stage-replay-divergence-follows-the-kv-cache-fixed-prefix-precision-controls-and-bidirectional-cache-transplantation-33d19298·1 events·first seen Aliases: Stage-Replay Divergence Follows the KV Cache: Fixed-Prefix Precision Controls and Bidirectional Cache Transplantation
A preprint from arXiv investigates why stage-replay diagnostics for LLMs diverge from live inference, finding that KV cache state—not token identity—is the causally sufficient carrier of divergent trajectories in a Qwen2.5-derived system. A 200-item controlled experiment shows BF16 precision produces disagreements on 166 suffixes while FP32 eliminates decoded disagreements entirely. Bidirectional transplantation of all 48 KV layers causes continuations to follow the cache donor with 100% fidelity across two checkpoints (24/24 and 43/43). The findings have direct implications for reproducibility and debugging of multi-step reasoning systems.