root/llama.cpp
1
0
Fork 0
mirror of https://github.com/ggerganov/llama.cpp.git synced 2024-11-11 21:39:52 +00:00
Code Issues Actions 12 Packages Projects Releases Wiki Activity

Update server.cpp

Browse Source
This commit is contained in:
Maximilian Winter 2024-05-12 12:06:36 +02:00
parent b228aba91a
commit f547c4f54a
1 changed files with 18 additions and 1 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

19
examples/server/server.cpp
View File

@ -2029,7 +2029,24 @@ struct server_context {
// reuse any previously computed tokens that are common with the new prompt // reuse any previously computed tokens that are common with the new prompt
slot.n_past = common_part(slot.cache_tokens, prompt_tokens); slot.n_past = common_part(slot.cache_tokens, prompt_tokens);
if (slot.ga_n != 1)
{
int ga_i = 0;
int32_t ga_n = slot.ga_n;
int32_t ga_w = slot.ga_w;
int32_t slot_npast = 0;
for (int k = 0; k < slot.n_past; ++k)
{
while (slot_npast >= ga_i + ga_w) {
const int bd = (ga_w/ga_n)*(ga_n - 1);
slot_npast -= bd;
ga_i += ga_w/ga_n;
}
slot_npast++;
}
slot.n_past_se = slot_npast;
slot.ga_i = ga_i;
}
// push the prompt into the sampling context (do not apply grammar) // push the prompt into the sampling context (do not apply grammar)
for (int i = 0; i < slot.n_past; ++i) { for (int i = 0; i < slot.n_past; ++i) {
llama_sampling_accept(slot.ctx_sampling, ctx, slot.cache_tokens[i], false); llama_sampling_accept(slot.ctx_sampling, ctx, slot.cache_tokens[i], false);