ikawrakow/ik_llama.cpp
1
0
Fork 0
mirror of https://github.com/ikawrakow/ik_llama.cpp.git synced 2026-02-23 06:34:13 +00:00
Code Issues Packages Projects Releases Wiki Activity

Correctly accumulate adaptive_p sampling time

Browse Source
This commit is contained in:
Kawrakow
2026-01-19 10:00:19 +00:00
parent 4df3251b12
commit bd2434945d
5 changed files with 25 additions and 17 deletions
Download Patch File Download Diff File Expand all files Collapse all files

9
include/llama.h
View File

@@ -1389,15 +1389,14 @@ LLAMA_API struct llama_grammar* llama_sampler_init_grammar_lazy_patterns(
const float decay,
const uint32_t seed);
void llama_prep_adaptive_p(
void llama_prep_adaptive_p(struct llama_context * ctx,
llama_token_data_array * candidates,
struct llama_sampler_adaptive_p * adapt_p_ctx);
/// @details Adaptive p sampler described in https://github.com/MrJackSpade/adaptive-p-docs/blob/main/README.md
void llama_sample_adaptive_p(
struct llama_context * ctx,
llama_token_data_array * candidates,
struct llama_sampler_adaptive_p * adapt_p_ctx);
void llama_sample_adaptive_p(struct llama_context * ctx,
llama_token_data_array * candidates,
struct llama_sampler_adaptive_p * adapt_p_ctx);
/// @details Mirostat 1.0 algorithm described in the paper https://arxiv.org/abs/2007.14966. Uses tokens instead of words.