You've already forked adk-python
mirror of
https://github.com/encounter/adk-python.git
synced 2026-03-30 10:57:20 -07:00
Xiang (Sean) Zhou
14f118899d
see https://github.com/google/adk-python/issues/2764 PiperOrigin-RevId: 807972596
106 lines
3.6 KiB
Python
106 lines
3.6 KiB
Python
# Copyright 2025 Google LLC
|
|
#
|
|
# Licensed under the Apache License, Version 2.0 (the "License");
|
|
# you may not use this file except in compliance with the License.
|
|
# You may obtain a copy of the License at
|
|
#
|
|
# http://www.apache.org/licenses/LICENSE-2.0
|
|
#
|
|
# Unless required by applicable law or agreed to in writing, software
|
|
# distributed under the License is distributed on an "AS IS" BASIS,
|
|
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
|
# See the License for the specific language governing permissions and
|
|
# limitations under the License.
|
|
|
|
"""Sample agent demonstrating log probability usage.
|
|
|
|
This agent shows how to access log probabilities from language model responses.
|
|
The after_model_callback appends confidence information to demonstrate how
|
|
logprobs can be extracted and used.
|
|
"""
|
|
|
|
from google.adk.agents.callback_context import CallbackContext
|
|
from google.adk.agents.llm_agent import Agent
|
|
from google.adk.models.llm_response import LlmResponse
|
|
from google.genai import types
|
|
|
|
|
|
async def append_logprobs_to_response(
|
|
callback_context: CallbackContext, llm_response: LlmResponse
|
|
) -> LlmResponse:
|
|
"""After-model callback that appends log probability information to response.
|
|
|
|
This callback demonstrates how to access avg_logprobs and logprobs_result
|
|
from the LlmResponse and append the information to the response content.
|
|
|
|
Args:
|
|
callback_context: The current callback context
|
|
llm_response: The LlmResponse containing logprobs data
|
|
|
|
Returns:
|
|
Modified LlmResponse with logprobs information appended
|
|
"""
|
|
# Build log probability analysis
|
|
if llm_response.avg_logprobs is None:
|
|
print("⚠️ No log probability data available")
|
|
logprobs_info = (
|
|
"\n\n[LOG PROBABILITY ANALYSIS]\n⚠️ No log probability data available"
|
|
)
|
|
else:
|
|
print(f"📊 Average log probability: {llm_response.avg_logprobs:.4f}")
|
|
|
|
# Build confidence analysis
|
|
confidence_level = (
|
|
"High"
|
|
if llm_response.avg_logprobs >= -0.5
|
|
else "Medium"
|
|
if llm_response.avg_logprobs >= -1.0
|
|
else "Low"
|
|
)
|
|
|
|
logprobs_info = f"""
|
|
|
|
[LOG PROBABILITY ANALYSIS]
|
|
📊 Average Log Probability: {llm_response.avg_logprobs:.4f}
|
|
🎯 Confidence Level: {confidence_level}
|
|
📈 Confidence Score: {100 * (2 ** llm_response.avg_logprobs):.1f}%"""
|
|
|
|
# Optionally include detailed logprobs_result information
|
|
if (
|
|
llm_response.logprobs_result
|
|
and llm_response.logprobs_result.top_candidates
|
|
):
|
|
logprobs_info += (
|
|
"\n🔍 Top alternatives analyzed:"
|
|
f" {len(llm_response.logprobs_result.top_candidates)}"
|
|
)
|
|
|
|
# Append logprobs analysis to the response
|
|
if llm_response.content and llm_response.content.parts:
|
|
llm_response.content.parts.append(types.Part(text=logprobs_info))
|
|
|
|
return llm_response
|
|
|
|
|
|
# Create a simple agent that demonstrates logprobs usage
|
|
root_agent = Agent(
|
|
model="gemini-2.0-flash",
|
|
name="logprobs_demo_agent",
|
|
description=(
|
|
"A simple agent that demonstrates log probability extraction and"
|
|
" display."
|
|
),
|
|
instruction="""
|
|
You are a helpful AI assistant. Answer user questions normally and naturally.
|
|
|
|
After you respond, you'll see log probability analysis appended to your response.
|
|
You don't need to include the log probability analysis in your response yourself.
|
|
""",
|
|
generate_content_config=types.GenerateContentConfig(
|
|
response_logprobs=True, # Enable log probability collection
|
|
logprobs=5, # Collect top 5 alternatives for analysis
|
|
temperature=0.7, # Moderate temperature for varied responses
|
|
),
|
|
after_model_callback=append_logprobs_to_response,
|
|
)
|