File size: 9,490 Bytes
f1adc24 | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 | #!/usr/bin/env python3
"""
Evaluation script for pattern-based baseline prover.
Evaluates on extrinsic propositional logic problemset.
"""
import os
import json
import sys
from typing import Optional
from concurrent.futures import ProcessPoolExecutor, as_completed
from functools import partial
import problems
from learning.exp_pattern_prover.pattern_prover import PatternProver, PatternProverConfig
from proofsearch import format_blocks_with_indent
try:
from tqdm import tqdm
except ImportError:
# Fallback if tqdm not available
def tqdm(iterable, **kwargs):
return iterable
def _evaluate_single_problem(args):
"""Worker function to evaluate a single problem (for multiprocessing)."""
problem, problemset_id, max_depth, max_iterations, verbose, log_dir = args
try:
# Load problemset (each worker needs its own copy)
problemset = problems.load_problemset(problemset_id)
# Create log file path if logging is enabled
log_file = None
if log_dir:
os.makedirs(log_dir, exist_ok=True)
log_file = os.path.join(log_dir, f"{problem}_log.jsonl")
# Create pattern prover
config = PatternProverConfig(
max_depth=max_depth,
max_iterations=max_iterations,
verbose=verbose,
log_file=log_file
)
prover = PatternProver(config)
# Initialize and solve
state = problemset.initialize_problem(problem)
result = prover.proof_search(problem, state)
# Extract proof if successful
proof_text = None
if result.success:
try:
# Get solution actions first to verify structure
solution_actions = result.root.get_solution_actions()
if solution_actions is None or (isinstance(solution_actions, list) and len(solution_actions) == 0):
proof_text = "<proof reconstruction failed: No solution actions found>"
else:
# Make a copy since reconstruct_proof modifies the list
actions_copy = [a for a in solution_actions] if isinstance(solution_actions, list) else solution_actions
proof_text = format_blocks_with_indent(result.root.reconstruct_proof())
except Exception as e:
import traceback
proof_text = f"<proof reconstruction failed: {e}>\n{traceback.format_exc()}"
return {
'problem': problem,
'success': result.success,
'iterations': result.iterations,
'proof': proof_text,
'error': None
}
except Exception as e:
import traceback
return {
'problem': problem,
'success': False,
'iterations': 0,
'proof': None,
'error': f"{str(e)}\n{traceback.format_exc()}"
}
def evaluate_pattern_prover(
problemset_id: str = 'extrinsic-propositional-logic',
max_problems: Optional[int] = None,
max_depth: int = 30,
max_iterations: int = 5000,
verbose: bool = False,
output_dir: str = '.',
num_workers: Optional[int] = None,
log_dir: Optional[str] = None,
):
"""
Evaluate pattern prover on a problemset.
Args:
problemset_id: ID of problemset to evaluate on
max_problems: Maximum number of problems to evaluate (None = all)
max_depth: Maximum search depth
max_iterations: Maximum iterations
verbose: Print verbose output
output_dir: Directory to save results
num_workers: Number of parallel workers (None = use all CPU cores)
"""
# Load problemset to get problem list
print(f"Loading problemset: {problemset_id}")
problemset = problems.load_problemset(problemset_id)
print(f"Loaded {len(problemset)} problems")
# Get problems to evaluate
problem_names = problemset.problem_names()
if max_problems is not None:
problem_names = problem_names[:max_problems]
# Determine number of workers
if num_workers is None:
import multiprocessing
num_workers = multiprocessing.cpu_count()
print(f"Evaluating {len(problem_names)} problems using {num_workers} workers...")
print()
# Set up log directory
if log_dir is None:
log_dir = os.path.join(output_dir, "logs")
# Prepare arguments for workers
worker_args = [
(problem, problemset_id, max_depth, max_iterations, verbose, log_dir)
for problem in problem_names
]
# Evaluate in parallel
results = {
"results": [],
"num_proved": 0,
"num_total": len(problem_names),
"solved_problems": []
}
if num_workers == 1:
# Sequential execution (useful for debugging)
for args in worker_args:
result = _evaluate_single_problem(args)
results["results"].append(result)
if result['success']:
results["num_proved"] += 1
results["solved_problems"].append(result['problem'])
if verbose:
status = "✓" if result['success'] else "✗"
print(f"{status} {result['problem']}: {result['iterations']} iterations")
else:
# Parallel execution
with ProcessPoolExecutor(max_workers=num_workers) as executor:
futures = {
executor.submit(_evaluate_single_problem, args): args[0]
for args in worker_args
}
for future in tqdm(as_completed(futures), total=len(futures), desc="Evaluating"):
problem_name = futures[future]
try:
result = future.result()
results["results"].append(result)
if result['success']:
results["num_proved"] += 1
results["solved_problems"].append(result['problem'])
if verbose:
status = "✓" if result['success'] else "✗"
print(f"{status} {result['problem']}: {result['iterations']} iterations")
except Exception as e:
print(f"Error evaluating {problem_name}: {e}")
results["results"].append({
'problem': problem_name,
'success': False,
'iterations': 0,
'proof': None,
'error': str(e)
})
# Sort results by problem name for consistency
results["results"].sort(key=lambda x: x['problem'])
# Print summary
print()
print("=" * 60)
print("SUMMARY")
print("=" * 60)
print(f"Solved: {results['num_proved']}/{results['num_total']}")
if results['num_total'] > 0:
print(f"Pass rate: {results['num_proved']/results['num_total']*100:.1f}%")
print()
if results['solved_problems']:
print("Solved problems:")
for p in sorted(results['solved_problems']):
print(f" - {p}")
print()
# Save results
os.makedirs(output_dir, exist_ok=True)
output_file = os.path.join(output_dir, "pattern_prover_results.json")
with open(output_file, 'w') as f:
json.dump(results, f, indent=2)
print(f"Results saved to: {output_file}")
return results
def main():
"""Main entry point."""
import argparse
parser = argparse.ArgumentParser(description="Evaluate pattern-based prover")
parser.add_argument(
'--problemset',
type=str,
default='extrinsic-propositional-logic',
help='Problemset ID (default: extrinsic-propositional-logic)'
)
parser.add_argument(
'--max-problems',
type=int,
default=None,
help='Maximum number of problems to evaluate (default: all)'
)
parser.add_argument(
'--max-depth',
type=int,
default=30,
help='Maximum search depth (default: 30)'
)
parser.add_argument(
'--max-iterations',
type=int,
default=5000,
help='Maximum iterations (default: 5000)'
)
parser.add_argument(
'--verbose',
action='store_true',
help='Print verbose output'
)
parser.add_argument(
'--output-dir',
type=str,
default='.',
help='Output directory for results (default: current directory)'
)
parser.add_argument(
'--num-workers',
type=int,
default=None,
help='Number of parallel workers (default: use all CPU cores)'
)
parser.add_argument(
'--log-dir',
type=str,
default=None,
help='Directory for detailed logs (default: output_dir/logs)'
)
args = parser.parse_args()
evaluate_pattern_prover(
problemset_id=args.problemset,
max_problems=args.max_problems,
max_depth=args.max_depth,
max_iterations=args.max_iterations,
verbose=args.verbose,
output_dir=args.output_dir,
num_workers=args.num_workers,
log_dir=args.log_dir
)
if __name__ == '__main__':
main()
#command to run:
# python eval_pattern_prover.py --problemset extrinsic-95--output-dir results/pattern_prover --num-workers 4 --log-dir results/pattern_prover_95 |