fix(os_symphony_evaluation) (#410)

* fix(os_symphony)

* Update desktop_env_os_symphony.py

* fix(os_symphony_desktop)

* fix(os_symphony_start)

* Add docstring to run_multienv_os_symphony.py

Added documentation header for the evaluation script.

desktop_env/desktop_env_os_symphony.py

@@ -333,6 +333,8 @@ class DesktopEnv(gym.Env):
 
     def _set_evaluator_info(self, task_config: Dict[str, Any]):
         """Set evaluator information from task config"""
+        if "evaluator" not in task_config:
+            return
         # evaluator dict
         # func -> metric function string, or list of metric function strings
         # conj -> conjunction of multiple metrics if func is a list with length > 1, "and"/"or"

run_multienv_os_symphony.py

@@ -1,3 +1,16 @@
+"""
+OS-Symphony Official Evaluation Script
+
+This script serves as the official evaluation entry point for OS-Symphony.
+It handles the setup of the desktop environment, agent initialization, and 
+execution of evaluation tasks.
+
+For detailed evaluation metrics, configuration options, and usage instructions, 
+please refer to the official repository:
+https://github.com/OS-Copilot/OS-Symphony
+"""
+
+
 import argparse
 import copy
 import datetime

run_os_symphony.sh

@@ -1,57 +1,58 @@
-
+EXP_NAME="xxx"
-EXP_NAME="os-osworld-origin-nogdrive-gpt5-gta1-32b-step50-20251220-ybw"
+export AWS_SECRET_ACCESS_KEY="xxx"
-# enable_rewrite_instruction
+export AWS_ACCESS_KEY_ID="xxx"
+export AWS_REGION="us-east-1"
+export AWS_SUBNET_ID="xxx"
+export AWS_SECURITY_GROUP_ID="xxx"
+# >> logs/${EXP_NAME}.log 2>&1
 python run_multienv_os_symphony.py \
-  --provider_name "docker" \
+  --provider_name "aws" \
-  --path_to_vm "xxx" \
+  --region "us-east-1" \
+  --client_password "osworld-public-evaluation" \
   --headless \
-  --num_envs 1 \
+  --num_envs 7 \
   --max_steps 50 \
   --benchmark osworld \
   --domain "all" \
   --test_all_meta_path evaluation_examples/test_nogdrive.json \
   --result_dir "results" \
-  --region "us-east-1" \
   --tool_config mm_agents/os_symphony/tool/all_tool_config.yaml \
   --orchestrator_provider "openai" \
   --orchestrator_model "gpt-5" \
-  --orchestrator_url "https://api.boyuerichdata.opensphereai.com/v1" \
+  --orchestrator_url "xxx" \
   --orchestrator_api_key "xxx" \
   --orchestrator_temperature 0.1 \
   --orchestrator_keep_first_image \
   --max_trajectory_length 8 \
   --grounder_provider "vllm" \
-  --grounder_model "gta1_32b" \
+  --grounder_model "UI-TARS-1.5-7B" \
   --grounder_api_key "none" \
-  --grounder_url "https://h.pjlab.org.cn/kapi/workspace.kubebrain.io/ailab-intern11/dingzichen-7jzkt-932268-worker-0.dingzichen/18080/v1/" \
+  --grounder_url "xxx" \
   --grounding_smart_resize \
-  --grounding_width 1280 \
+  --grounding_width 1920 \
-  --grounding_height 800 \
+  --grounding_height 1080 \
   --coder_provider "openai" \
   --coder_model "gpt-5" \
-  --coder_url "https://api.boyuerichdata.opensphereai.com/v1" \
+  --coder_url "xxx" \
   --coder_api_key "xxx" \
   --coder_temperature 0.1 \
   --coder_budget 20 \
   --memoryer_provider "openai" \
   --memoryer_model "gpt-5" \
-  --memoryer_url "https://api.boyuerichdata.opensphereai.com/v1" \
+  --memoryer_url "xxx" \
   --memoryer_api_key "xxx" \
   --memoryer_temperature 0.1 \
   --memoryer_max_images 8 \
   --searcher_provider "openai" \
   --searcher_model "gpt-5" \
-  --searcher_url "https://api.boyuerichdata.opensphereai.com/v1" \
+  --searcher_url "xxx" \
   --searcher_api_key "xxx" \
   --searcher_temperature 0.1 \
   --searcher_type "vlm" \
-  --searcher_engine "duckduckgo" \
+  --searcher_engine "google" \
-  --searcher_budget 20\
+  --searcher_budget 20 \
   --searcher_screen_width 1920 \
   --searcher_screen_height 1080 \
-  --searcher_path_to_vm "xxx" \
   --sleep_after_execution 3 \
   --exp_name ${EXP_NAME} \
-  --enable_reflection
+  --enable_reflection >> logs/${EXP_NAME}.log 2>&1
-
-# bash scripts/remove_all_osworld_container.sh > logs/${EXP_NAME}.log 2>&1   --enable_rewrite_instruction   --grounding_smart_resize