root/llama.cpp
1
0
Fork 0
mirror of https://github.com/ggerganov/llama.cpp.git synced 2024-12-24 02:14:35 +00:00
Code Issues Actions 11 Packages Projects Releases Wiki Activity

sampling: fix top_k <= 0 (#5388)

Browse Source 
* sampling: fix top_k <= 0

* Update llama.cpp

Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>

---------

Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>
This commit is contained in:
Johannes Gäßler 2024-02-08 09:46:30 +01:00 committed by GitHub
parent 8504d2d0da
commit 26d4efd11e
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
3 changed files with 7 additions and 1 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
common/sampling.cpp
View File

@ -132,7 +132,7 @@ static void sampler_queue(
const float temp = params.temp;
const float dynatemp_range = params.dynatemp_range;
const float dynatemp_exponent = params.dynatemp_exponent;
const int32_t top_k = params.top_k <= 0 ? n_vocab : params.top_k;
const int32_t top_k = params.top_k;
const float top_p = params.top_p;
const float min_p = params.min_p;
const float tfs_z = params.tfs_z;

4
llama.cpp
View File

@ -8585,6 +8585,10 @@ void llama_sample_top_k(struct llama_context * ctx, llama_token_data_array * can
// }
const int64_t t_start_sample_us = ggml_time_us();
if (k <= 0) {
k = candidates->size;
}
k = std::max(k, (int) min_keep);
k = std::min(k, (int) candidates->size);

2
tests/test-sampling.cpp
View File

@ -235,6 +235,8 @@ int main(void) {
test_top_k({0.1f, 0.2f, 0.3f, 0.4f}, {0.4f}, 1);
test_top_k({0.1f, 0.2f, 0.3f, 0.4f}, {0.4f, 0.3f, 0.2f}, 3);
test_top_k({0.1f, 0.2f, 0.3f, 0.4f}, {0.4f, 0.3f, 0.2f, 0.1f}, 4);
test_top_k({0.1f, 0.2f, 0.3f, 0.4f}, {0.4f, 0.3f, 0.2f, 0.1f}, 0);
test_top_p({0.1f, 0.2f, 0.3f, 0.4f}, {0.4f}, 0);
test_top_p({0.1f, 0.2f, 0.3f, 0.4f}, {0.4f, 0.3f}, 0.7f);