🌟 The Multi-Agent Framework: First AI Software Company, Towards Natural Language Programming
1
kitd5bd342b90a7
diff --git a/examples/di/InfiAgent-DABench/README.md b/examples/di/InfiAgent-DABench/README.mdindex 814a1937b..74783c9d1 100644--- a/examples/di/InfiAgent-DABench/README.md+++ b/examples/di/InfiAgent-DABench/README.md@@ -1,12 +1,42 @@# InfiAgent-DABenchThis example is used to solve the InfiAgent-DABench using Data Interpreter (DI), and obtains 94.93% accuracy using gpt-4o.-## Dataset+## Dataset download```cd /examples/di/InfiAgent-DABenchgit clone https://github.com/InfiAgent/InfiAgent.gitmv InfiAgent/examples/DA-Agent/data ./```+## Special note:+When doing DABench testing, you need to set the ExecuteNbCode() init to:+```+class ExecuteNbCode(Action):+ """execute notebook code block, return result to llm, and display it."""++ nb: NotebookNode+ nb_client: NotebookClient+ console: Console+ interaction: str+ timeout: int = 600++ def __init__(+ self,+ nb=nbformat.v4.new_notebook(),+ timeout=600,+ ):+ super().__init__(+ nb=nbformat.v4.new_notebook(),#nb,+ nb_client=NotebookClient(nb, timeout=timeout),+ timeout=timeout,+ console=Console(),+ interaction=("ipython" if self.is_ipython() else "terminal"),+ )+```+The path of ExecuteNbCode() is:+```+metagpt.actions.di.execute_nb_code+```+Instead of using the original nb initialization by default.## How to run```python run_InfiAgent-DABench_single.py --id x # run a task, x represents the id of the question you want to test