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

Reenable tokenizer test for LLaMa

Browse source
This commit is contained in:
goerch 2023-09-09 12:21:06 +02:00
parent 9a953a42c4
commit 89a727774c
6 changed files with 133 additions and 255 deletions
Download patch file Download diff file Expand all files Collapse all files

7
llama.cpp
View file

@ -3121,10 +3121,9 @@ struct llm_tokenizer_spm {
while (offs < text.size()) { while (offs < text.size()) {
llm_symbol sym; llm_symbol sym;
size_t len = utf8_len(text[offs]); size_t len = utf8_len(text[offs]);
GGML_ASSERT(offs + len <= text.size());
sym.text = text.c_str() + offs; sym.text = text.c_str() + offs;
sym.n = len; sym.n = std::min(len, text.size() - offs);
offs += len; offs += sym.n;
sym.prev = index - 1; sym.prev = index - 1;
sym.next = offs == text.size() ? -1 : index + 1; sym.next = offs == text.size() ? -1 : index + 1;
index++; index++;
@ -6218,7 +6217,7 @@ int llama_tokenize_with_model(
auto res = llama_tokenize_internal(model->vocab, text, add_bos); auto res = llama_tokenize_internal(model->vocab, text, add_bos);
if (n_max_tokens < (int) res.size()) { if (n_max_tokens < (int) res.size()) {
LLAMA_LOG_ERROR("%s: too many tokens\n", __func__); // LLAMA_LOG_ERROR("%s: too many tokens\n", __func__);
return -((int) res.size()); return -((int) res.size());
} }

5
tests/CMakeLists.txt
View file

@ -29,9 +29,8 @@ llama_build_executable(test-tokenizer-0-llama.cpp)
llama_test_executable (test-tokenizer-0-llama test-tokenizer-0-llama.cpp ${CMAKE_CURRENT_SOURCE_DIR}/../models/ggml-vocab-llama.gguf) llama_test_executable (test-tokenizer-0-llama test-tokenizer-0-llama.cpp ${CMAKE_CURRENT_SOURCE_DIR}/../models/ggml-vocab-llama.gguf)
llama_build_executable(test-tokenizer-0-falcon.cpp) llama_build_executable(test-tokenizer-0-falcon.cpp)
#llama_test_executable (test-tokenizer-0-falcon test-tokenizer-0-falcon.cpp ${CMAKE_CURRENT_SOURCE_DIR}/../models/ggml-vocab-falcon.gguf) #llama_test_executable (test-tokenizer-0-falcon test-tokenizer-0-falcon.cpp ${CMAKE_CURRENT_SOURCE_DIR}/../models/ggml-vocab-falcon.gguf)
llama_build_executable(test-tokenizer-1.cpp) llama_build_executable(test-tokenizer-1-llama.cpp)
# test-tokenizer-1 requires a BPE vocab. re-enable when we have one. llama_test_executable (test-tokenizer-1-llama test-tokenizer-1-llama.cpp ${CMAKE_CURRENT_SOURCE_DIR}/../models/ggml-vocab-llama.gguf)
#llama_test_executable (test-tokenizer-1.llama test-tokenizer-1.cpp ${CMAKE_CURRENT_SOURCE_DIR}/../models/ggml-vocab-falcon.gguf)
#llama_test_executable(test-tokenizer-1.aquila test-tokenizer-1.cpp ${CMAKE_CURRENT_SOURCE_DIR}/../models/ggml-vocab-aquila.gguf) #llama_test_executable(test-tokenizer-1.aquila test-tokenizer-1.cpp ${CMAKE_CURRENT_SOURCE_DIR}/../models/ggml-vocab-aquila.gguf)
llama_build_and_test_executable(test-grammar-parser.cpp) llama_build_and_test_executable(test-grammar-parser.cpp)
llama_build_and_test_executable(test-llama-grammar.cpp) llama_build_and_test_executable(test-llama-grammar.cpp)

4
tests/test-tokenizer-0-llama.cpp
View file

@ -1,5 +1,6 @@
#include "llama.h" #include "llama.h"
#include "common.h" #include "common.h"
#include "console.h"
#include <cstdio> #include <cstdio>
#include <string> #include <string>
@ -89,6 +90,9 @@ int main(int argc, char **argv) {
return 2; return 2;
} }
console::init(false, false);
atexit([]() { console::cleanup(); });
bool success = true; bool success = true;
for (const auto & test_kv : k_tests()) { for (const auto & test_kv : k_tests()) {

140
tests/test-tokenizer-0.cpp
View file

@ -1,140 +0,0 @@
#include "llama.h"
#include "common.h"
#include <cstdio>
#include <string>
#include <map>
#include <vector>
static std::string unescape_whitespace(llama_context* ctx, const std::vector<llama_token>& tokens) {
std::string result;
for (size_t i = 0; i < tokens.size(); ++i) {
result += llama_token_to_str(ctx, tokens[i]);
}
return result;
}
static const std::map<std::string, std::vector<llama_token>> & k_tests() {
static std::map<std::string, std::vector<llama_token>> _k_tests = {
{ " ", {1, 259, }, },
{ " ", { 1, 1678, }, },
{ " ", { 1, 268, }, },
{ "\t", { 1, 29871, 12, }, },
{ "\n", { 1, 29871, 13, }, },
{ "\t\n", { 1, 29871, 12, 13, }, },
{ "Hello world", { 1, 15043, 3186, }, },
{ " Hello world", { 1, 29871, 15043, 3186, }, },
{ "Hello World", { 1, 15043, 2787, }, },
{ " Hello World", { 1, 29871, 15043, 2787, }, },
{ " Hello World!", { 1, 29871, 15043, 2787, 29991, }, },
{ " this is 🦙.cpp", { 1, 29871, 445, 338, 29871, 243, 162, 169, 156, 29889, 8223, }, },
{ "w048 7tuijk dsdfhu", { 1, 281, 29900, 29946, 29947, 29871, 29955, 9161, 13535, 18031, 2176, 6905, }, },
{ "нещо на Български", { 1, 1538, 4851, 665, 1386, 29713, 1305, }, },
{ "កាន់តែពិសេសអាចខលចេញ", { 1, 29871, 31849, 31324, 31934, 228, 162, 142, 228, 161,
146, 228, 162, 133, 228, 161, 153, 228, 161, 186,
31708, 228, 162, 132, 31708, 228, 161, 165, 31324, 228,
161, 136, 228, 161, 132, 228, 161, 158, 228, 161,
136, 228, 162, 132, 228, 161, 140, }, },
{ "🚀 (normal) 😶‍🌫️ (multiple emojis concatenated) ✅ (only emoji that has its own token)",
{ 1, 29871, 243, 162, 157, 131, 313, 8945, 29897, 29871,
243, 162, 155, 185, 30722, 243, 162, 143, 174, 30598,
313, 20787, 953, 3848, 275, 16125, 630, 29897, 29871, 31681,
313, 6194, 953, 29877, 2397, 393, 756, 967, 1914, 5993, 29897, }, },
{ "Hello", { 1, 15043 }, },
{ " Hello", { 1, 29871, 15043 }, },
{ " Hello", { 1, 259, 15043 }, },
{ " Hello", { 1, 1678, 15043 }, },
{ " Hello", { 1, 268, 15043 }, },
{ " Hello\n Hello", { 1, 268, 15043, 13, 1678, 15043 }, },
};
return _k_tests;
}
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
{
auto lparams = llama_context_default_params();
lparams.vocab_only = true;
model = llama_load_model_from_file(fname.c_str(), lparams);
if (model == NULL) {
fprintf(stderr, "%s: error: failed to load vocab '%s'\n", __func__, fname.c_str());
return 1;
}
ctx = llama_new_context_with_model(model, lparams);
if (ctx == NULL) {
fprintf(stderr, "%s: error: failed to load vocab '%s'\n", __func__, fname.c_str());
llama_free_model(model);
return 1;
}
}
const int n_vocab = llama_n_vocab(ctx);
if (n_vocab != 32000) {
fprintf(stderr, "%s : expected 32000 tokens, got %d\n", __func__, n_vocab);
llama_free_model(model);
llama_free(ctx);
return 2;
}
bool success = true;
for (const auto & test_kv : k_tests()) {
std::vector<llama_token> res = llama_tokenize(ctx, test_kv.first, true);
fprintf(stderr, "%s : '%s' tokenized to '%s'\n",
__func__, test_kv.first.c_str(), unescape_whitespace(ctx, res).c_str());
bool correct = res.size() == test_kv.second.size();
for (int i = 0; i < (int) res.size() && correct; ++i) {
if (res[i] != test_kv.second[i]) {
correct = false;
}
}
if (!correct) {
fprintf(stderr, "%s : failed test: '%s'\n", __func__, test_kv.first.c_str());
fprintf(stderr, "%s : detokenized to: '%s' instead of '%s'\n", __func__,
unescape_whitespace(ctx, res).c_str(), unescape_whitespace(ctx, test_kv.second).c_str());
fprintf(stderr, "%s : expected tokens: ", __func__);
for (const auto & t : test_kv.second) {
fprintf(stderr, "%6d, ", t);
}
fprintf(stderr, "\n");
fprintf(stderr, "%s : got tokens: ", __func__);
for (const auto & t : res) {
fprintf(stderr, "%6d, ", t);
}
fprintf(stderr, "\n");
success = false;
}
}
llama_free_model(model);
llama_free(ctx);
llama_backend_free();
return success ? 0 : 3;
}

124
tests/test-tokenizer-1-llama.cpp Normal file
View file

@ -0,0 +1,124 @@
#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;
std::string codepoint_to_utf8(codepoint cp) {
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
{
auto lparams = llama_context_default_params();
lparams.vocab_only = true;
model = llama_load_model_from_file(fname.c_str(), lparams);
if (model == NULL) {
fprintf(stderr, "%s: error: failed to load vocab '%s'\n", __func__, fname.c_str());
return 1;
}
ctx = llama_new_context_with_model(model, lparams);
if (ctx == NULL) {
fprintf(stderr, "%s: error: failed to load vocab '%s'\n", __func__, fname.c_str());
llama_free_model(model);
return 1;
}
}
GGML_ASSERT(llama_vocab_type(ctx) == LLAMA_VOCAB_TYPE_SPM);
console::init(false, false);
atexit([]() { console::cleanup(); });
const int n_vocab = llama_n_vocab(ctx);
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) {
fprintf(stderr, "%s : error: token %d detokenizes to >%s<(%d) but tokenization of this detokenizes to >%s<(%d)\n",
__func__, i, str.c_str(), str.length(), check.c_str(), check.length());
if(i != 3)
return 2;
}
}
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);
if (str != check) {
fprintf(stderr, "%s : error: codepoint %d detokenizes to >%s<(%d) instead of >%s<(%d)\n",
__func__, cp, check.c_str(), check.length(), str.c_str(), str.length());
if(cp != 0 && cp != 9601)
return 3;
}
}
}
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) {
fprintf(stderr, "%s : error: codepoint %d detokenizes to >%s<(%d) instead of >%s<(%d)\n",
__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;
}

108
tests/test-tokenizer-1.cpp
View file

@ -1,108 +0,0 @@
#include "llama.h"
#include "common.h"
#include <cassert>
#include <cstdio>
#include <cstring>
#include <string>
#include <codecvt>
#include <map>
#include <vector>
#include <locale>
static std::string escape_whitespace(const std::string& text) {
std::string result = "\xe2\x96\x81";
for (size_t offs = 0; offs < text.length(); ++offs) {
if (text[offs] == ' ') {
result += "\xe2\x96\x81";
} else {
result += text[offs];
}
}
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
{
auto lparams = llama_context_default_params();
lparams.vocab_only = true;
model = llama_load_model_from_file(fname.c_str(), lparams);
if (model == NULL) {
fprintf(stderr, "%s: error: failed to load vocab '%s'\n", __func__, fname.c_str());
return 1;
}
ctx = llama_new_context_with_model(model, lparams);
if (ctx == NULL) {
fprintf(stderr, "%s: error: failed to load vocab '%s'\n", __func__, fname.c_str());
llama_free_model(model);
return 1;
}
}
GGML_ASSERT(llama_vocab_type(ctx) == LLAMA_VOCAB_TYPE_BPE);
const int n_vocab = llama_n_vocab(ctx);
for (int i = 0; i < n_vocab; ++i) {
std::string forward = llama_token_to_piece(ctx, i);
std::vector<llama_token> tokens = llama_tokenize(ctx, forward, false);
if (tokens.size() == 1) {
if (i != tokens[0]) {
std::string backward = llama_token_to_piece(ctx, tokens[0]);
fprintf(stderr, "%s : error: token %d is string %s but bpe returns token %d %s\n",
__func__, i, llama_token_to_piece(ctx, i).c_str(), tokens[0], backward.c_str());
return 2;
}
}
}
#ifdef _WIN32
std::wstring_convert<typename std::codecvt_utf8<char16_t>, char16_t> u16converter;
for (char16_t ch = 0x0000; ch < 0xffff; ++ch) {
std::u16string u16str(1, ch);
std::string str = u16converter.to_bytes(u16str);
std::vector<llama_token> tokens = llama_tokenize(ctx, escape_whitespace(str).c_str(), false);
if (tokens.size() == 1) {
fprintf(stderr, "%s : info: %s tokenized to %d \n",
__func__, str.c_str(), tokens[0]);
}
}
std::wstring_convert<typename std::codecvt_utf8<char32_t>, char32_t> u32converter;
for (char32_t ch = 0x0000; ch < 0x0010ffff; ++ch) {
std::u32string u32str(1, ch);
std::string str = u32converter.to_bytes(u32str);
std::vector<llama_token> tokens = llama_tokenize(ctx, escape_whitespace(str).c_str(), false);
if (tokens.size() == 1) {
fprintf(stderr, "%s : info: %s tokenized to %d \n", __func__, str.c_str(), tokens[0]);
}
}
#endif
llama_free_model(model);
llama_free(ctx);
llama_backend_free();
return 0;
}