llama : optimize vocabulary loading (#30679)

* llama : optimize vocabulary loading

* Update src/unicode.h

Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>

* CR : added reserve in gguf_kv_to_str

* CR : trailing space fix

---------

Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>
This commit is contained in:
Evgen MykhailenkoandGeorgi Gerganov authored and GitHub committed 2026-10-21 12:03:27 +03:00
1 parent eb182d85c0
commit 824eaecf48
7 files changed
+75 -39

No files matched your search

+2 -2
View File
@@ -179,13 +179,13 @@ static void llama_adapter_lora_init_impl(llama_model & model, FILE * file, llama
? format("%s[%s,%zu]", gguf_type_name(type), gguf_type_name(gguf_get_arr_type(gguf_ctx, i)), gguf_get_arr_n(gguf_ctx, i))
: gguf_type_name(type);
const char * name = gguf_get_key(gguf_ctx, i);
const std::string value = gguf_kv_to_str(gguf_ctx, i);
const size_t MAX_VALUE_LEN = 40;
const std::string value = gguf_kv_to_str(gguf_ctx, i, MAX_VALUE_LEN);
if (type != GGUF_TYPE_ARRAY) {
adapter.gguf_kv.emplace(name, value);
}
const size_t MAX_VALUE_LEN = 40;
std::string print_value = value.size() > MAX_VALUE_LEN ? format("%s...", value.substr(0, MAX_VALUE_LEN - 3).c_str()) : value;
replace_all(print_value, "\n", "\\n");