vrr/koboldcpp
2
0
Fork 0
mirror of https://github.com/LostRuins/koboldcpp.git synced 2025-09-11 09:34:37 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions

Merge branch 'upstream' into concedo_experimental

Browse source 
# Conflicts:
#	common/sampling.h
#	llama.h
#	tests/test-chat-template.cpp
This commit is contained in:
Concedo 2024-04-24 21:29:07 +08:00
commit a681cdd9ef
20 changed files with 788 additions and 355 deletions
Download patch file Download diff file Expand all files Collapse all files

13
common/sampling.cpp
View file

@ -1,4 +1,6 @@
#define LLAMA_API_INTERNAL
#include "sampling.h"
#include <random>
struct llama_sampling_context * llama_sampling_init(const struct llama_sampling_params & params) {
struct llama_sampling_context * result = new llama_sampling_context();
@ -33,6 +35,8 @@ struct llama_sampling_context * llama_sampling_init(const struct llama_sampling_
result->prev.resize(params.n_prev);
llama_sampling_set_rng_seed(result, params.seed);
return result;
}
@ -62,6 +66,13 @@ void llama_sampling_reset(llama_sampling_context * ctx) {
ctx->cur.clear();
}
void llama_sampling_set_rng_seed(struct llama_sampling_context * ctx, uint32_t seed) {
if (seed == LLAMA_DEFAULT_SEED) {
seed = time(NULL);
}
ctx->rng.seed(seed);
}
void llama_sampling_cp(llama_sampling_context * src, llama_sampling_context * dst) {
if (dst->grammar) {
llama_grammar_free(dst->grammar);
@ -203,7 +214,7 @@ static llama_token llama_sampling_sample_impl(
sampler_queue(ctx_main, params, cur_p, min_keep);
id = llama_sample_token(ctx_main, &cur_p);
id = llama_sample_token_with_rng(ctx_main, &cur_p, ctx_sampling->rng);
//{
// const int n_top = 10;