Spaces:
Sleeping
Sleeping
File size: 12,776 Bytes
5e1a30c |
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 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 |
"""
Performance Optimization for Neural Reranking.
This module provides performance optimization capabilities including
caching, batch processing, latency optimization, and resource management
to ensure neural reranking meets the <200ms additional latency target.
Simplified from reranking/performance_optimizer.py for integration with
the enhanced neural reranker in the rerankers/ component.
"""
import logging
import time
import hashlib
from typing import List, Dict, Any, Optional, Tuple
from collections import OrderedDict
import threading
from src.core.interfaces import Document
logger = logging.getLogger(__name__)
class LRUCache:
"""Thread-safe LRU cache for neural reranking scores."""
def __init__(self, max_size: int, ttl_seconds: int = 3600):
"""
Initialize LRU cache.
Args:
max_size: Maximum number of entries
ttl_seconds: Time-to-live for cache entries
"""
self.max_size = max_size
self.ttl_seconds = ttl_seconds
self.cache = OrderedDict()
self.timestamps = {}
self._lock = threading.Lock()
self.stats = {
"hits": 0,
"misses": 0,
"evictions": 0,
"ttl_expirations": 0
}
def get(self, key: str) -> Optional[List[float]]:
"""Get cached scores by key."""
with self._lock:
current_time = time.time()
# Check if key exists and is not expired
if key in self.cache:
if current_time - self.timestamps[key] < self.ttl_seconds:
# Move to end (most recently used)
value = self.cache.pop(key)
self.cache[key] = value
self.stats["hits"] += 1
return value
else:
# Expired
del self.cache[key]
del self.timestamps[key]
self.stats["ttl_expirations"] += 1
self.stats["misses"] += 1
return None
def put(self, key: str, value: List[float]):
"""Put scores in cache."""
with self._lock:
current_time = time.time()
# Remove if already exists
if key in self.cache:
del self.cache[key]
del self.timestamps[key]
# Add new entry
self.cache[key] = value
self.timestamps[key] = current_time
# Check size limit
while len(self.cache) > self.max_size:
# Remove least recently used
oldest_key = next(iter(self.cache))
del self.cache[oldest_key]
del self.timestamps[oldest_key]
self.stats["evictions"] += 1
def clear(self):
"""Clear all cache entries."""
with self._lock:
self.cache.clear()
self.timestamps.clear()
def get_stats(self) -> Dict[str, Any]:
"""Get cache statistics."""
with self._lock:
total_requests = self.stats["hits"] + self.stats["misses"]
hit_rate = self.stats["hits"] / total_requests if total_requests > 0 else 0
return {
**self.stats,
"size": len(self.cache),
"max_size": self.max_size,
"hit_rate": hit_rate,
"ttl_seconds": self.ttl_seconds
}
class BatchProcessor:
"""Optimized batch processing for neural reranking."""
def __init__(
self,
min_batch_size: int = 1,
max_batch_size: int = 64,
timeout_ms: int = 50
):
"""
Initialize batch processor.
Args:
min_batch_size: Minimum batch size
max_batch_size: Maximum batch size
timeout_ms: Batch timeout in milliseconds
"""
self.min_batch_size = min_batch_size
self.max_batch_size = max_batch_size
self.timeout_ms = timeout_ms
self.stats = {
"batches_processed": 0,
"total_items": 0,
"avg_batch_size": 0,
"timeout_batches": 0
}
def optimize_batch_size(
self,
items: List[Any],
target_latency_ms: int = 150
) -> int:
"""
Optimize batch size based on item count and latency targets.
Args:
items: Items to process
target_latency_ms: Target latency in milliseconds
Returns:
Optimal batch size
"""
item_count = len(items)
# Start with configured max batch size
optimal_size = min(self.max_batch_size, item_count)
# Adjust based on latency target
if target_latency_ms < 100:
# Very tight latency - use smaller batches
optimal_size = min(optimal_size, 16)
elif target_latency_ms > 300:
# Looser latency - can use larger batches
optimal_size = min(optimal_size, 64)
# Ensure minimum batch size
optimal_size = max(self.min_batch_size, optimal_size)
return optimal_size
def create_batches(
self,
items: List[Any],
batch_size: Optional[int] = None
) -> List[List[Any]]:
"""
Create optimized batches from items.
Args:
items: Items to batch
batch_size: Override batch size (optional)
Returns:
List of batches
"""
if not items:
return []
if batch_size is None:
batch_size = self.optimize_batch_size(items)
batches = []
for i in range(0, len(items), batch_size):
batch = items[i:i + batch_size]
batches.append(batch)
# Update statistics
self.stats["batches_processed"] += len(batches)
self.stats["total_items"] += len(items)
if self.stats["batches_processed"] > 0:
self.stats["avg_batch_size"] = self.stats["total_items"] / self.stats["batches_processed"]
return batches
def get_stats(self) -> Dict[str, Any]:
"""Get batch processing statistics."""
return self.stats.copy()
class PerformanceOptimizer:
"""
Performance optimizer for neural reranking.
Provides caching, batch processing, and latency optimization
to ensure neural reranking meets performance targets.
"""
def __init__(
self,
max_latency_ms: int = 200,
target_latency_ms: int = 150,
enable_caching: bool = True,
cache_ttl_seconds: int = 3600,
max_cache_size: int = 10000,
dynamic_batching: bool = True,
min_batch_size: int = 1,
max_batch_size: int = 64
):
"""
Initialize performance optimizer.
Args:
max_latency_ms: Maximum allowed latency
target_latency_ms: Target latency for optimization
enable_caching: Whether to enable caching
cache_ttl_seconds: Cache time-to-live
max_cache_size: Maximum cache entries
dynamic_batching: Whether to enable dynamic batching
min_batch_size: Minimum batch size
max_batch_size: Maximum batch size
"""
self.max_latency_ms = max_latency_ms
self.target_latency_ms = target_latency_ms
self.enable_caching = enable_caching
self.dynamic_batching = dynamic_batching
# Initialize cache
self.cache = LRUCache(max_cache_size, cache_ttl_seconds) if enable_caching else None
# Initialize batch processor
self.batch_processor = BatchProcessor(
min_batch_size, max_batch_size
) if dynamic_batching else None
self.stats = {
"optimizations": 0,
"cache_enabled": enable_caching,
"batching_enabled": dynamic_batching,
"fallbacks": 0,
"latency_violations": 0
}
logger.info(f"PerformanceOptimizer initialized (cache={enable_caching}, batching={dynamic_batching})")
def generate_cache_key(
self,
query: str,
documents: List[Document],
model_name: str
) -> str:
"""
Generate cache key for query-documents-model combination.
Args:
query: Search query
documents: List of documents
model_name: Model name
Returns:
Cache key string
"""
# Create a hash of query + document IDs + model name
content = f"{query}|{model_name}"
# Add document identifiers
doc_ids = []
for doc in documents:
if hasattr(doc, 'id') and doc.id:
doc_ids.append(str(doc.id))
else:
# Fallback to content hash
doc_hash = hashlib.md5(doc.content.encode()).hexdigest()[:8]
doc_ids.append(doc_hash)
content += "|" + ",".join(doc_ids)
return hashlib.md5(content.encode()).hexdigest()
def get_cached_scores(
self,
query: str,
documents: List[Document],
model_name: str
) -> Optional[List[float]]:
"""
Get cached scores if available.
Args:
query: Search query
documents: List of documents
model_name: Model name
Returns:
Cached scores or None
"""
if not self.enable_caching or not self.cache:
return None
cache_key = self.generate_cache_key(query, documents, model_name)
return self.cache.get(cache_key)
def cache_scores(
self,
query: str,
documents: List[Document],
model_name: str,
scores: List[float]
):
"""
Cache scores for future use.
Args:
query: Search query
documents: List of documents
model_name: Model name
scores: Scores to cache
"""
if not self.enable_caching or not self.cache:
return
cache_key = self.generate_cache_key(query, documents, model_name)
self.cache.put(cache_key, scores)
def optimize_batch_size(
self,
query_doc_pairs: List[List[str]]
) -> int:
"""
Optimize batch size for processing.
Args:
query_doc_pairs: Query-document pairs
Returns:
Optimal batch size
"""
if not self.dynamic_batching or not self.batch_processor:
return len(query_doc_pairs)
return self.batch_processor.optimize_batch_size(
query_doc_pairs, self.target_latency_ms
)
def should_use_fallback(self, estimated_latency_ms: float) -> bool:
"""
Determine if fallback should be used based on latency estimate.
Args:
estimated_latency_ms: Estimated processing latency
Returns:
True if fallback should be used
"""
return estimated_latency_ms > self.max_latency_ms
def record_latency(self, actual_latency_ms: float):
"""
Record actual latency for optimization learning.
Args:
actual_latency_ms: Measured latency
"""
if actual_latency_ms > self.max_latency_ms:
self.stats["latency_violations"] += 1
self.stats["optimizations"] += 1
def get_stats(self) -> Dict[str, Any]:
"""Get performance optimization statistics."""
stats = self.stats.copy()
if self.cache:
stats["cache"] = self.cache.get_stats()
if self.batch_processor:
stats["batch_processor"] = self.batch_processor.get_stats()
# Calculate performance metrics
if self.stats["optimizations"] > 0:
stats["latency_violation_rate"] = self.stats["latency_violations"] / self.stats["optimizations"]
return stats
def reset_stats(self) -> None:
"""Reset optimization statistics."""
self.stats = {
"optimizations": 0,
"cache_enabled": self.enable_caching,
"batching_enabled": self.dynamic_batching,
"fallbacks": 0,
"latency_violations": 0
} |