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

common : remove duplicate function llama_should_add_bos_token (#8778)

Browse source
This commit is contained in:
Zhenwei Jin 2024-08-15 15:23:23 +08:00 • committed by GitHub
parent 6bda7ce6c3
commit 4af8420afb
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
14 changed files with 26 additions and 40 deletions
Download patch file Download diff file Expand all files Collapse all files

2
examples/cvector-generator/cvector-generator.cpp
View file

@ -271,7 +271,7 @@ struct tokenized_prompt {
size_t max_seq_len;
tokenized_prompt(llama_context * ctx, std::string pos, std::string neg) {
const bool add_bos = llama_should_add_bos_token(llama_get_model(ctx));
const bool add_bos = llama_add_bos_token(llama_get_model(ctx));
tokens_pos = ::llama_tokenize(ctx, pos, add_bos, true);
tokens_neg = ::llama_tokenize(ctx, neg, add_bos, true);
max_seq_len = std::max(tokens_pos.size(), tokens_neg.size());

2
examples/eval-callback/eval-callback.cpp
View file

@ -127,7 +127,7 @@ static bool ggml_debug(struct ggml_tensor * t, bool ask, void * user_data) {
}
static bool run(llama_context * ctx, const gpt_params & params) {
const bool add_bos = llama_should_add_bos_token(llama_get_model(ctx));
const bool add_bos = llama_add_bos_token(llama_get_model(ctx));
std::vector<llama_token> tokens = ::llama_tokenize(ctx, params.prompt, add_bos);

4
examples/imatrix/imatrix.cpp
View file

@ -433,8 +433,8 @@ static void process_logits(
}
static bool compute_imatrix(llama_context * ctx, const gpt_params & params) {
const bool add_bos = llama_should_add_bos_token(llama_get_model(ctx));
GGML_ASSERT(llama_add_eos_token(llama_get_model(ctx)) != 1);
const bool add_bos = llama_add_bos_token(llama_get_model(ctx));
GGML_ASSERT(!llama_add_eos_token(llama_get_model(ctx)));
const int n_ctx = llama_n_ctx(ctx);
auto tim1 = std::chrono::high_resolution_clock::now();

4
examples/infill/infill.cpp
View file

@ -203,8 +203,8 @@ int main(int argc, char ** argv) {
LOG_TEE("\n");
LOG_TEE("%s\n", gpt_params_get_system_info(params).c_str());
}
const bool add_bos = llama_should_add_bos_token(model);
GGML_ASSERT(llama_add_eos_token(model) != 1);
const bool add_bos = llama_add_bos_token(model);
GGML_ASSERT(!llama_add_eos_token(model));
LOG("add_bos: %d\n", add_bos);
std::vector<llama_token> embd_inp;

4
examples/main/main.cpp
View file

@ -267,9 +267,9 @@ int main(int argc, char ** argv) {
}
}
const bool add_bos = llama_should_add_bos_token(model);
const bool add_bos = llama_add_bos_token(model);
if (!llama_model_has_encoder(model)) {
GGML_ASSERT(llama_add_eos_token(model) != 1);
GGML_ASSERT(!llama_add_eos_token(model));
}
LOG("add_bos: %d\n", add_bos);

12
examples/perplexity/perplexity.cpp
View file

@ -340,8 +340,8 @@ static results_perplexity perplexity_v2(llama_context * ctx, const gpt_params &
// Output: `perplexity: 13.5106 [114/114]`
// BOS tokens will be added for each chunk before eval
const bool add_bos = llama_should_add_bos_token(llama_get_model(ctx));
GGML_ASSERT(llama_add_eos_token(llama_get_model(ctx)) != 1);
const bool add_bos = llama_add_bos_token(llama_get_model(ctx));
GGML_ASSERT(!llama_add_eos_token(llama_get_model(ctx)));
fprintf(stderr, "%s: tokenizing the input ..\n", __func__);
@ -480,8 +480,8 @@ static results_perplexity perplexity(llama_context * ctx, const gpt_params & par
// Output: `perplexity: 13.5106 [114/114]`
// BOS tokens will be added for each chunk before eval
const bool add_bos = llama_should_add_bos_token(llama_get_model(ctx));
GGML_ASSERT(llama_add_eos_token(llama_get_model(ctx)) != 1);
const bool add_bos = llama_add_bos_token(llama_get_model(ctx));
GGML_ASSERT(!llama_add_eos_token(llama_get_model(ctx)));
std::ofstream logits_stream;
if (!params.logits_file.empty()) {
@ -1733,8 +1733,8 @@ static void kl_divergence(llama_context * ctx, const gpt_params & params) {
const int n_batch = params.n_batch;
const int num_batches = (n_ctx + n_batch - 1)/n_batch;
const int nv = 2*((n_vocab + 1)/2) + 4;
const bool add_bos = llama_should_add_bos_token(llama_get_model(ctx));
GGML_ASSERT(llama_add_eos_token(llama_get_model(ctx)) != 1);
const bool add_bos = llama_add_bos_token(llama_get_model(ctx));
GGML_ASSERT(!llama_add_eos_token(llama_get_model(ctx)));
std::vector<uint16_t> log_probs_uint16(size_t(n_ctx - 1 - n_ctx/2) * nv);
std::vector<float> kld_values(size_t(n_ctx - 1 - n_ctx/2)*n_chunk);

7
examples/server/server.cpp
View file

@ -693,9 +693,8 @@ struct server_context {
n_ctx = llama_n_ctx(ctx);
add_bos_token = llama_should_add_bos_token(model);
has_eos_token = llama_add_eos_token(model) != 1;
add_bos_token = llama_add_bos_token(model);
has_eos_token = !llama_add_eos_token(model);
return true;
}
@ -2038,7 +2037,7 @@ struct server_context {
slot.t_start_generation = 0;
if (slot.infill) {
const bool add_bos = llama_should_add_bos_token(model);
const bool add_bos = llama_add_bos_token(model);
bool suff_rm_leading_spc = true;
if (params.input_suffix.find_first_of(' ') == 0 && params.input_suffix.size() > 1) {
params.input_suffix.erase(0, 1);

2
examples/tokenize/tokenize.cpp
View file

@ -362,7 +362,7 @@ int main(int raw_argc, char ** raw_argv) {
prompt = stdin_buffer.str();
}
const bool model_wants_add_bos = llama_should_add_bos_token(model);
const bool model_wants_add_bos = llama_add_bos_token(model);
const bool add_bos = model_wants_add_bos && !no_bos;
const bool parse_special = !no_parse_special;