furyhawk/llama.cpp
1
0
Fork 0
mirror of https://github.com/ggml-org/llama.cpp.git synced 2026-07-21 02:05:51 +00:00
Code Issues Packages Projects Releases Wiki Activity

server : add trace logs for slot prefix similarity computation

Browse Source 
Log the computed prefix similarity values for each slot during slot
selection, including the common prefix length, task token count,
current similarity, best similarity so far, and the threshold.

Assisted-by: pi:llama.cpp/Qwen3.6-27B
This commit is contained in:
Georgi Gerganov
2026-07-17 13:01:28 +03:00
parent 635cdd5fcc
commit 2c3ed6086f
1 changed files with 5 additions and 1 deletions
Download Patch File Download Diff File Expand all files Collapse all files
tools/server/server-context.cpp
+5 -1
View File
@@ -1557,7 +1557,11 @@ private:
}
// fraction of the Longest Common Prefix length with respect to the input prompt length
const float sim_cur = float(tokens.get_common_prefix(task.tokens)) / task.tokens.size();
const int32_t n_prefix = tokens.get_common_prefix(task.tokens);
const float sim_cur = float(n_prefix) / task.tokens.size();
SLT_TRC(slot, "prefix similarity check: n_prefix = %d, n_task_tokens = %d, sim_cur = %.3f, sim_best = %.3f, threshold = %.3f\n",
n_prefix, (int) task.tokens.size(), sim_cur, sim_best, slot_prompt_similarity);
// select the current slot if the criteria match
if (sim_cur > sim_best && sim_cur > slot_prompt_similarity) {