Unverified50% confidenceFactExact time
In the bidirectional KV transfer mechanism, prefill passes the initial KV to decode, and the incremental KV produced during decode generation also flows back to the prefill side, so that each round performs incremental prefill on top of the previous round rather than full recomputation.
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
10/19/2026
First Seen
Valid until: 1/17/2027
Sources
Related Entities
Related Claims
UnverifiedPrefix cache reuse causes KV values to be computed in a different batch than a fresh prefill, so the KV values from a cache hit may differ at the floating-point level from those of a fresh prefill, and the difference propagates through subsequent tokens.65% similarVerifiedLarge model inference is divided into two phases, Prefill and Decode; Prefill is a major component of Time to First Token (TTFT), while Decode corresponds to Time Per Output Token (TPOT).64% similarUnverifiedCoDR only re-masks and regenerates tokens that the model no longer endorses, rather than discarding and redoing the entire output.63% similarUnverifiedThe KV cache, as the carrier of the inference state of Transformer-type models, naturally mixes information from different sources and lacks a mechanism for isolation by relationship.63% similarVerifiedOutput token pricing is typically higher than input token pricing because the generation process requires autoregressive decoding token by token.62% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1070062API
curl https://kongchang.com/api/v1/knowledge/claims/1070062MCP
get_claim(id=1070062)