CS348Project/llama.cpp
5
0
Fork 0
mirror of https://github.com/ggml-org/llama.cpp.git synced 2025-11-03 09:22:01 +00:00
Code Issues Packages Projects Releases Wiki Activity

simple : add parallel decoding support

Browse Source
This commit is contained in:
Georgi Gerganov
2023-09-20 13:06:34 +03:00
parent addae65fd4
commit b377bf2266
7 changed files with 187 additions and 76 deletions
Download Patch File Download Diff File Expand all files Collapse all files

8
examples/embd-input/embd-input-lib.cpp
View File

@@ -79,7 +79,7 @@ bool eval_float(void * model, float * input, int N){
if (n_eval > n_batch) {
n_eval = n_batch;
}
llama_batch batch = { uint32_t(n_eval), nullptr, (input+i*n_emb), nullptr, nullptr, nullptr, n_past, 1, 0, };
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)) {
fprintf(stderr, "%s : failed to eval\n", __func__);
return false;
@@ -183,11 +183,11 @@ llama_token sampling_id(struct MyModel* mymodel) {
if (mirostat == 1) {
static float mirostat_mu = 2.0f * mirostat_tau;
const int mirostat_m = 100;
llama_sample_temperature(ctx, &candidates_p, temp);
llama_sample_temp(ctx, &candidates_p, temp);
id = llama_sample_token_mirostat(ctx, &candidates_p, mirostat_tau, mirostat_eta, mirostat_m, &mirostat_mu);
} else if (mirostat == 2) {
static float mirostat_mu = 2.0f * mirostat_tau;
llama_sample_temperature(ctx, &candidates_p, temp);
llama_sample_temp(ctx, &candidates_p, temp);
id = llama_sample_token_mirostat_v2(ctx, &candidates_p, mirostat_tau, mirostat_eta, &mirostat_mu);
} else {
// Temperature sampling
@@ -195,7 +195,7 @@ llama_token sampling_id(struct MyModel* mymodel) {
llama_sample_tail_free(ctx, &candidates_p, tfs_z, 1);
llama_sample_typical(ctx, &candidates_p, typical_p, 1);
llama_sample_top_p(ctx, &candidates_p, top_p, 1);
llama_sample_temperature(ctx, &candidates_p, temp);
llama_sample_temp(ctx, &candidates_p, temp);
id = llama_sample_token(ctx, &candidates_p);
}
}