changes to make it work with Azure hosted openAI models

Author: uniperJim

config.yaml

@@ -0,0 +1,112 @@
+# OpenEvolve Default Configuration
+# This file contains all available configuration options with sensible defaults
+# You can use this as a template for your own configuration
+
+# General settings
+max_iterations: 1000                  # Maximum number of evolution iterations
+checkpoint_interval: 50               # Save checkpoints every N iterations
+log_level: "INFO"                     # Logging level (DEBUG, INFO, WARNING, ERROR, CRITICAL)
+log_dir: null                         # Custom directory for logs (default: output_dir/logs)
+random_seed: null                     # Random seed for reproducibility (null = random)
+
+# Evolution settings
+diff_based_evolution: true            # Use diff-based evolution (true) or full rewrites (false)
+allow_full_rewrites: false            # Allow occasional full rewrites even in diff-based mode
+max_code_length: 10000                # Maximum allowed code length in characters
+
+# LLM configuration
+llm:
+  models:
+    - name: "o1"
+      weight: 1.0
+
+  evaluator_models:
+    - name: "o1"
+      weight: 1.0
+
+  # Azure endpoint *root* – no path, no query string
+  api_base: "https://<YOUR_BASE>.openai.azure.com/openai/deployments/<YOUR_DEPLOYMENT_eg_o1"
+
+  # Tell the SDK which API flavour and version to use
+  # api_type: "azure"
+  # api_version: "2025-01-01-preview"
+  api_key: YOUR_API_KEY  # Or provide it directly here
+  temperature: 0.7
+  top_p: 0.95
+  max_tokens: 4096
+  timeout: 60
+  retries: 3
+  retry_delay: 5
+
+
+# Prompt configuration
+prompt:
+  template_dir: null                  # Custom directory for prompt templates
+  system_message: "You are an expert coder helping to improve programs through evolution."
+  evaluator_system_message: "You are an expert code reviewer."
+
+  # Number of examples to include in the prompt
+  num_top_programs: 3                 # Number of top-performing programs to include
+  num_diverse_programs: 2             # Number of diverse programs to include
+
+  # Template stochasticity
+  use_template_stochasticity: true    # Use random variations in templates for diversity
+  template_variations:                # Different phrasings for parts of the template
+    improvement_suggestion:
+      - "Here's how we could improve this code:"
+      - "I suggest the following improvements:"
+      - "We can enhance this code by:"
+
+  # Note: meta-prompting features are not yet implemented
+
+# Database configuration
+database:
+  # General settings
+  db_path: null                       # Path to persist database (null = in-memory only)
+  in_memory: true                     # Keep database in memory for faster access
+
+  # Evolutionary parameters
+  population_size: 1000               # Maximum number of programs to keep in memory
+  archive_size: 100                   # Size of elite archive
+  num_islands: 5                      # Number of islands for island model (separate populations)
+
+  # Island-based evolution parameters
+  # Islands provide diversity by maintaining separate populations that evolve independently.
+  # Migration periodically shares the best solutions between adjacent islands.
+  migration_interval: 50              # Migrate between islands every N generations
+  migration_rate: 0.1                 # Fraction of top programs to migrate (0.1 = 10%)
+
+  # Selection parameters
+  elite_selection_ratio: 0.1          # Ratio of elite programs to select
+  exploration_ratio: 0.2              # Ratio of exploration vs exploitation
+  exploitation_ratio: 0.7             # Ratio of exploitation vs random selection
+  # Note: diversity_metric is fixed to "edit_distance" (feature_based not implemented)
+
+  # Feature map dimensions for MAP-Elites
+  feature_dimensions:                 # Dimensions for MAP-Elites feature map
+    - "score"                         # Performance score
+    - "complexity"                    # Code complexity (length)
+  feature_bins: 10                    # Number of bins per dimension
+
+# Evaluator configuration
+evaluator:
+  # General settings
+  timeout: 300                        # Maximum evaluation time in seconds
+  max_retries: 3                      # Maximum number of retries for evaluation
+
+  # Note: resource limits (memory_limit_mb, cpu_limit) are not yet implemented
+
+  # Evaluation strategies
+  cascade_evaluation: true            # Use cascade evaluation to filter bad solutions early
+  cascade_thresholds:                 # Thresholds for advancing to next evaluation stage
+    - 0.5                             # First stage threshold
+    - 0.75                            # Second stage threshold
+    - 0.9                             # Third stage threshold
+
+  # Parallel evaluation
+  parallel_evaluations: 4             # Number of parallel evaluations
+  # Note: distributed evaluation is not yet implemented
+
+  # LLM-based feedback (experimental)
+  use_llm_feedback: false             # Use LLM to evaluate code quality
+  llm_feedback_weight: 0.1            # Weight for LLM feedback in final score

openevolve/llm/openai.py

@@ -12,6 +12,8 @@
 from openevolve.config import LLMConfig
 from openevolve.llm.base import LLMInterface
 
+from openai import AzureOpenAI
+
 logger = logging.getLogger(__name__)
 
 
@@ -34,9 +36,13 @@ def __init__(
         self.api_key = model_cfg.api_key
 
         # Set up API client
+        print("##########################")
+    
         self.client = openai.OpenAI(
-            api_key=self.api_key,
-            base_url=self.api_base,
+            api_key      = self.api_key,
+            base_url     = self.api_base,
+            default_headers={"api-key": self.api_key},
+            default_query = {"api-version": "2025-01-01-preview"},
         )
 
         logger.info(f"Initialized OpenAI LLM with model: {self.model}")
@@ -102,6 +108,16 @@ async def generate_with_context(
 
     async def _call_api(self, params: Dict[str, Any]) -> str:
         """Make the actual API call"""
+        # ----- Azure o-series models need max_completion_tokens -----
+        if "max_tokens" in params:
+            params = params.copy()                               # don’t mutate caller’s dict
+            params["extra_body"] = {"max_completion_tokens": params.pop("max_tokens")}
+        # -----------------------------------------------------------
+        # ⬇ NEW: drop sampling knobs that o-series refuses
+        for unsupported in ("temperature", "top_p",
+                            "frequency_penalty", "presence_penalty"):
+            params.pop(unsupported, None)
+        # ------------------------------------------------------------------
         # Use asyncio to run the blocking API call in a thread pool
         loop = asyncio.get_event_loop()
         response = await loop.run_in_executor(