2023-09-13 13:19:44 +00:00
|
|
|
#include "llama.h"
|
|
|
|
#include "common.h"
|
|
|
|
#include "console.h"
|
|
|
|
|
|
|
|
#include <cassert>
|
|
|
|
#include <cstdio>
|
|
|
|
#include <cstring>
|
|
|
|
#include <string>
|
|
|
|
#include <codecvt>
|
|
|
|
#include <map>
|
|
|
|
#include <vector>
|
|
|
|
#include <locale>
|
|
|
|
|
|
|
|
typedef int codepoint;
|
|
|
|
|
2023-09-15 19:38:27 +00:00
|
|
|
static std::string codepoint_to_utf8(codepoint cp) {
|
2023-09-13 13:19:44 +00:00
|
|
|
std::string result;
|
|
|
|
if (0x00 <= cp && cp <= 0x7f) {
|
|
|
|
result.push_back(cp);
|
|
|
|
} else if (0x80 <= cp && cp <= 0x7ff) {
|
|
|
|
result.push_back(0xc0 | ((cp >> 6) & 0x1f));
|
|
|
|
result.push_back(0x80 | (cp & 0x3f));
|
|
|
|
} else if (0x800 <= cp && cp <= 0xffff) {
|
|
|
|
result.push_back(0xe0 | ((cp >> 12) & 0x0f));
|
|
|
|
result.push_back(0x80 | ((cp >> 6) & 0x3f));
|
|
|
|
result.push_back(0x80 | (cp & 0x3f));
|
|
|
|
} else if (0x10000 <= cp && cp <= 0x10ffff) {
|
|
|
|
result.push_back(0xf0 | ((cp >> 18) & 0x07));
|
|
|
|
result.push_back(0x80 | ((cp >> 12) & 0x3f));
|
|
|
|
result.push_back(0x80 | ((cp >> 6) & 0x3f));
|
|
|
|
result.push_back(0x80 | (cp & 0x3f));
|
|
|
|
} else {
|
|
|
|
throw std::invalid_argument("invalid codepoint");
|
|
|
|
}
|
|
|
|
return result;
|
|
|
|
}
|
|
|
|
|
|
|
|
int main(int argc, char **argv) {
|
|
|
|
if (argc < 2) {
|
|
|
|
fprintf(stderr, "Usage: %s <vocab-file>\n", argv[0]);
|
|
|
|
return 1;
|
|
|
|
}
|
|
|
|
|
|
|
|
const std::string fname = argv[1];
|
|
|
|
|
|
|
|
fprintf(stderr, "%s : reading vocab from: '%s'\n", __func__, fname.c_str());
|
|
|
|
|
|
|
|
llama_model * model;
|
|
|
|
llama_context * ctx;
|
|
|
|
|
|
|
|
llama_backend_init(false);
|
|
|
|
|
|
|
|
// load the vocab
|
|
|
|
{
|
2023-09-28 19:42:38 +00:00
|
|
|
auto mparams = llama_model_default_params();
|
2023-09-13 13:19:44 +00:00
|
|
|
|
2023-09-28 19:42:38 +00:00
|
|
|
mparams.vocab_only = true;
|
2023-09-13 13:19:44 +00:00
|
|
|
|
2023-09-28 19:42:38 +00:00
|
|
|
model = llama_load_model_from_file(fname.c_str(), mparams);
|
2023-09-13 13:19:44 +00:00
|
|
|
|
|
|
|
if (model == NULL) {
|
|
|
|
fprintf(stderr, "%s: error: failed to load vocab '%s'\n", __func__, fname.c_str());
|
|
|
|
return 1;
|
|
|
|
}
|
|
|
|
|
2023-09-28 19:42:38 +00:00
|
|
|
auto cparams = llama_context_default_params();
|
|
|
|
|
|
|
|
ctx = llama_new_context_with_model(model, cparams);
|
2023-09-13 13:19:44 +00:00
|
|
|
|
|
|
|
if (ctx == NULL) {
|
|
|
|
fprintf(stderr, "%s: error: failed to load vocab '%s'\n", __func__, fname.c_str());
|
|
|
|
llama_free_model(model);
|
|
|
|
return 1;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2023-09-28 19:42:38 +00:00
|
|
|
GGML_ASSERT(llama_vocab_type(model) == LLAMA_VOCAB_TYPE_SPM);
|
2023-09-13 13:19:44 +00:00
|
|
|
|
|
|
|
#ifdef _WIN32
|
|
|
|
// We need this for unicode console support
|
|
|
|
console::init(false, false);
|
|
|
|
atexit([]() { console::cleanup(); });
|
|
|
|
#endif
|
|
|
|
|
2023-09-28 19:42:38 +00:00
|
|
|
const int n_vocab = llama_n_vocab(model);
|
2023-09-13 13:19:44 +00:00
|
|
|
|
|
|
|
for (int i = 0; i < n_vocab; ++i) {
|
|
|
|
std::string str = llama_detokenize_spm(ctx, std::vector<int>(1, i));
|
|
|
|
std::vector<llama_token> tokens = llama_tokenize(ctx, str, false);
|
|
|
|
std::string check = llama_detokenize_spm(ctx, tokens);
|
|
|
|
if (check != str) {
|
2023-09-16 11:41:33 +00:00
|
|
|
fprintf(stderr, "%s : error: token %d detokenizes to '%s'(%zu) but tokenization of this detokenizes to '%s'(%zu)\n",
|
2023-09-13 13:19:44 +00:00
|
|
|
__func__, i, str.c_str(), str.length(), check.c_str(), check.length());
|
2023-09-16 11:41:33 +00:00
|
|
|
return 2;
|
2023-09-13 13:19:44 +00:00
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
for (codepoint cp = 0x0000; cp < 0xffff; ++cp) {
|
|
|
|
if (cp < 0xd800 || cp > 0xdfff) {
|
|
|
|
std::string str = codepoint_to_utf8(cp);
|
|
|
|
std::vector<llama_token> tokens = llama_tokenize(ctx, str, false);
|
|
|
|
std::string check = llama_detokenize_spm(ctx, tokens);
|
2023-09-16 11:41:33 +00:00
|
|
|
if (cp != 9601 && str != check) {
|
|
|
|
fprintf(stderr, "%s : error: codepoint %d detokenizes to '%s'(%zu) instead of '%s'(%zu)\n",
|
2023-09-13 13:19:44 +00:00
|
|
|
__func__, cp, check.c_str(), check.length(), str.c_str(), str.length());
|
2023-09-16 11:41:33 +00:00
|
|
|
return 3;
|
2023-09-13 13:19:44 +00:00
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
for (codepoint cp = 0x10000; cp < 0x0010ffff; ++cp) {
|
|
|
|
std::string str = codepoint_to_utf8(cp);
|
|
|
|
std::vector<llama_token> tokens = llama_tokenize(ctx, str, false);
|
|
|
|
std::string check = llama_detokenize_spm(ctx, tokens);
|
|
|
|
if (str != check) {
|
2023-09-16 11:41:33 +00:00
|
|
|
fprintf(stderr, "%s : error: codepoint %d detokenizes to '%s'(%zu) instead of '%s'(%zu)\n",
|
2023-09-13 13:19:44 +00:00
|
|
|
__func__, cp, check.c_str(), check.length(), str.c_str(), str.length());
|
|
|
|
return 4;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
llama_free_model(model);
|
|
|
|
llama_free(ctx);
|
|
|
|
|
|
|
|
llama_backend_free();
|
|
|
|
|
|
|
|
return 0;
|
|
|
|
}
|