File size: 6,182 Bytes
b8ab4a2 |
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 |
"""
Validation module for the Math Expert model
"""
import os
import json
from pathlib import Path
import hashlib
import datetime
from typing import Dict, Any, List, Optional
import numpy as np
from sympy import simplify, Eq
class MathValidator:
def __init__(self, checkpoint_dir: str = "checkpoints"):
self.checkpoint_dir = Path(checkpoint_dir)
self.checkpoint_dir.mkdir(exist_ok=True)
self.validation_dir = self.checkpoint_dir / "validation"
self.validation_dir.mkdir(exist_ok=True)
# Initialize validation metrics
self.metrics = {
"accuracy": [],
"equation_simplification": [],
"proof_validation": [],
"memory_usage": []
}
def validate_equation(self, equation: str, expected_result: str) -> Dict[str, Any]:
"""Validate mathematical equation correctness"""
try:
# Try to simplify both sides
lhs = simplify(equation)
rhs = simplify(expected_result)
# Check if simplified forms are equal
is_correct = lhs == rhs
return {
"is_correct": is_correct,
"simplified_lhs": str(lhs),
"simplified_rhs": str(rhs),
"validation_score": float(is_correct)
}
except Exception as e:
return {
"is_correct": False,
"error": str(e),
"validation_score": 0.0
}
def validate_proof(self, proof_steps: List[str], expected_theorem: str) -> Dict[str, Any]:
"""Validate mathematical proof steps"""
try:
# Check if each step logically follows from previous steps
current_context = set()
validation_score = 1.0
for step in proof_steps:
# Try to parse the step as an equation
try:
lhs, rhs = step.split('=')
if not Eq(simplify(lhs), simplify(rhs)):
validation_score *= 0.9 # Penalize incorrect steps
except:
pass # Not all steps are equations
# Update context
current_context.add(step)
# Check if final step matches expected theorem
final_step = proof_steps[-1]
matches_theorem = expected_theorem in final_step
return {
"is_valid": validation_score > 0.5,
"validation_score": validation_score,
"matches_theorem": matches_theorem,
"context_size": len(current_context)
}
except Exception as e:
return {
"is_valid": False,
"error": str(e),
"validation_score": 0.0
}
def create_checkpoint(self, data: Dict[str, Any], name: str = None) -> str:
"""Create a checkpoint of validation data"""
if name is None:
name = datetime.datetime.now().strftime("%Y%m%d_%H%M%S")
checkpoint_path = self.validation_dir / f"checkpoint_{name}.json"
# Add timestamp and hash
data["timestamp"] = str(datetime.datetime.now())
data["hash"] = hashlib.sha256(str(data).encode()).hexdigest()
with open(checkpoint_path, 'w') as f:
json.dump(data, f, indent=2)
return str(checkpoint_path)
def load_checkpoint(self, name: str) -> Optional[Dict[str, Any]]:
"""Load a validation checkpoint"""
checkpoint_path = self.validation_dir / f"checkpoint_{name}.json"
if not checkpoint_path.exists():
return None
with open(checkpoint_path, 'r') as f:
return json.load(f)
def validate_dataset(self, dataset: List[Dict[str, Any]]) -> Dict[str, Any]:
"""Validate a complete dataset"""
results = []
error_count = 0
for idx, example in enumerate(dataset):
try:
# Validate equations
if "equation" in example:
eq_result = self.validate_equation(
example["equation"],
example.get("expected_result", "")
)
results.append(eq_result)
# Validate proofs
if "proof_steps" in example:
proof_result = self.validate_proof(
example["proof_steps"],
example.get("theorem", "")
)
results.append(proof_result)
except Exception as e:
error_count += 1
results.append({
"error": str(e),
"validation_score": 0.0
})
# Calculate overall metrics
scores = [r["validation_score"] for r in results if "validation_score" in r]
if scores:
avg_score = np.mean(scores)
else:
avg_score = 0.0
return {
"total_examples": len(dataset),
"processed_examples": len(results),
"error_count": error_count,
"average_score": float(avg_score),
"detailed_results": results
}
def save_validation_report(self, report: Dict[str, Any], name: str = None) -> str:
"""Save a validation report"""
if name is None:
name = datetime.datetime.now().strftime("%Y%m%d_%H%M%S")
report_path = self.validation_dir / f"report_{name}.json"
# Add timestamp and summary metrics
report["timestamp"] = str(datetime.datetime.now())
report["summary"] = {
"accuracy": report.get("average_score", 0.0),
"error_rate": report.get("error_count", 0) / report.get("total_examples", 1)
}
with open(report_path, 'w') as f:
json.dump(report, f, indent=2)
return str(report_path)
|