vbatts/llama.cpp
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

minor : comments + rename

Browse source 
ggml-ci
This commit is contained in:
Georgi Gerganov 2023-10-16 18:17:31 +03:00
parent 1c626e2fe1
commit 373d782d42
No known key found for this signature in database
GPG key ID: 449E073F9DC10735
4 changed files with 13 additions and 7 deletions
Download patch file Download diff file Expand all files Collapse all files

6
examples/main/main.cpp
View file

@ -612,8 +612,14 @@ int main(int argc, char ** argv) {
LOG("embd_inp.size(): %d, n_consumed: %d\n", (int) embd_inp.size(), n_consumed);
while ((int) embd_inp.size() > n_consumed) {
embd.push_back(embd_inp[n_consumed]);
// GG: I'm not sure it's a good idea to push the prompt tokens into the sampling context
// Most likely will remove this in the future to avoid exposing "prev"
// Same thing is done in "server". If we stop pushing the prompt tokens, then the repetition
// penalty will be applied only based on the tokens generated by the model.
ctx_sampling->prev.erase(ctx_sampling->prev.begin());
ctx_sampling->prev.push_back(embd_inp[n_consumed]);
++n_consumed;
if ((int) embd.size() >= params.n_batch) {
break;