Values that help initialize a Session.""" |
| 59 | + |
| 60 | + app_name: str |
| 61 | + """The name of the app.""" |
| 62 | + |
| 63 | + user_id: str |
| 64 | + """The user id.""" |
| 65 | + |
| 66 | + state: dict[str, Any] = Field(default_factory=dict) |
| 67 | + """The state of the session.""" |
| 68 | + |
| 69 | + |
| 70 | +class EvalCase(BaseModel): |
| 71 | + """An eval case.""" |
| 72 | + |
| 73 | + eval_id: str |
| 74 | + """Unique identifier for the evaluation case.""" |
| 75 | + |
| 76 | + conversation: list[Invocation] |
| 77 | + """A conversation between the user and the Agent. The conversation can have any number of invocations.""" |
| 78 | + |
| 79 | + session_input: SessionInput |
| 80 | + """Session input that will be passed on to the Agent during eval. |
| 81 | + It is common for Agents state to be initialized to some initial/default value, |
| 82 | + for example, your agent may need to know today's date. |
| 83 | + """ |
| 84 | + |
| 85 | + creation_timestamp: float = 0.0 |
| 86 | + """The time at which this eval case was created.""" |
0 commit comments