encounter/adk-python
0
0
Fork 0
mirror of https://github.com/encounter/adk-python.git synced 2026-07-09 18:19:28 -07:00
Code Issues Packages Projects Releases Wiki Activity

ADK changes

Browse Source 
PiperOrigin-RevId: 829136628
This commit is contained in:
Google Team Member
2025-11-06 15:55:26 -08:00
committed by Copybara-Service
parent f3d6fcf444
commit f1f44675e4
16 changed files with 6481 additions and 50 deletions
Download Patch File Download Diff File Expand all files Collapse all files
contributing/samples/gepa/README.md
+14 -1
View File
@@ -12,7 +12,9 @@ The goal of this demo is to take an agent with a simple, underperforming prompt
and automatically improve it using GEPA, increasing the agent's reliability on a
customer support task.
## Tau-Bench Retail Environment
## Examples
### Tau-Bench Retail Environment
We use the `'retail'` environment from
[Tau-bench](https://github.com/sierra-research/tau-bench), a benchmark designed
@@ -26,6 +28,17 @@ tool-calling strategy. It receives the conversation history and a list of
available tools, and it must decide whether to respond to the user or call a
tool.
The easiest way to run this demo is through the provided Colab notebook:
[`gepa_tau_bench.ipynb`](https://colab.research.google.com/github/google/adk-python/blob/main/contributing/samples/gepa/gepa_tau_bench.ipynb).
### Improving a voter Agent's PII filtering ability
This demo notebook ([`voter_agent/gepa.ipynb`](https://colab.research.google.com/github/google/adk-python/blob/main/contributing/samples/gepa/voter_agent/gepa.ipynb)) walks you through optimizing an AI
agent's prompt using the Genetic-Pareto (GEPA) algorithm. We'll use the Google
Agent Development Kit (ADK) to build and evaluate a "Vote Taker" agent designed
to collect audience votes while filtering sensitive information.
## GEPA Overview
**GEPA (Genetic-Pareto)** is a prompt optimization algorithm that learns from