Spaces:
Runtime error
Runtime error
File size: 12,133 Bytes
12af533 |
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 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 |
"""
REALIZATION ENGINE
==================
Implementation of the crystallization framework discovered in our conversation.
Core Concepts:
- Realizations have quality scores (Q) based on 6 features
- Realizations crystallize into layers based on Q scores
- Layers form a hierarchy (0 > 1 > 2 > N)
- Retrieval follows O(log n) pattern: check highest layer first, descend if not found
"""
import json
import re
from typing import Dict, List, Optional, Tuple
from dataclasses import dataclass, asdict
from datetime import datetime
import hashlib
@dataclass
class RealizationFeatures:
"""The 6 features that determine realization quality"""
grounding: float # 0-1: How rooted in facts/rules
certainty: float # 0-1: Precision auto quality (self-certifying)
structure: float # 0-1: Crystallization clarity
applicability: float # 0-1: Actionability/usefulness
coherence: float # 0-1: Consistency with prior layers
generativity: float # 0-1: Daughters potential (بنات افكار)
def validate(self):
"""Ensure all features are in valid range"""
for name, value in asdict(self).items():
if not 0 <= value <= 1:
raise ValueError(f"{name} must be between 0 and 1, got {value}")
@dataclass
class Realization:
"""A single realization with metadata"""
id: str
content: str
features: RealizationFeatures
q_score: float
layer: int
timestamp: str
parents: List[str] # IDs of realizations this builds on
children: List[str] # IDs of realizations spawned from this
turn_number: int
# Metadata
context: str = "" # Surrounding conversation
evidence: List[str] = None # Supporting facts
def __post_init__(self):
if self.evidence is None:
self.evidence = []
class RealizationEngine:
"""
The core engine for managing realizations.
Implements:
- Q-score calculation
- Layer assignment
- Hierarchical storage
- O(log n) retrieval
- Invalidation strategies
"""
# Feature weights for Q-score calculation
WEIGHTS = {
'grounding': 0.18,
'certainty': 0.22, # Highest - certainty IS the realization signal
'structure': 0.20,
'applicability': 0.18,
'coherence': 0.12,
'generativity': 0.10
}
# Layer thresholds
LAYER_THRESHOLDS = {
0: 0.95, # Universal rules (rarely achieved)
1: 0.92, # Domain facts
2: 0.85, # Patterns
3: 0.75, # Situational insights
'N': 0.0 # Everything else (ephemeral)
}
def __init__(self):
# Storage: layer -> {id -> Realization}
self.layers = {
0: {}, # Universal rules
1: {}, # Domain facts
2: {}, # Patterns
3: {}, # Situational
'N': {} # Ephemeral
}
# Index for fast lookup
self.index = {} # id -> Realization
# Metadata
self.stats = {
'total_realizations': 0,
'layer_distribution': {0: 0, 1: 0, 2: 0, 3: 0, 'N': 0},
'avg_q_score': 0.0
}
def calculate_q_score(self, features: RealizationFeatures) -> Tuple[float, str]:
"""
Calculate quality score using weighted sum.
Returns:
(q_score, calculation_string)
"""
features.validate()
calc_parts = []
total = 0.0
for name, weight in self.WEIGHTS.items():
value = getattr(features, name)
contribution = weight * value
total += contribution
calc_parts.append(f"{weight}×{value:.2f}")
calc_string = " + ".join(calc_parts) + f" = {total:.4f}"
return round(total, 4), calc_string
def assign_layer(self, q_score: float, features: RealizationFeatures) -> int:
"""
Assign realization to appropriate layer based on Q-score and features.
Layer assignment rules:
- Q ≥ 0.95 AND Grounding ≥ 0.90 → Layer 0 (Universal Rule)
- Q ≥ 0.92 → Layer 1 (Domain Fact)
- Q ≥ 0.85 → Layer 2 (Pattern)
- Q ≥ 0.75 → Layer 3 (Situational)
- Q < 0.75 → Layer N (Ephemeral)
"""
if q_score >= 0.95 and features.grounding >= 0.90:
return 0
elif q_score >= 0.92:
return 1
elif q_score >= 0.85:
return 2
elif q_score >= 0.75:
return 3
else:
return 'N'
def generate_id(self, content: str) -> str:
"""Generate unique ID for realization based on content hash"""
hash_obj = hashlib.sha256(content.encode())
return f"R_{hash_obj.hexdigest()[:8]}"
def add_realization(
self,
content: str,
features: RealizationFeatures,
turn_number: int,
parents: List[str] = None,
context: str = "",
evidence: List[str] = None
) -> Realization:
"""
Add a new realization to the system.
Automatically calculates Q-score and assigns to layer.
"""
if parents is None:
parents = []
# Calculate Q-score
q_score, calc_string = self.calculate_q_score(features)
# Assign layer
layer = self.assign_layer(q_score, features)
# Generate ID
r_id = self.generate_id(content)
# Create realization
realization = Realization(
id=r_id,
content=content,
features=features,
q_score=q_score,
layer=layer,
timestamp=datetime.now().isoformat(),
parents=parents,
children=[],
turn_number=turn_number,
context=context,
evidence=evidence or []
)
# Store in appropriate layer
self.layers[layer][r_id] = realization
self.index[r_id] = realization
# Update parent-child relationships
for parent_id in parents:
if parent_id in self.index:
self.index[parent_id].children.append(r_id)
# Update stats
self.stats['total_realizations'] += 1
self.stats['layer_distribution'][layer] += 1
self._update_avg_q()
print(f"✅ Crystallized: {content[:60]}...")
print(f" Q = {q_score:.4f} ({calc_string})")
print(f" Layer {layer}")
print()
return realization
def retrieve(self, query: str, similarity_threshold: float = 0.5) -> List[Realization]:
"""
Retrieve realizations matching query.
Uses hierarchical search: start at Layer 0, descend if needed.
"""
results = []
# Search from highest layer to lowest
for layer in [0, 1, 2, 3, 'N']:
layer_results = self._search_layer(layer, query, similarity_threshold)
results.extend(layer_results)
# If we found high-quality results, stop (optimization)
if layer_results and layer in [0, 1]:
break
# Sort by Q-score descending
results.sort(key=lambda r: r.q_score, reverse=True)
return results
def _search_layer(self, layer: int, query: str, threshold: float) -> List[Realization]:
"""Search within a specific layer"""
results = []
query_lower = query.lower()
for realization in self.layers[layer].values():
# Simple keyword matching (could be enhanced with embeddings)
content_lower = realization.content.lower()
# Check for keyword matches
query_words = set(query_lower.split())
content_words = set(content_lower.split())
overlap = len(query_words & content_words)
if overlap > 0 or query_lower in content_lower:
results.append(realization)
return results
def get_realization_tree(self, r_id: str, depth: int = 3) -> Dict:
"""
Get realization and its family tree (parents + children).
Returns hierarchical structure showing بنات افكار (daughters of ideas).
"""
if r_id not in self.index:
return None
realization = self.index[r_id]
tree = {
'id': r_id,
'content': realization.content,
'q_score': realization.q_score,
'layer': realization.layer,
'parents': [],
'children': []
}
if depth > 0:
# Get parents
for parent_id in realization.parents:
parent_tree = self.get_realization_tree(parent_id, depth - 1)
if parent_tree:
tree['parents'].append(parent_tree)
# Get children (بنات افكار)
for child_id in realization.children:
child_tree = self.get_realization_tree(child_id, depth - 1)
if child_tree:
tree['children'].append(child_tree)
return tree
def _update_avg_q(self):
"""Update average Q-score statistic"""
if self.stats['total_realizations'] == 0:
self.stats['avg_q_score'] = 0.0
else:
total_q = sum(r.q_score for r in self.index.values())
self.stats['avg_q_score'] = total_q / self.stats['total_realizations']
def export_state(self) -> Dict:
"""Export entire state as JSON-serializable dict"""
return {
'layers': {
str(k): {r_id: self._realization_to_dict(r)
for r_id, r in v.items()}
for k, v in self.layers.items()
},
'stats': self.stats,
'timestamp': datetime.now().isoformat()
}
def _realization_to_dict(self, r: Realization) -> Dict:
"""Convert Realization to dict"""
return {
'id': r.id,
'content': r.content,
'features': asdict(r.features),
'q_score': r.q_score,
'layer': r.layer,
'timestamp': r.timestamp,
'parents': r.parents,
'children': r.children,
'turn_number': r.turn_number,
'context': r.context,
'evidence': r.evidence
}
def print_stats(self):
"""Print system statistics"""
print("\n" + "="*60)
print("REALIZATION ENGINE STATISTICS")
print("="*60)
print(f"Total Realizations: {self.stats['total_realizations']}")
print(f"Average Q-Score: {self.stats['avg_q_score']:.4f}")
print("\nLayer Distribution:")
for layer in [0, 1, 2, 3, 'N']:
count = self.stats['layer_distribution'][layer]
pct = (count / self.stats['total_realizations'] * 100) if self.stats['total_realizations'] > 0 else 0
layer_name = {
0: "Universal Rules",
1: "Domain Facts",
2: "Patterns",
3: "Situational",
'N': "Ephemeral"
}[layer]
print(f" Layer {layer} ({layer_name}): {count} ({pct:.1f}%)")
print("="*60 + "\n")
if __name__ == "__main__":
# Quick test
engine = RealizationEngine()
# Test realization
features = RealizationFeatures(
grounding=0.95,
certainty=0.93,
structure=0.92,
applicability=0.90,
coherence=0.95,
generativity=0.92
)
r = engine.add_realization(
content="Realizations crystallize into layers (بنات افكار)",
features=features,
turn_number=1,
evidence=["Observable in conversation", "Matches how knowledge accumulates"]
)
engine.print_stats()
|