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

common : reimplement logging (#9418)

Browse source 
https://github.com/ggerganov/llama.cpp/pull/9418
This commit is contained in:
Georgi Gerganov 2024-09-15 20:46:12 +03:00 • committed by GitHub
parent e6deac31f7
commit 6262d13e0b
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
54 changed files with 2092 additions and 2419 deletions
Download patch file Download diff file Expand all files Collapse all files

43
examples/simple/simple.cpp
View file

@ -1,16 +1,14 @@
#include "arg.h"
#include "common.h"
#include "log.h"
#include "llama.h"
#include <cmath>
#include <cstdio>
#include <string>
#include <vector>
static void print_usage(int, char ** argv) {
LOG_TEE("\nexample usage:\n");
LOG_TEE("\n %s -m model.gguf -p \"Hello my name is\" -n 32\n", argv[0]);
LOG_TEE("\n");
LOG("\nexample usage:\n");
LOG("\n %s -m model.gguf -p \"Hello my name is\" -n 32\n", argv[0]);
LOG("\n");
}
int main(int argc, char ** argv) {
@ -23,6 +21,8 @@ int main(int argc, char ** argv) {
return 1;
}
gpt_init();
// total length of the sequence including the prompt
const int n_predict = params.n_predict;
@ -69,25 +69,24 @@ int main(int argc, char ** argv) {
const int n_ctx = llama_n_ctx(ctx);
const int n_kv_req = tokens_list.size() + (n_predict - tokens_list.size());
LOG_TEE("\n%s: n_predict = %d, n_ctx = %d, n_kv_req = %d\n", __func__, n_predict, n_ctx, n_kv_req);
LOG("\n");
LOG_INF("%s: n_predict = %d, n_ctx = %d, n_kv_req = %d\n", __func__, n_predict, n_ctx, n_kv_req);
// make sure the KV cache is big enough to hold all the prompt and generated tokens
if (n_kv_req > n_ctx) {
LOG_TEE("%s: error: n_kv_req > n_ctx, the required KV cache size is not big enough\n", __func__);
LOG_TEE("%s: either reduce n_predict or increase n_ctx\n", __func__);
LOG_ERR("%s: error: n_kv_req > n_ctx, the required KV cache size is not big enough\n", __func__);
LOG_ERR("%s: either reduce n_predict or increase n_ctx\n", __func__);
return 1;
}
// print the prompt token-by-token
fprintf(stderr, "\n");
LOG("\n");
for (auto id : tokens_list) {
fprintf(stderr, "%s", llama_token_to_piece(ctx, id).c_str());
LOG("%s", llama_token_to_piece(ctx, id).c_str());
}
fflush(stderr);
// create a llama_batch with size 512
// we use this object to submit token data for decoding
@ -102,7 +101,7 @@ int main(int argc, char ** argv) {
batch.logits[batch.n_tokens - 1] = true;
if (llama_decode(ctx, batch) != 0) {
LOG_TEE("%s: llama_decode() failed\n", __func__);
LOG("%s: llama_decode() failed\n", __func__);
return 1;
}
@ -116,16 +115,16 @@ int main(int argc, char ** argv) {
while (n_cur <= n_predict) {
// sample the next token
{
const llama_token new_token_id = llama_sampler_sample(smpl, ctx, batch.n_tokens - 1);
const llama_token new_token_id = llama_sampler_sample(smpl, ctx, -1);
// is it an end of generation?
if (llama_token_is_eog(model, new_token_id) || n_cur == n_predict) {
LOG_TEE("\n");
LOG("\n");
break;
}
LOG_TEE("%s", llama_token_to_piece(ctx, new_token_id).c_str());
LOG("%s", llama_token_to_piece(ctx, new_token_id).c_str());
fflush(stdout);
// prepare the next batch
@ -141,23 +140,23 @@ int main(int argc, char ** argv) {
// evaluate the current batch with the transformer model
if (llama_decode(ctx, batch)) {
fprintf(stderr, "%s : failed to eval, return code %d\n", __func__, 1);
LOG_ERR("%s : failed to eval, return code %d\n", __func__, 1);
return 1;
}
}
LOG_TEE("\n");
LOG("\n");
const auto t_main_end = ggml_time_us();
LOG_TEE("%s: decoded %d tokens in %.2f s, speed: %.2f t/s\n",
LOG_INF("%s: decoded %d tokens in %.2f s, speed: %.2f t/s\n",
__func__, n_decode, (t_main_end - t_main_start) / 1000000.0f, n_decode / ((t_main_end - t_main_start) / 1000000.0f));
LOG_TEE("\n");
LOG("\n");
llama_perf_sampler_print(smpl);
llama_perf_context_print(ctx);
fprintf(stderr, "\n");
LOG("\n");
llama_batch_free(batch);
llama_sampler_free(smpl);