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

llama.cpp : split llama_context_params into model and context params (#3301)

Browse source 
* llama.cpp : split llama_context_params into model and context params

ggml-ci

* fix metal build

* fix freq_base/scale default to model value

* llama-bench : keep the same model between tests when possible

* move n_threads to llama_context_params, add n_threads_batch

* fix mpi build

* remove kv_size(), cuda scratch fixes

* remove low-vram option

* add n_threads_batch to system info, refactor to get_system_info()

* add documentation about --threads-batch to the READMEs

* llama-bench fix

* main : fix rope freq/scale warning

* llama.cpp : add llama_get_model
common : add llama_tokenize from model

* remove duplicated ctx/model functions

ggml-ci

* cuda : print total VRAM used
This commit is contained in:
slaren 2023-09-28 21:42:38 +02:00 • committed by GitHub
parent 0512d66670
commit 16bc66d947
No known key found for this signature in database
GPG key ID: 4AEE18F83AFDEB23
27 changed files with 713 additions and 633 deletions
Download patch file Download diff file Expand all files Collapse all files

13
examples/embd-input/embd-input-lib.cpp
View file

@ -48,8 +48,7 @@ struct MyModel* create_mymodel(int argc, char ** argv) {
// print system information
{
fprintf(stderr, "\n");
fprintf(stderr, "system_info: n_threads = %d / %d | %s\n",
params.n_threads, std::thread::hardware_concurrency(), llama_print_system_info());
fprintf(stderr, "%s\n", get_system_info(params).c_str());
}
struct MyModel * ret = new MyModel();
ret->ctx = ctx;
@ -71,7 +70,7 @@ bool eval_float(void * model, float * input, int N){
MyModel * mymodel = (MyModel*)model;
llama_context * ctx = mymodel->ctx;
gpt_params params = mymodel->params;
int n_emb = llama_n_embd(ctx);
int n_emb = llama_n_embd(llama_get_model(ctx));
int n_past = mymodel->n_past;
int n_batch = N; // params.n_batch;
@ -81,7 +80,7 @@ bool eval_float(void * model, float * input, int N){
n_eval = n_batch;
}
llama_batch batch = { int32_t(n_eval), nullptr, (input+i*n_emb), nullptr, nullptr, nullptr, n_past, 1, 0, };
if (llama_decode(ctx, batch, params.n_threads)) {
if (llama_decode(ctx, batch)) {
fprintf(stderr, "%s : failed to eval\n", __func__);
return false;
}
@ -102,7 +101,7 @@ bool eval_tokens(void * model, std::vector<llama_token> tokens) {
if (n_eval > params.n_batch) {
n_eval = params.n_batch;
}
if (llama_decode(ctx, llama_batch_get_one(&tokens[i], n_eval, n_past, 0), params.n_threads)) {
if (llama_decode(ctx, llama_batch_get_one(&tokens[i], n_eval, n_past, 0))) {
fprintf(stderr, "%s : failed to eval\n", __func__);
return false;
}
@ -133,7 +132,7 @@ llama_token sampling_id(struct MyModel* mymodel) {
// out of user input, sample next token
const float temp = params.temp;
const int32_t top_k = params.top_k <= 0 ? llama_n_vocab(ctx) : params.top_k;
const int32_t top_k = params.top_k <= 0 ? llama_n_vocab(llama_get_model(ctx)) : params.top_k;
const float top_p = params.top_p;
const float tfs_z = params.tfs_z;
const float typical_p = params.typical_p;
@ -149,7 +148,7 @@ llama_token sampling_id(struct MyModel* mymodel) {
llama_token id = 0;
{
auto logits = llama_get_logits(ctx);
auto n_vocab = llama_n_vocab(ctx);
auto n_vocab = llama_n_vocab(llama_get_model(ctx));
// Apply params.logit_bias map
for (auto it = params.logit_bias.begin(); it != params.logit_bias.end(); it++) {