[Chore] Optimize L2Cache tokens comparison

Haiyang Shi · Haiyang Shi · commit bf4cc2140243 · 2025-06-05T10:31:23.000-07:00
Signed-off-by: Haiyang Shi &lt;haiyang.shi@bytedance.com&gt;
diff --git a/python/aibrix_kvcache/aibrix_kvcache/l2/l2_cache.py b/python/aibrix_kvcache/aibrix_kvcache/l2/l2_cache.py
@@ -530,7 +530,7 @@ def _tokens_match(
             if len(tokens_in_mr) != self.block_ntokens:
                 return False
             all_tokens = np_array_concat(prefix_in_mr, tokens_in_mr)
-            is_identical = all(all_tokens == real_key)
+            is_identical = np.all(all_tokens == real_key)
             if is_identical:
                 return True
             else:
