Validate migration didn't create inconsistencies Checks that: 1. Program island metadata matches actual island assignment

Apply diff blocks line-wise and return (new_text, applied_count)

Calculate custom features for MAP-Elites IMPORTANT: Returns raw continuous values, not bin indices. The database handles all scaling and

Extract diff blocks from the diff text Args: diff_text: Diff in the SEARCH/REPLACE format diff_pattern: Regex pattern for th

Generate synthetic regression data with outliers.

Get the description of a task.

Generates the vertices of a regular hexagon. Args: center_x: x-coordinate of the center. center_y: y-coordinate of the center. side_le

Generates the vertices of a regular hexagon. Args: center_x: x-coordinate of the center. center_y: y-coordinate of the center. side_le

Unified data loading interface. Loads and processes data from Hugging Face Hub. Each task's dataset is grouped by a 'group' key. The functio

Run the complete benchmark suite

Run a function with a timeout using concurrent.futures Args: func: Function to run args: Arguments to pass to the function

Run a function with a timeout using concurrent.futures Args: func: Function to run args: Arguments to pass to the function

Run a function with a timeout using concurrent.futures Args: func: Function to run args: Arguments to pass to the function

Run a function with a timeout using concurrent.futures Args: func: Function to run args: Arguments to pass to the function

Run a function with a timeout using concurrent.futures Args: func: Function to run args: Arguments to pass to the function

Run a function with a timeout using concurrent.futures Args: func: Function to run args: Arguments to pass to the function

Run a function with a timeout using concurrent.futures Args: func: Function to run args: Arguments to pass to the function

Run a function with a timeout using concurrent.futures Args: func: Function to run args: Arguments to pass to the function

Check if migration should occur based on generation counters

Start the process pool

Update model parameters for all models

Applies a scaled tanh to map latent parameters to the desired range.

Get safer benchmark configurations for GPU protection

Resolve run root from a path that may point to: - <run_root> - <run_root>/checkpoints - <run_root>/checkpoints/checkpoint_123

Convert BenchmarkResult to dictionary

Sample a parent for exploration (from current island)

Sample a parent from a specific island using fitness-weighted selection Args: island_id: The island to sample from

Serialize feature_stats for JSON storage Returns: Dictionary that can be JSON-serialized

Kill process (and its group on POSIX) and await .wait()

Apply a diff to the original code Args: original_code: Original source code diff_text: Diff in the SEARCH/REPLACE format

Calculate improvement deltas between parent and child metrics

Evaluate a pipeline configuration program. Args: program_path: Path to the Python file containing configure_pipeline() Returns:

Main evaluation function called by OpenEvolve

Evolve a Python function based on test cases Args: func: Initial function to evolve test_cases: List of (input, expected_out

Export traces to JSON format with metadata Args: traces: List of trace objects with to_dict() method output_path: Path to sa

Extract parameter information from a function signature. Args: signature: Function signature string Returns: List of pa

Get size of a specific artifact in bytes

Get current tracing statistics

Run evolution with process-based parallelism

Solve the psd_cone_projection problem. Args: problem: Dictionary containing problem data specific to psd_cone_pr

Stop the process pool

Apply security filtering to artifact text Args: text: Input text Returns: Filtered text with potent

Calculate overall safety score based on error statistics

Make the actual API call

Run cascade evaluation with increasingly challenging test cases Args: program_path: Path to the program file Re

Remove stale island best program references Cleans up references to programs that no longer exist in the database or are not

Create rich error context for cascade failures Args: stage: The stage where the error occurred error: The ex

Create a serializable snapshot of the database state

Create long context prompt

Create very long context prompt to test KV cache scaling

Extract class information from the task file with improved method extraction.

Extract unique features of an inspiration program Args: program: Program dictionary Returns: String

Get comprehensive error statistics

Load all templates and fragments from a directory

Calculates the C6 lower bound using jnp.unique

Check if metrics pass a threshold Uses 'combined_score' if available (for consistency with evolution), otherwise falls back

Convert BenchmarkResult to dictionary

Sample inspiration programs for the next evolution step. For proper island-based evolution, inspirations are sampled ONLY from the

Save a program to disk Args: program: Program to save base_path: Base path to save to (uses config.db_path i

Update the reference set for diversity calculation

Analyze MLIR IR for performance-relevant characteristics

Benchmark the optimized implementation

Calculate the average of non-None times If more than 5% of the elements are None, return None

Check what MLIR tools are available

Return True if: - cities is ndarray of shape (n, 2), n >= 3, all finite - solution is 1D ndarray of length n - solution contain

Returns the squared norm of an integer vector using exact computation.

Create OpenEvolve files for a specific task. Args: task_name: Name of the AlgoTune task output_dir:

Core evaluation logic: fits a model or evaluates it on test data.

The complex function we're trying to minimize

Evaluate a prompt on a subset of the dataset.

Evolve arbitrary code with a custom evaluator Args: initial_code: Initial code to evolve evaluator: Function that takes a pr

Export traces to specified format Args: traces: List of trace objects output_path: Path to save the file format: Out