From 9d533a77d0c3850ce09d736bc1baa67fd6ad27b3 Mon Sep 17 00:00:00 2001 From: Georgi Gerganov Date: Tue, 27 Feb 2024 14:35:51 +0200 Subject: llama : fix defrag bugs + add parameter (#5735) * llama : fix defrag bugs + enable by default ggml-ci * llama : add defrag_thold parameter ggml-ci * llama : cont * llama : disable log message ggml-ci * llama : fix graph size check during defrag --- examples/passkey/passkey.cpp | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) (limited to 'examples/passkey/passkey.cpp') diff --git a/examples/passkey/passkey.cpp b/examples/passkey/passkey.cpp index 47de67a9..2cbc9e1f 100644 --- a/examples/passkey/passkey.cpp +++ b/examples/passkey/passkey.cpp @@ -182,7 +182,7 @@ int main(int argc, char ** argv) { llama_kv_cache_seq_rm (ctx, 0, n_keep , n_keep + n_discard); llama_kv_cache_seq_add(ctx, 0, n_keep + n_discard, n_ctx, -n_discard); - llama_kv_cache_defrag (ctx); + //llama_kv_cache_defrag (ctx); llama_kv_cache_update (ctx); n_past = llama_kv_cache_seq_pos_max(ctx, 0) + 1; @@ -213,7 +213,7 @@ int main(int argc, char ** argv) { llama_kv_cache_seq_rm (ctx, 0, n_keep , n_keep + n_discard); llama_kv_cache_seq_add(ctx, 0, n_keep + n_discard, n_ctx, -n_discard); - llama_kv_cache_defrag (ctx); + //llama_kv_cache_defrag (ctx); llama_kv_cache_update (ctx); n_past = llama_kv_cache_seq_pos_max(ctx, 0) + 1; -- cgit v1.2.3