[None][fix] Fix compute token accounting for KV cache reuse with context chunking - #12682

[None][fix] Fix compute token accounting for KV cache reuse with context chunking#12682

lancelly merged 2 commits into

NVIDIA:feat/bench_yNVIDIA/TensorRT-LLM:feat/bench_yfrom

lancelly:reuse-budget-fix-on-feat-bench-ylancelly/TensorRT-LLM:reuse-budget-fix-on-feat-bench-yCopy head branch name to clipboard