From 4eb126fff09449ab957c1646fa9e876efe15c13d Mon Sep 17 00:00:00 2001 From: Georgi Gerganov Date: Sun, 24 Nov 2024 15:39:07 +0200 Subject: [PATCH] common : change defaults [no ci] --- common/common.h | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/common/common.h b/common/common.h index f354a5fbe..c9fb2b62a 100644 --- a/common/common.h +++ b/common/common.h @@ -157,8 +157,8 @@ struct common_params_sampling { struct common_params_speculative { int32_t n_ctx = 4096; // draft context size - int32_t n_max = 5; // maximum number of tokens to draft during speculative decoding - int32_t n_min = 0; // minimum number of draft tokens to use for speculative decoding + int32_t n_max = 16; // maximum number of tokens to draft during speculative decoding + int32_t n_min = 5; // minimum number of draft tokens to use for speculative decoding int32_t n_gpu_layers = -1; // number of layers to store in VRAM for the draft model (-1 - use default) float p_split = 0.1f; // speculative decoding split probability float p_min = 0.9f; // minimum speculative decoding probability (greedy)