flashinfer.cake_fmha.cake_batch_context_with_kv_cache

flashinfer.cake_fmha.cake_batch_context_with_kv_cache(*args, **kwargs)

Run the FlashInfer TRTLLM paged-context ABI through Cake FMHA.

Parameters and return values match flashinfer.trtllm_batch_context_with_kv_cache(). The selection is explicit and never replaces FlashInfer’s conventional default.