ikawrakow/ik_llama.cpp
1
0
Fork 0
mirror of https://github.com/ikawrakow/ik_llama.cpp.git synced 2026-02-28 00:54:09 +00:00
Code Issues Packages Projects Releases Wiki Activity

A hopefully more efficient adaptive_p sampling (#1161)

Browse Source 
* A hopefully more efficient adaptive_p sampling

* Once at it, lets fix the formatting too

* More formatting

* Correctly accumulate sampling time for adaptive_p
This commit is contained in:
Kawrakow
2026-01-19 15:01:55 +02:00
committed by GitHub
parent 6a5c180be9
commit fa58c20c42
5 changed files with 96 additions and 53 deletions
Download Patch File Download Diff File Expand all files Collapse all files

16
src/llama.cpp
View File

@@ -7690,14 +7690,12 @@ void llama_sample_dry([[maybe_unused]] struct llama_context* ctx, struct llama_s
void llama_sample_adaptive_p(
[[maybe_unused]] struct llama_context * ctx,
llama_token_data_array * candidates,
struct llama_sampler_adaptive_p * adapt_p_ctx)
{
llama_sample_adaptive_p_impl(candidates, adapt_p_ctx);
struct llama_sampler_adaptive_p * adapt_p_ctx) {
llama_sample_adaptive_p_impl(&ctx->sampling, candidates, adapt_p_ctx);
}
void llama_prep_adaptive_p(llama_token_data_array * candidates, struct llama_sampler_adaptive_p * adapt_p_ctx)
{
llama_prep_adaptive_p_impl(candidates, adapt_p_ctx);
void llama_prep_adaptive_p(struct llama_context * ctx, llama_token_data_array * candidates, struct llama_sampler_adaptive_p * adapt_p_ctx) {
llama_prep_adaptive_p_impl(&ctx->sampling, candidates, adapt_p_ctx);
}
@@ -7743,8 +7741,7 @@ llama_token llama_sample_token(struct llama_context * ctx, llama_token_data_arra
llama_token llama_sample_token_adaptive_p(
struct llama_context * ctx,
llama_token_data_array * candidates,
struct llama_sampler_adaptive_p * adapt_p_ctx)
{
struct llama_sampler_adaptive_p * adapt_p_ctx) {
return llama_sample_token_adaptive_p_impl(&ctx->sampling, candidates, adapt_p_ctx);
}
@@ -7800,8 +7797,7 @@ void llama_sampler_dry_accept(struct llama_sampler_dry* smpl, llama_token token)
}
struct llama_sampler_adaptive_p * llama_init_adaptive_p(const float target, const float decay, const uint32_t seed)
{
struct llama_sampler_adaptive_p * llama_init_adaptive_p(const float target, const float decay, const uint32_t seed) {
return llama_init_adaptive_p_impl(target, decay, seed);
}