Merge pull request algorithmicsuperintelligence#163 from tonyxty/main

Add configurable program length check

Author: codelion

configs/default_config.yaml

@@ -68,6 +68,13 @@ prompt:
   max_artifact_bytes: 20480           # Maximum artifact size in bytes (20KB default)
   artifact_security_filter: true      # Apply security filtering to artifacts
 
+  # Feature extraction and program labeling thresholds
+  # These control how the LLM perceives and categorizes programs
+  suggest_simplification_after_chars: 500     # Suggest simplifying if program exceeds this many characters
+  include_changes_under_chars: 100           # Include change descriptions in features if under this length  
+  concise_implementation_max_lines: 10        # Label as "concise" if program has this many lines or fewer
+  comprehensive_implementation_min_lines: 50  # Label as "comprehensive" if program has this many lines or more
+
   # Note: meta-prompting features are not yet implemented
 
 # Database configuration

openevolve/config.py

@@ -142,6 +142,15 @@ class PromptConfig:
     max_artifact_bytes: int = 20 * 1024  # 20KB in prompt
     artifact_security_filter: bool = True
 
+    # Feature extraction and program labeling
+    suggest_simplification_after_chars: Optional[int] = 500  # Suggest simplifying if program exceeds this many characters
+    include_changes_under_chars: Optional[int] = 100  # Include change descriptions in features if under this length
+    concise_implementation_max_lines: Optional[int] = 10  # Label as "concise" if program has this many lines or fewer
+    comprehensive_implementation_min_lines: Optional[int] = 50  # Label as "comprehensive" if program has this many lines or more
+    
+    # Backward compatibility - deprecated
+    code_length_threshold: Optional[int] = None  # Deprecated: use suggest_simplification_after_chars
+
 
 @dataclass
 class DatabaseConfig:

openevolve/prompt/sampler.py

@@ -171,7 +171,9 @@ def _identify_improvement_areas(
         improvement_areas = []
 
         # Check program length
-        if len(current_program) > 500:
+        # Support both old and new parameter names for backward compatibility
+        threshold = self.config.suggest_simplification_after_chars or self.config.code_length_threshold
+        if threshold and len(current_program) > threshold:
             improvement_areas.append(
                 "Consider simplifying the code to improve readability and maintainability"
             )
@@ -499,7 +501,7 @@ def _extract_unique_features(self, program: Dict[str, Any]) -> str:
         metadata = program.get("metadata", {})
         if "changes" in metadata:
             changes = metadata["changes"]
-            if isinstance(changes, str) and len(changes) < 100:
+            if isinstance(changes, str) and self.config.include_changes_under_chars and len(changes) < self.config.include_changes_under_chars:
                 features.append(f"Modification: {changes}")
 
         # Analyze metrics for standout characteristics
@@ -521,9 +523,9 @@ def _extract_unique_features(self, program: Dict[str, Any]) -> str:
                 features.append("NumPy-based implementation")
             if "for" in code_lower and "while" in code_lower:
                 features.append("Mixed iteration strategies")
-            if len(code.split("\n")) < 10:
+            if self.config.concise_implementation_max_lines and len(code.split("\n")) <= self.config.concise_implementation_max_lines:
                 features.append("Concise implementation")
-            elif len(code.split("\n")) > 50:
+            elif self.config.comprehensive_implementation_min_lines and len(code.split("\n")) >= self.config.comprehensive_implementation_min_lines:
                 features.append("Comprehensive implementation")
 
         # Default if no specific features found