From 9dd5463e0687f5514b4765de425fe6b1dc5a4243 Mon Sep 17 00:00:00 2001 From: Xingyao Wang Date: Thu, 2 Jan 2025 16:39:17 -0500 Subject: [PATCH] Set default value of use_microagents to False to prevent breaking eval (#5976) Co-authored-by: openhands --- evaluation/benchmarks/EDA/run_infer.py | 2 ++ evaluation/benchmarks/agent_bench/run_infer.py | 2 ++ evaluation/benchmarks/aider_bench/run_infer.py | 2 ++ evaluation/benchmarks/biocoder/run_infer.py | 2 ++ evaluation/benchmarks/bird/run_infer.py | 2 ++ evaluation/benchmarks/browsing_delegation/run_infer.py | 2 ++ evaluation/benchmarks/discoverybench/run_infer.py | 2 ++ evaluation/benchmarks/gaia/run_infer.py | 2 ++ evaluation/benchmarks/gorilla/run_infer.py | 2 ++ evaluation/benchmarks/gpqa/run_infer.py | 2 ++ evaluation/benchmarks/humanevalfix/run_infer.py | 2 ++ evaluation/benchmarks/logic_reasoning/run_infer.py | 2 ++ evaluation/benchmarks/mint/run_infer.py | 2 ++ evaluation/benchmarks/ml_bench/run_infer.py | 2 ++ evaluation/benchmarks/toolqa/run_infer.py | 2 ++ evaluation/benchmarks/webarena/run_infer.py | 2 ++ 16 files changed, 32 insertions(+) diff --git a/evaluation/benchmarks/EDA/run_infer.py b/evaluation/benchmarks/EDA/run_infer.py index e8cee3df3e20..b5a021a0b853 100644 --- a/evaluation/benchmarks/EDA/run_infer.py +++ b/evaluation/benchmarks/EDA/run_infer.py @@ -75,6 +75,8 @@ def get_config( workspace_mount_path=None, ) config.set_llm_config(metadata.llm_config) + agent_config = config.get_agent_config(metadata.agent_class) + agent_config.use_microagents = False return config diff --git a/evaluation/benchmarks/agent_bench/run_infer.py b/evaluation/benchmarks/agent_bench/run_infer.py index a64c66f22cdc..cf1dc6bba97c 100644 --- a/evaluation/benchmarks/agent_bench/run_infer.py +++ b/evaluation/benchmarks/agent_bench/run_infer.py @@ -59,6 +59,8 @@ def get_config( workspace_mount_path=None, ) config.set_llm_config(metadata.llm_config) + agent_config = config.get_agent_config(metadata.agent_class) + agent_config.use_microagents = False return config diff --git a/evaluation/benchmarks/aider_bench/run_infer.py b/evaluation/benchmarks/aider_bench/run_infer.py index bc850dbc6261..f0e86f30380e 100644 --- a/evaluation/benchmarks/aider_bench/run_infer.py +++ b/evaluation/benchmarks/aider_bench/run_infer.py @@ -67,6 +67,8 @@ def get_config( workspace_mount_path=None, ) config.set_llm_config(metadata.llm_config) + agent_config = config.get_agent_config(metadata.agent_class) + agent_config.use_microagents = False # copy 'draft_editor' config if exists config_copy = copy.deepcopy(config) diff --git a/evaluation/benchmarks/biocoder/run_infer.py b/evaluation/benchmarks/biocoder/run_infer.py index c33c75e5a221..ba8eb4d17b20 100644 --- a/evaluation/benchmarks/biocoder/run_infer.py +++ b/evaluation/benchmarks/biocoder/run_infer.py @@ -73,6 +73,8 @@ def get_config( workspace_mount_path=None, ) config.set_llm_config(metadata.llm_config) + agent_config = config.get_agent_config(metadata.agent_class) + agent_config.use_microagents = False return config diff --git a/evaluation/benchmarks/bird/run_infer.py b/evaluation/benchmarks/bird/run_infer.py index 14946ebacb2f..45ddf582dc64 100644 --- a/evaluation/benchmarks/bird/run_infer.py +++ b/evaluation/benchmarks/bird/run_infer.py @@ -86,6 +86,8 @@ def get_config( workspace_mount_path=None, ) config.set_llm_config(metadata.llm_config) + agent_config = config.get_agent_config(metadata.agent_class) + agent_config.use_microagents = False return config diff --git a/evaluation/benchmarks/browsing_delegation/run_infer.py b/evaluation/benchmarks/browsing_delegation/run_infer.py index 016b6c3f582e..3313c9ff4c3d 100644 --- a/evaluation/benchmarks/browsing_delegation/run_infer.py +++ b/evaluation/benchmarks/browsing_delegation/run_infer.py @@ -50,6 +50,8 @@ def get_config( workspace_mount_path=None, ) config.set_llm_config(metadata.llm_config) + agent_config = config.get_agent_config(metadata.agent_class) + agent_config.use_microagents = False return config diff --git a/evaluation/benchmarks/discoverybench/run_infer.py b/evaluation/benchmarks/discoverybench/run_infer.py index 0d5b47410c2d..05ff44003517 100644 --- a/evaluation/benchmarks/discoverybench/run_infer.py +++ b/evaluation/benchmarks/discoverybench/run_infer.py @@ -77,6 +77,8 @@ def get_config( workspace_mount_path=None, ) config.set_llm_config(metadata.llm_config) + agent_config = config.get_agent_config(metadata.agent_class) + agent_config.use_microagents = False agent_config = AgentConfig( function_calling=False, codeact_enable_jupyter=True, diff --git a/evaluation/benchmarks/gaia/run_infer.py b/evaluation/benchmarks/gaia/run_infer.py index 8aaa479e92be..7974a092903c 100644 --- a/evaluation/benchmarks/gaia/run_infer.py +++ b/evaluation/benchmarks/gaia/run_infer.py @@ -62,6 +62,8 @@ def get_config( workspace_mount_path=None, ) config.set_llm_config(metadata.llm_config) + agent_config = config.get_agent_config(metadata.agent_class) + agent_config.use_microagents = False return config diff --git a/evaluation/benchmarks/gorilla/run_infer.py b/evaluation/benchmarks/gorilla/run_infer.py index e453b1f570ba..740a3c3ada8f 100644 --- a/evaluation/benchmarks/gorilla/run_infer.py +++ b/evaluation/benchmarks/gorilla/run_infer.py @@ -55,6 +55,8 @@ def get_config( workspace_mount_path=None, ) config.set_llm_config(metadata.llm_config) + agent_config = config.get_agent_config(metadata.agent_class) + agent_config.use_microagents = False return config diff --git a/evaluation/benchmarks/gpqa/run_infer.py b/evaluation/benchmarks/gpqa/run_infer.py index 08e66827924e..eb1c808ec8a4 100644 --- a/evaluation/benchmarks/gpqa/run_infer.py +++ b/evaluation/benchmarks/gpqa/run_infer.py @@ -76,6 +76,8 @@ def get_config( workspace_mount_path=None, ) config.set_llm_config(metadata.llm_config) + agent_config = config.get_agent_config(metadata.agent_class) + agent_config.use_microagents = False return config diff --git a/evaluation/benchmarks/humanevalfix/run_infer.py b/evaluation/benchmarks/humanevalfix/run_infer.py index b2fb6d677a9c..5ab5af818f90 100644 --- a/evaluation/benchmarks/humanevalfix/run_infer.py +++ b/evaluation/benchmarks/humanevalfix/run_infer.py @@ -97,6 +97,8 @@ def get_config( workspace_mount_path=None, ) config.set_llm_config(metadata.llm_config) + agent_config = config.get_agent_config(metadata.agent_class) + agent_config.use_microagents = False return config diff --git a/evaluation/benchmarks/logic_reasoning/run_infer.py b/evaluation/benchmarks/logic_reasoning/run_infer.py index d84c5f8ca8cb..ee48f5ea76c8 100644 --- a/evaluation/benchmarks/logic_reasoning/run_infer.py +++ b/evaluation/benchmarks/logic_reasoning/run_infer.py @@ -61,6 +61,8 @@ def get_config( workspace_mount_path=None, ) config.set_llm_config(metadata.llm_config) + agent_config = config.get_agent_config(metadata.agent_class) + agent_config.use_microagents = False return config diff --git a/evaluation/benchmarks/mint/run_infer.py b/evaluation/benchmarks/mint/run_infer.py index a98fa8d91805..61223572ae83 100644 --- a/evaluation/benchmarks/mint/run_infer.py +++ b/evaluation/benchmarks/mint/run_infer.py @@ -119,6 +119,8 @@ def get_config( workspace_mount_path=None, ) config.set_llm_config(metadata.llm_config) + agent_config = config.get_agent_config(metadata.agent_class) + agent_config.use_microagents = False return config diff --git a/evaluation/benchmarks/ml_bench/run_infer.py b/evaluation/benchmarks/ml_bench/run_infer.py index 1c084fc14916..7f15476423c0 100644 --- a/evaluation/benchmarks/ml_bench/run_infer.py +++ b/evaluation/benchmarks/ml_bench/run_infer.py @@ -92,6 +92,8 @@ def get_config( workspace_mount_path=None, ) config.set_llm_config(metadata.llm_config) + agent_config = config.get_agent_config(metadata.agent_class) + agent_config.use_microagents = False return config diff --git a/evaluation/benchmarks/toolqa/run_infer.py b/evaluation/benchmarks/toolqa/run_infer.py index 6f6f1a0e2048..8586f9a7bb7c 100644 --- a/evaluation/benchmarks/toolqa/run_infer.py +++ b/evaluation/benchmarks/toolqa/run_infer.py @@ -56,6 +56,8 @@ def get_config( workspace_mount_path=None, ) config.set_llm_config(metadata.llm_config) + agent_config = config.get_agent_config(metadata.agent_class) + agent_config.use_microagents = False return config diff --git a/evaluation/benchmarks/webarena/run_infer.py b/evaluation/benchmarks/webarena/run_infer.py index ac51a201a712..c35c79ba2cce 100644 --- a/evaluation/benchmarks/webarena/run_infer.py +++ b/evaluation/benchmarks/webarena/run_infer.py @@ -77,6 +77,8 @@ def get_config( workspace_mount_path=None, ) config.set_llm_config(metadata.llm_config) + agent_config = config.get_agent_config(metadata.agent_class) + agent_config.use_microagents = False return config