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

main : wip logs

Browse source
This commit is contained in:
Georgi Gerganov 2023-08-29 14:23:39 +03:00
parent 5c978f48a4
commit c72d344c1a
No known key found for this signature in database
GPG key ID: 449E073F9DC10735
1 changed files with 29 additions and 28 deletions
Download patch file Download diff file Expand all files Collapse all files

57
examples/main/main.cpp
View file

@ -115,12 +115,12 @@ int main(int argc, char ** argv) {
#ifndef LOG_DISABLE_LOGS
LOG_SET_TARGET(LOG_FILENAME_GENERATOR("main", "log"));
LOG_TEE("Log start\n")
LOG_TEE("Log start\n");
LOG_DUMP_CMDLINE(argc,argv);
#endif // LOG_DISABLE_LOGS
// TODO: Dump params ?
//LOG("Params perplexity: %s\n", LOG_TOSTR(params.perplexity))
//LOG("Params perplexity: %s\n", LOG_TOSTR(params.perplexity));
// save choice to use color for later
// (note for later: this is a slightly awkward choice)
@ -172,7 +172,7 @@ int main(int argc, char ** argv) {
params.prompt = gpt_random_prompt(rng);
}
LOG("llama backend init\n")
LOG("%s: llama backend init\n", __func__);
llama_backend_init(params.numa);
llama_model * model;
@ -182,7 +182,7 @@ int main(int argc, char ** argv) {
g_ctx = &ctx;
// load the model and apply lora adapter, if any
LOG("load the model and apply lora adapter, if any\n")
LOG("%s: load the model and apply lora adapter, if any\n", __func__);
std::tie(model, ctx) = llama_init_from_gpt_params(params);
if (params.cfg_scale > 1.f) {
struct llama_context_params lparams = llama_context_params_from_gpt_params(params);
@ -253,27 +253,26 @@ int main(int argc, char ** argv) {
}
}
// Add BOS if SPM tokenizer
const bool add_bos = llama_vocab_type(ctx) == LLAMA_VOCAB_TYPE_SPM;
LOG("add_bos: %d\n", add_bos);
// tokenize the prompt
LOG("Tokenize the prompt\n")
std::vector<llama_token> embd_inp;
if (params.interactive_first || params.instruct || !params.prompt.empty() || session_tokens.empty()) {
LOG("tokenize the prompt\n");
embd_inp = ::llama_tokenize(ctx, params.prompt, add_bos);
} else {
LOG("use session tokens\n");
embd_inp = session_tokens;
}
LOG("prompt: \"%s\"\n", LOG_TOSTR(params.prompt))
//LOG("Prompt tokenized: %s\n", LOG_TOSTR(embd_inp))
LOG("embd_inp tokenized: %s\n", LOG_TOKENS_TOSTR_PRETTY(embd_inp,ctx))
LOG("prompt: \"%s\"\n", LOG_TOSTR(params.prompt));
LOG("tokens: %s\n", LOG_TOKENS_TOSTR_PRETTY(ctx, embd_inp));
// Should not run without any tokens
if (embd_inp.empty()) {
embd_inp.push_back(llama_token_bos(ctx));
LOG("embd_inp was considered empty and bos was added: %s\n", LOG_TOKENS_TOSTR_PRETTY(embd_inp,ctx))
LOG("embd_inp was considered empty and bos was added: %s\n", LOG_TOKENS_TOSTR_PRETTY(ctx, embd_inp));
}
// Tokenize negative prompt
@ -281,22 +280,22 @@ int main(int argc, char ** argv) {
int guidance_offset = 0;
int original_prompt_len = 0;
if (ctx_guidance) {
LOG("cfg_negative_prompt: \"%s\"\n", LOG_TOSTR(params.cfg_negative_prompt))
LOG("cfg_negative_prompt: \"%s\"\n", LOG_TOSTR(params.cfg_negative_prompt));
guidance_inp = ::llama_tokenize(ctx_guidance, params.cfg_negative_prompt, add_bos);
LOG("guidance_inp tokenized: %s\n", LOG_TOKENS_TOSTR_PRETTY(guidance_inp,ctx_guidance))
LOG("guidance_inp tokenized: %s\n", LOG_TOKENS_TOSTR_PRETTY(ctx_guidance, guidance_inp));
std::vector<llama_token> original_inp = ::llama_tokenize(ctx, params.prompt, add_bos);
LOG("original_inp tokenized: %s\n", LOG_TOKENS_TOSTR_PRETTY(original_inp,ctx))
LOG("original_inp tokenized: %s\n", LOG_TOKENS_TOSTR_PRETTY(ctx, original_inp));
original_prompt_len = original_inp.size();
guidance_offset = (int)guidance_inp.size() - original_prompt_len;
LOG("original_prompt_len: %s", LOG_TOSTR(original_prompt_len))
LOG("guidance_offset: %s", LOG_TOSTR(guidance_offset))
LOG("original_prompt_len: %s", LOG_TOSTR(original_prompt_len));
LOG("guidance_offset: %s", LOG_TOSTR(guidance_offset));
}
const int n_ctx = llama_n_ctx(ctx);
LOG("n_ctx: %d\n", n_ctx)
LOG("n_ctx: %d\n", n_ctx);
if ((int) embd_inp.size() > n_ctx - 4) {
LOG_TEE("%s: error: prompt is too long (%d tokens, max %d)\n", __func__, (int) embd_inp.size(), n_ctx - 4);
@ -305,7 +304,7 @@ int main(int argc, char ** argv) {
// debug message about similarity of saved session, if applicable
size_t n_matching_session_tokens = 0;
if (session_tokens.size()) {
if (session_tokens.size() > 0) {
for (llama_token id : session_tokens) {
if (n_matching_session_tokens >= embd_inp.size() || id != embd_inp[n_matching_session_tokens]) {
break;
@ -327,14 +326,12 @@ int main(int argc, char ** argv) {
LOGLN(
"recalculate the cached logits (check): embd_inp.empty() %s, n_matching_session_tokens %zu, embd_inp.size() %zu, session_tokens.size() %zu, embd_inp.size() %zu",
LOG_TOSTR(embd_inp.empty()), n_matching_session_tokens, embd_inp.size(), session_tokens.size(), embd_inp.size())
LOG_TOSTR(embd_inp.empty()), n_matching_session_tokens, embd_inp.size(), session_tokens.size(), embd_inp.size());
// if we will use the cache for the full prompt without reaching the end of the cache, force
// reevaluation of the last token token to recalculate the cached logits
if (!embd_inp.empty() && n_matching_session_tokens == embd_inp.size() &&
session_tokens.size() > embd_inp.size()) {
LOGLN("recalculate the cached logits (do): session_tokens.resize( %zu )", embd_inp.size() - 1)
if (!embd_inp.empty() && n_matching_session_tokens == embd_inp.size() && session_tokens.size() > embd_inp.size()) {
LOGLN("recalculate the cached logits (do): session_tokens.resize( %zu )", embd_inp.size() - 1);
session_tokens.resize(embd_inp.size() - 1);
}
@ -348,6 +345,9 @@ int main(int argc, char ** argv) {
const auto inp_pfx = ::llama_tokenize(ctx, "\n\n### Instruction:\n\n", add_bos);
const auto inp_sfx = ::llama_tokenize(ctx, "\n\n### Response:\n\n", false);
LOG("inp_pfx: %s\n", LOG_TOKENS_TOSTR_PRETTY(ctx, inp_pfx));
LOG("inp_sfx: %s\n", LOG_TOKENS_TOSTR_PRETTY(ctx, inp_sfx));
// in instruct mode, we inject a prefix and a suffix to each input by the user
if (params.instruct) {
params.interactive_first = true;
@ -403,7 +403,7 @@ int main(int argc, char ** argv) {
LOG_TEE("%s: interactive mode on.\n", __func__);
if (params.antiprompt.size()) {
for (auto antiprompt : params.antiprompt) {
for (const auto & antiprompt : params.antiprompt) {
LOG_TEE("Reverse prompt: '%s'\n", antiprompt.c_str());
}
}
@ -463,11 +463,11 @@ int main(int argc, char ** argv) {
" - To return control without starting a new line, end your input with '/'.\n"
" - If you want to submit another line, end your input with '\\'.\n";
}
fprintf(stderr, "== Running in interactive mode. ==\n"
LOG_TEE("== Running in interactive mode. ==\n");
#if defined (__unix__) || (defined (__APPLE__) && defined (__MACH__)) || defined (_WIN32)
" - Press Ctrl+C to interject at any time.\n"
LOG_TEE( " - Press Ctrl+C to interject at any time.\n");
#endif
"%s\n", control_message);
LOG_TEE( "%s\n", control_message);
is_interacting = params.interactive_first;
}
@ -492,8 +492,9 @@ int main(int argc, char ** argv) {
std::vector<llama_token> embd;
std::vector<llama_token> embd_guidance;
// do one empty run to warm up the model
{
LOG("warming up the model with an empty run\n");
const std::vector<llama_token> tmp = { llama_token_bos(ctx), };
llama_eval(ctx, tmp.data(), tmp.size(), 0, params.n_threads);
llama_reset_timings(ctx);