2023-10-03 07:16:26 +00:00
|
|
|
#include "llama.h"
|
|
|
|
#include "common.h"
|
|
|
|
#include "unicode.h"
|
|
|
|
#include "console.h"
|
|
|
|
|
|
|
|
#include <cassert>
|
2024-02-13 13:14:22 +00:00
|
|
|
#include <codecvt>
|
2023-10-03 07:16:26 +00:00
|
|
|
#include <cstdio>
|
|
|
|
#include <cstring>
|
2024-02-13 13:14:22 +00:00
|
|
|
#include <locale>
|
2023-10-03 07:16:26 +00:00
|
|
|
#include <string>
|
2024-02-13 13:14:22 +00:00
|
|
|
#include <thread>
|
2023-10-03 07:16:26 +00:00
|
|
|
#include <vector>
|
2024-07-05 17:01:35 +00:00
|
|
|
#include <atomic>
|
2023-10-03 07:16:26 +00:00
|
|
|
|
|
|
|
int main(int argc, char **argv) {
|
2024-05-11 08:12:06 +00:00
|
|
|
if (argc < 2 || argc > 3) {
|
|
|
|
fprintf(stderr, "Usage: %s <vocab-file> [--ignore-merges]\n", argv[0]);
|
2023-10-03 07:16:26 +00:00
|
|
|
return 1;
|
|
|
|
}
|
|
|
|
|
|
|
|
const std::string fname = argv[1];
|
2024-05-11 08:12:06 +00:00
|
|
|
bool ignore_merges = false;
|
|
|
|
if (argc == 3) {
|
|
|
|
if (std::strcmp(argv[2], "--ignore-merges") != 0) {
|
|
|
|
fprintf(stderr, "Usage: %s <vocab-file> [--ignore-merges]\n", argv[0]);
|
|
|
|
return 1;
|
|
|
|
}
|
|
|
|
ignore_merges = true;
|
|
|
|
}
|
2023-10-03 07:16:26 +00:00
|
|
|
|
|
|
|
fprintf(stderr, "%s : reading vocab from: '%s'\n", __func__, fname.c_str());
|
|
|
|
|
2024-05-11 08:12:06 +00:00
|
|
|
if (ignore_merges) {
|
|
|
|
fprintf(stderr, "%s : ignoring merges for tokens inside vocab\n", __func__);
|
|
|
|
}
|
|
|
|
|
2023-10-03 07:16:26 +00:00
|
|
|
llama_model * model;
|
|
|
|
llama_context * ctx;
|
|
|
|
|
2024-02-16 09:31:07 +00:00
|
|
|
llama_backend_init();
|
2023-10-03 07:16:26 +00:00
|
|
|
|
|
|
|
// load the vocab
|
|
|
|
{
|
|
|
|
auto mparams = llama_model_default_params();
|
|
|
|
|
|
|
|
mparams.vocab_only = true;
|
|
|
|
|
|
|
|
model = llama_load_model_from_file(fname.c_str(), mparams);
|
|
|
|
|
|
|
|
if (model == NULL) {
|
|
|
|
fprintf(stderr, "%s: error: failed to load vocab '%s'\n", __func__, fname.c_str());
|
|
|
|
return 1;
|
|
|
|
}
|
|
|
|
|
|
|
|
auto cparams = llama_context_default_params();
|
|
|
|
|
|
|
|
ctx = llama_new_context_with_model(model, cparams);
|
|
|
|
|
|
|
|
if (ctx == NULL) {
|
|
|
|
fprintf(stderr, "%s: error: failed to load vocab '%s'\n", __func__, fname.c_str());
|
|
|
|
llama_free_model(model);
|
|
|
|
return 1;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2024-07-05 17:01:35 +00:00
|
|
|
//GGML_ASSERT(llama_vocab_type(model) == LLAMA_VOCAB_TYPE_BPE);
|
|
|
|
if (llama_vocab_type(model) != LLAMA_VOCAB_TYPE_BPE) {
|
|
|
|
return 99;
|
|
|
|
}
|
2023-10-03 07:16:26 +00:00
|
|
|
|
|
|
|
#ifdef _WIN32
|
|
|
|
// We need this for unicode console support
|
|
|
|
console::init(false, false);
|
|
|
|
atexit([]() { console::cleanup(); });
|
|
|
|
#endif
|
|
|
|
|
|
|
|
const int n_vocab = llama_n_vocab(model);
|
|
|
|
|
|
|
|
for (int i = 0; i < n_vocab; ++i) {
|
2024-07-05 17:01:35 +00:00
|
|
|
std::string str = llama_detokenize(ctx, std::vector<int>(1, i));
|
2023-10-03 07:16:26 +00:00
|
|
|
try {
|
2024-03-11 15:47:47 +00:00
|
|
|
auto cps = unicode_cpts_from_utf8(str);
|
2024-05-11 08:12:06 +00:00
|
|
|
std::vector<llama_token> tokens = llama_tokenize(ctx, str, false, true);
|
|
|
|
if (ignore_merges && tokens.size() > 1) {
|
|
|
|
fprintf(stderr,
|
|
|
|
"%s : error: token %d detokenizes to '%s'(%zu) but "
|
|
|
|
"tokenization of this to multiple tokens: [",
|
|
|
|
__func__, i, str.c_str(), str.length());
|
|
|
|
fprintf(stderr, "%d", tokens[0]);
|
|
|
|
for (size_t i = 1; i < tokens.size(); i++) {
|
|
|
|
fprintf(stderr, ", %d", tokens[i]);
|
|
|
|
}
|
|
|
|
fprintf(stderr, "]\n");
|
|
|
|
return 2;
|
|
|
|
}
|
2024-07-05 17:01:35 +00:00
|
|
|
std::string check = llama_detokenize(ctx, tokens);
|
2023-10-03 07:16:26 +00:00
|
|
|
if (check != str) {
|
|
|
|
fprintf(stderr, "%s : error: token %d detokenizes to '%s'(%zu) but tokenization of this detokenizes to '%s'(%zu)\n",
|
|
|
|
__func__, i, str.c_str(), str.length(), check.c_str(), check.length());
|
|
|
|
return 2;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
catch (const std::invalid_argument &) {
|
2024-02-13 13:14:22 +00:00
|
|
|
//fprintf(stderr, "%s : info: utf8 conversion %d '%s'\n", __func__, i, str.c_str());
|
2023-10-03 07:16:26 +00:00
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2024-02-13 13:14:22 +00:00
|
|
|
// unicode
|
|
|
|
{
|
|
|
|
const int nthread = std::thread::hardware_concurrency();
|
|
|
|
|
|
|
|
std::vector<std::thread> threads(nthread);
|
|
|
|
|
2024-07-05 17:01:35 +00:00
|
|
|
std::atomic_int errcode = {};
|
|
|
|
|
2024-02-13 13:14:22 +00:00
|
|
|
for (int i = 0; i < nthread; ++i) {
|
2024-07-05 17:01:35 +00:00
|
|
|
threads[i] = std::thread([i, nthread, ctx, &errcode]() {
|
|
|
|
for (uint32_t cp = i; !errcode && cp < 0x00110000; cp += nthread) {
|
|
|
|
if ((0x0000D800 <= cp && cp <= 0x0000DFFF) || // surrogates \p{Cs}
|
|
|
|
(0x00040000 <= cp && cp <= 0x000E0000)) { // undefined \p{Cn}
|
2024-02-13 13:14:22 +00:00
|
|
|
continue;
|
|
|
|
}
|
|
|
|
|
2024-03-11 15:47:47 +00:00
|
|
|
std::string str = unicode_cpt_to_utf8(cp);
|
2024-02-13 13:14:22 +00:00
|
|
|
std::vector<llama_token> tokens = llama_tokenize(ctx, str, false);
|
2024-07-05 17:01:35 +00:00
|
|
|
std::string check = llama_detokenize(ctx, tokens);
|
2024-02-13 13:14:22 +00:00
|
|
|
if (cp != 9601 && str != check) {
|
2024-07-05 17:01:35 +00:00
|
|
|
fprintf(stderr, "error: codepoint 0x%x detokenizes to '%s'(%zu) instead of '%s'(%zu)\n",
|
2024-02-13 13:14:22 +00:00
|
|
|
cp, check.c_str(), check.length(), str.c_str(), str.length());
|
2024-07-05 17:01:35 +00:00
|
|
|
errcode = 3;
|
2024-02-13 13:14:22 +00:00
|
|
|
}
|
|
|
|
}
|
|
|
|
});
|
2023-10-24 07:17:17 +00:00
|
|
|
}
|
2024-02-13 13:14:22 +00:00
|
|
|
|
|
|
|
for (auto & t : threads) {
|
|
|
|
t.join();
|
2023-10-03 07:16:26 +00:00
|
|
|
}
|
2024-07-05 17:01:35 +00:00
|
|
|
|
|
|
|
if (errcode) {
|
|
|
|
return errcode;
|
|
|
|
}
|
2023-10-03 07:16:26 +00:00
|
|
|
}
|
2024-02-13 13:14:22 +00:00
|
|
|
|
2023-10-03 07:16:26 +00:00
|
|
|
llama_free_model(model);
|
|
|
|
llama_free(ctx);
|
|
|
|
|
|
|
|
llama_backend_free();
|
|
|
|
|
|
|
|
return 0;
|
|
|
|
}
|