format SPO code and add figure from paper

examples/spo/README.md

@@ -4,6 +4,10 @@ # SPO 🤖 | Self-Supervised Prompt PromptOptimizer
 
 A next-generation prompt engineering system implementing **Self-Supervised Prompt Optimization (SPO)**. Achieves state-of-the-art performance with 17.8-90.9× higher cost efficiency than conventional methods. 🚀
 
+<p align="center">
+<a href=""><img src="../../docs/resources/spo/SPO-method.png" alt="Framework of AFlow" title="Framework of AFlow <sub>1</sub>" width="80%"></a>
+</p>
+
 ## ✨ Core Advantages
 
 - 💸 **Ultra-Low Cost** - _$0.15 per task optimization_
@@ -11,6 +15,25 @@ ## ✨ Core Advantages
 - ⚡ **Universal Adaptation** - _Closed & open-ended tasks supported_
 - 🔄 **Self-Evolving** - _Auto-optimization via LLM-as-judge mechanism_
 
+[Read our paper on arXiv](coming soon)
+
+## 📊 Experiment
+
+###  Closed Tasks
+<p align="center">
+<a href=""><img src="../../docs/resources/spo/SPO-closed_task_table.png" alt="Framework of AFlow" title="Framework of AFlow <sub>1</sub>" width="80%"></a>
+<a href=""><img src="../../docs/resources/spo/SPO-closed_task_figure.png" alt="Framework of AFlow" title="Framework of AFlow <sub>1</sub>" width="80%"></a>
+</p>
+
+*SPO demonstrates superior cost efficiency, requiring only 1.1% to 5.6% of the cost of state-of-the-art methods while maintaining competitive performance.*
+
+### Open-ended Tasks
+<p align="center">
+<a href=""><img src="../../docs/resources/spo/SPO-open-ended _task_figure.png" alt="Framework of AFlow" title="Framework of AFlow <sub>1</sub>" width="80%"></a>
+</p>
+
+*SPO significantly improves model performance across all model configurations in open-ended tasks.*
+
 ## 🚀 Quick Start
 
 ### 1. Configure Your API Key ⚙️

examples/spo/optimize.py

@@ -1,38 +1,27 @@
 import argparse
+
 from metagpt.ext.spo.components.optimizer import PromptOptimizer
 from metagpt.ext.spo.utils.llm_client import SPO_LLM
 
 
 def parse_args():
-    parser = argparse.ArgumentParser(description='SPO PromptOptimizer CLI')
+    parser = argparse.ArgumentParser(description="SPO PromptOptimizer CLI")
 
     # LLM parameter
-    parser.add_argument('--opt-model', type=str, default='claude-3-5-sonnet-20240620',
-                       help='Model for optimization')
-    parser.add_argument('--opt-temp', type=float, default=0.7,
-                       help='Temperature for optimization')
-    parser.add_argument('--eval-model', type=str, default='gpt-4o-mini',
-                       help='Model for evaluation')
-    parser.add_argument('--eval-temp', type=float, default=0.3,
-                       help='Temperature for evaluation')
-    parser.add_argument('--exec-model', type=str, default='gpt-4o-mini',
-                       help='Model for execution')
-    parser.add_argument('--exec-temp', type=float, default=0,
-                       help='Temperature for execution')
+    parser.add_argument("--opt-model", type=str, default="claude-3-5-sonnet-20240620", help="Model for optimization")
+    parser.add_argument("--opt-temp", type=float, default=0.7, help="Temperature for optimization")
+    parser.add_argument("--eval-model", type=str, default="gpt-4o-mini", help="Model for evaluation")
+    parser.add_argument("--eval-temp", type=float, default=0.3, help="Temperature for evaluation")
+    parser.add_argument("--exec-model", type=str, default="gpt-4o-mini", help="Model for execution")
+    parser.add_argument("--exec-temp", type=float, default=0, help="Temperature for execution")
 
     # PromptOptimizer parameter
-    parser.add_argument('--workspace', type=str, default='workspace',
-                       help='Path for optimized output')
-    parser.add_argument('--initial-round', type=int, default=1,
-                       help='Initial round number')
-    parser.add_argument('--max-rounds', type=int, default=10,
-                       help='Maximum number of rounds')
-    parser.add_argument('--template', type=str, default='Poem.yaml',
-                       help='Template file name')
-    parser.add_argument('--name', type=str, default='Poem',
-                       help='Project name')
-    parser.add_argument('--no-iteration', action='store_false', dest='iteration',
-                       help='Disable iteration mode')
+    parser.add_argument("--workspace", type=str, default="workspace", help="Path for optimized output")
+    parser.add_argument("--initial-round", type=int, default=1, help="Initial round number")
+    parser.add_argument("--max-rounds", type=int, default=10, help="Maximum number of rounds")
+    parser.add_argument("--template", type=str, default="Poem.yaml", help="Template file name")
+    parser.add_argument("--name", type=str, default="Poem", help="Project name")
+    parser.add_argument("--no-iteration", action="store_false", dest="iteration", help="Disable iteration mode")
 
     return parser.parse_args()
 
@@ -41,18 +30,9 @@ def main():
     args = parse_args()
 
     SPO_LLM.initialize(
-        optimize_kwargs={
-            "model": args.opt_model,
-            "temperature": args.opt_temp
-        },
-        evaluate_kwargs={
-            "model": args.eval_model,
-            "temperature": args.eval_temp
-        },
-        execute_kwargs={
-            "model": args.exec_model,
-            "temperature": args.exec_temp
-        }
+        optimize_kwargs={"model": args.opt_model, "temperature": args.opt_temp},
+        evaluate_kwargs={"model": args.eval_model, "temperature": args.eval_temp},
+        execute_kwargs={"model": args.exec_model, "temperature": args.exec_temp},
     )
 
     optimizer = PromptOptimizer(