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

Optimization: eliminate addition of redundant stacks when advancing grammar.

Browse source
This commit is contained in:
Clint Herron 2024-04-11 13:57:10 -04:00
parent cbaadc9294
commit 80553e53c0
1 changed files with 7 additions and 2 deletions
Download patch file Download diff file Expand all files Collapse all files

9
llama.cpp
View file

@ -11861,7 +11861,9 @@ static void llama_grammar_advance_stack(
std::vector<std::vector<const llama_grammar_element *>> & new_stacks) { std::vector<std::vector<const llama_grammar_element *>> & new_stacks) {
if (stack.empty()) { if (stack.empty()) {
new_stacks.emplace_back(stack); if (std::find(new_stacks.begin(), new_stacks.end(), stack) == new_stacks.end()) {
new_stacks.emplace_back(stack);
}
return; return;
} }
@ -11898,7 +11900,10 @@ static void llama_grammar_advance_stack(
} }
case LLAMA_GRETYPE_CHAR: case LLAMA_GRETYPE_CHAR:
case LLAMA_GRETYPE_CHAR_NOT: case LLAMA_GRETYPE_CHAR_NOT:
new_stacks.emplace_back(stack); if (std::find(new_stacks.begin(), new_stacks.end(), stack) == new_stacks.end()) {
// only add the stack if it's not a duplicate of one we already have
new_stacks.emplace_back(stack);
}
break; break;
default: default:
// end of alternate (LLAMA_GRETYPE_END, LLAMA_GRETYPE_ALT) or middle of char range // end of alternate (LLAMA_GRETYPE_END, LLAMA_GRETYPE_ALT) or middle of char range