Files
modelTesting/outputs/muse-glimmer-28b-gguf-lfu.py
AygeaandClaude 2f99dd1e35 Grade muse-glimmer-28b full 7-prompt battery — new benchmark leader
7 entries (30→37 total). Muse Glimmer 28B (GGUF) avg 80.7 — the strongest
model in the benchmark, 6/7 prompts Minor Logic Flaws:
  lfu 76 | webhook 81 | automation 89 | rust 85 | data 88 | tts 58 | mcp 88

Standout results:
- rust 85 (KAT 36, Qwen3-Coder 54) — real tokio channels (mpsc::channel, not
  hallucinated mpsc::bounded), two-tier CancellationToken, zero clippy lints;
  one-line E0507 compile fix.
- automation 89 — first model to print a correct summary (98/2/0/100);
  atomic temp+fsync+rename checkpointing.
- data 88 edges out Gemma-26B's 86; mcp 88 sets the bar on a new prompt.
Only weak spot: tts 58 (backpressure raises instead of awaits, like Qwen3-Coder).

Captured via the native /api/v1/chat fix (real tok/sec + TTFT). Slow
deep-thinker: ~17-19 t/s, 5-9 min/prompt, ~5-9k tokens incl. reasoning.

Also gitignore checkpoint.json (automation test runtime artifact).

Co-Authored-By: Claude <noreply@anthropic.com>
2026-08-10 13:13:54 -07:00

271 lines
8.9 KiB
Python

# lfu_cache.py
import asyncio
import time
from typing import Any, Optional, Dict, Set
class _Node:
__slots__ = ("key", "value", "expire_at", "freq", "prev", "next")
def __init__(self, key=None, value=None, expire_at=None, freq=0):
self.key = key
self.value = value
self.expire_at = expire_at
self.freq = freq
self.prev = None
self.next = None
class _FreqList:
__slots__ = ("head", "tail")
def __init__(self):
self.head = _Node()
self.tail = _Node()
self.head.next = self.tail
self.tail.prev = self.head
class LFUCache:
def __init__(self, capacity: int, evict_interval: float = 0.5):
self.capacity = capacity
self.key_map: Dict[Any, _Node] = {}
self.freq_map: Dict[int, _FreqList] = {}
self.min_freq = 0
self.lock = asyncio.Lock()
self._evictor_task: Optional[asyncio.Task] = None
self._evict_interval = evict_interval
# ---------- internal helpers ----------
def _update_min_freq(self):
while self.min_freq not in self.freq_map:
self.min_freq += 1
def _add_node_to_head(self, freq: int, node: _Node):
lst = self.freq_map.get(freq)
if lst is None:
lst = _FreqList()
self.freq_map[freq] = lst
node.prev = lst.head
node.next = lst.head.next
lst.head.next.prev = node
lst.head.next = node
def _remove_node(self, node: _Node):
freq = node.freq
lst = self.freq_map.get(freq)
if lst:
node.prev.next = node.next
node.next.prev = node.prev
if lst.head.next is lst.tail:
del self.freq_map[freq]
if self.min_freq == freq:
self._update_min_freq()
if self.key_map.get(node.key) is node:
del self.key_map[node.key]
def _increment_freq(self, node: _Node):
old_freq = node.freq
lst = self.freq_map[old_freq]
node.prev.next = node.next
node.next.prev = node.prev
if lst.head.next is lst.tail:
del self.freq_map[old_freq]
if self.min_freq == old_freq:
self._update_min_freq()
new_freq = old_freq + 1
node.freq = new_freq
self._add_node_to_head(new_freq, node)
def _evict_one(self):
if not self.freq_map:
return
if self.min_freq not in self.freq_map:
self._update_min_freq()
lst = self.freq_map[self.min_freq]
node = lst.tail.prev
if node is lst.head:
return
self._remove_node(node)
def _is_expired(self, node: _Node) -> bool:
return node.expire_at is not None and time.monotonic() > node.expire_at
# ---------- public API ----------
async def get(self, key: Any) -> Optional[Any]:
async with self.lock:
node = self.key_map.get(key)
if not node:
return None
if self._is_expired(node):
self._remove_node(node)
return None
self._increment_freq(node)
return node.value
async def put(self, key: Any, value: Any, ttl_seconds: Optional[float] = None):
async with self.lock:
now = time.monotonic()
expire_at = now + ttl_seconds if ttl_seconds is not None else None
node = self.key_map.get(key)
if node:
if self._is_expired(node):
self._remove_node(node)
node = None
if node:
node.value = value
node.expire_at = expire_at
self._increment_freq(node)
return
if len(self.key_map) >= self.capacity:
self._evict_one()
node = _Node(key=key, value=value, expire_at=expire_at, freq=1)
self.key_map[key] = node
self._add_node_to_head(1, node)
if self.min_freq == 0 or 1 < self.min_freq:
self.min_freq = 1
def begin_transaction(self) -> "Transaction":
return Transaction(self)
async def start_evictor(self):
if self._evictor_task and not self._evictor_task.done():
return
self._evictor_task = asyncio.create_task(self._evictor_loop())
async def stop_evictor(self):
if self._evictor_task:
self._evictor_task.cancel()
try:
await self._evictor_task
except asyncio.CancelledError:
pass
self._evictor_task = None
async def _evictor_loop(self):
while True:
await asyncio.sleep(self._evict_interval)
async with self.lock:
now = time.monotonic()
batch = 100
count = 0
for k in list(self.key_map.keys()):
if count >= batch:
break
node = self.key_map.get(k)
if node and node.expire_at and now > node.expire_at:
self._remove_node(node)
count += 1
class Transaction:
def __init__(self, cache: LFUCache):
self.cache = cache
self.pending_puts: Dict[Any, tuple[Any, Optional[float]]] = {}
self.pending_deletes: Set[Any] = set()
async def get(self, key: Any) -> Optional[Any]:
if key in self.pending_deletes:
return None
if key in self.pending_puts:
val, _ = self.pending_puts[key]
return val
async with self.cache.lock:
node = self.cache.key_map.get(key)
if not node:
return None
if node.expire_at and time.monotonic() > node.expire_at:
self.cache._remove_node(node)
return None
return node.value
async def put(self, key: Any, value: Any, ttl_seconds: Optional[float] = None):
self.pending_puts[key] = (value, ttl_seconds)
self.pending_deletes.discard(key)
async def delete(self, key: Any):
self.pending_deletes.add(key)
self.pending_puts.pop(key, None)
async def commit(self):
async with self.cache.lock:
for key in self.pending_deletes:
node = self.cache.key_map.get(key)
if node:
self.cache._remove_node(node)
for key, (value, ttl_seconds) in self.pending_puts.items():
expire_at = None
if ttl_seconds is not None:
expire_at = time.monotonic() + ttl_seconds
node = self.cache.key_map.get(key)
if node:
node.value = value
node.expire_at = expire_at
self.cache._increment_freq(node)
else:
if len(self.cache.key_map) >= self.cache.capacity:
self.cache._evict_one()
node = _Node(key=key, value=value, expire_at=expire_at, freq=1)
self.cache.key_map[key] = node
self.cache._add_node_to_head(1, node)
if self.cache.min_freq == 0 or 1 < self.cache.min_freq:
self.cache.min_freq = 1
self.pending_puts.clear()
self.pending_deletes.clear()
def rollback(self):
self.pending_puts.clear()
self.pending_deletes.clear()
# ------------------- tests -------------------
async def main():
cache = LFUCache(capacity=3, evict_interval=0.2)
await cache.start_evictor()
print("=== LFU eviction order ===")
await cache.put("a", 1)
await cache.put("b", 2)
await cache.put("c", 3)
await cache.get("a")
await cache.get("a")
await cache.get("b")
await cache.put("d", 4) # should evict c
assert await cache.get("c") is None
assert await cache.get("a") == 1
print("LFU eviction OK")
print("=== TTL lazy vs background ===")
await cache.put("x", "temp", ttl_seconds=0.1)
await asyncio.sleep(0.15)
assert await cache.get("x") is None
await cache.put("y", "temp2", ttl_seconds=0.1)
await asyncio.sleep(0.3)
assert await cache.get("y") is None
print("TTL OK")
print("=== Transaction commit / rollback ===")
tx = cache.begin_transaction()
await tx.put("t1", 100)
assert await cache.get("t1") is None
await tx.commit()
assert await cache.get("t1") == 100
tx2 = cache.begin_transaction()
await tx2.put("t2", 200)
await tx2.rollback()
assert await cache.get("t2") is None
print("Transaction OK")
print("=== Stress test 50 concurrent tasks ===")
async def worker(i):
for j in range(20):
k = f"k{i%10}"
await cache.put(k, i*j, ttl_seconds=5)
await cache.get(k)
await asyncio.gather(*(worker(i) for i in range(50)))
print("Stress OK")
await cache.stop_evictor()
print("All tests passed")
if __name__ == "__main__":
asyncio.run(main())