warm_keep_util.h File Reference

gh#96 (v2.7.5) warm-keep / incremental-prefill decision logic. More...

#include <algorithm>
#include <cstddef>
#include <vector>

Include dependency graph for warm_keep_util.h:

This graph shows which files directly or indirectly include this file:

Go to the source code of this file.

Namespaces
namespace	entropic
	Activate model on GPU (WARM → ACTIVE).

Functions
template<typename Tok >
std::size_t	entropic::common_prefix_len (const std::vector< Tok > &a, const std::vector< Tok > &b)
	Length of the longest common token prefix of two sequences.
template<typename Tok >
std::size_t	entropic::warm_keep_cut (const std::vector< Tok > &resident, const std::vector< Tok > &incoming, long kv_pos_max)
	Decide how many resident-KV tokens warm-keep may reuse this turn.

Detailed Description

gh#96 (v2.7.5) warm-keep / incremental-prefill decision logic.

Pure functions (no llama.cpp, no I/O) so the correctness-critical reuse decision is unit-testable on the CPU pre-commit tier. The backend wraps these with the actual KV mutation (seq_rm + delta decode).

Version

2.7.5

Definition in file warm_keep_util.h.