CS348Project/llama.cpp
5
0
Fork 0
mirror of https://github.com/ggml-org/llama.cpp.git synced 2025-10-31 08:51:55 +00:00
Code Issues Packages Projects Releases Wiki Activity

fix instruct mode (#445)

Browse Source 
changes to EOS behavior in interactive and reverse prompt handling broke instruct mode by erroneously injecting instruct mode's reverse prompt and an extra newline.
This commit is contained in:
rabidcopy
2023-03-24 10:22:39 -05:00
committed by GitHub
parent afd220d9c6
commit 863f65e2e3
1 changed files with 7 additions and 3 deletions
Download Patch File Download Diff File Expand all files Collapse all files

6
main.cpp
View File

@@ -387,7 +387,7 @@ int main(int argc, char ** argv) {
} }
// replace end of text token with newline token when in interactive mode // replace end of text token with newline token when in interactive mode
if (id == llama_token_eos() && params.interactive) { if (id == llama_token_eos() && params.interactive && !params.instruct) {
id = llama_token_newline.front(); id = llama_token_newline.front();
if (params.antiprompt.size() != 0) { if (params.antiprompt.size() != 0) {
// tokenize and inject first reverse prompt // tokenize and inject first reverse prompt
@@ -488,9 +488,13 @@ int main(int argc, char ** argv) {
// end of text token // end of text token
if (embd.back() == llama_token_eos()) { if (embd.back() == llama_token_eos()) {
if (params.instruct) {
is_interacting = true;
} else {
fprintf(stderr, " [end of text]\n"); fprintf(stderr, " [end of text]\n");
break; break;
} }
}
// In interactive mode, respect the maximum number of tokens and drop back to user input when reached. // In interactive mode, respect the maximum number of tokens and drop back to user input when reached.
if (params.interactive && remaining_tokens <= 0) { if (params.interactive && remaining_tokens <= 0) {