llama : move vocab, grammar and sampling into separate files (#8508)

* llama : move sampling code into llama-sampling

ggml-ci

* llama : move grammar code into llama-grammar

ggml-ci

* cont

ggml-ci

* cont : pre-fetch rules

* cont

ggml-ci

* llama : deprecate llama_sample_grammar

* llama : move tokenizers into llama-vocab

ggml-ci

* make : update llama.cpp deps [no ci]

* llama : redirect external API to internal APIs

ggml-ci

* llama : suffix the internal APIs with "_impl"

ggml-ci

* llama : clean-up
This commit is contained in:
Georgi Gerganov 2024-07-23 13:10:17 +03:00 committed by GitHub
parent 751fcfc6c3
commit 938943cdbf
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
18 changed files with 3521 additions and 2968 deletions

View file

@ -16,20 +16,25 @@ static bool llama_sample_grammar_string(struct llama_grammar * grammar, const st
auto decoded = decode_utf8(input_str, {});
const auto & code_points = decoded.first;
const llama_grammar_rules & rules = llama_grammar_get_rules (grammar);
llama_grammar_stacks & cur_stacks = llama_grammar_get_stacks(grammar);
size_t pos = 0;
for (auto it = code_points.begin(), end = code_points.end() - 1; it != end; ++it) {
auto prev_stacks = grammar->stacks;
llama_grammar_accept(grammar->rules, prev_stacks, *it, grammar->stacks);
if (grammar->stacks.empty()) {
const llama_grammar_stacks prev_stacks = llama_grammar_get_stacks(grammar); // copy
llama_grammar_accept(rules, prev_stacks, *it, cur_stacks);
if (cur_stacks.empty()) {
error_pos = pos;
error_msg = "Unexpected character '" + unicode_cpt_to_utf8(*it) + "'";
grammar->stacks = prev_stacks;
cur_stacks = prev_stacks;
return false;
}
++pos;
}
for (const auto & stack : grammar->stacks) {
for (const auto & stack : cur_stacks) {
if (stack.empty()) {
return true;
}