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

common : add llama_batch_add() and llama_batch_clear() helpers

Browse source
This commit is contained in:
Georgi Gerganov 2023-10-16 12:41:33 +03:00
parent 005949109d
commit 360a333145
No known key found for this signature in database
GPG key ID: 449E073F9DC10735
10 changed files with 98 additions and 122 deletions
Download patch file Download diff file Expand all files Collapse all files

24
examples/batched/batched.cpp
View file

@ -99,19 +99,13 @@ int main(int argc, char ** argv) {
// create a llama_batch
// we use this object to submit token data for decoding
llama_batch batch = llama_batch_init(std::max(tokens_list.size(), (size_t)n_parallel), 0, 1);
// evaluate the initial prompt
batch.n_tokens = tokens_list.size();
for (int32_t i = 0; i < batch.n_tokens; i++) {
batch.token[i] = tokens_list[i];
batch.pos[i] = i;
batch.n_seq_id[i] = 1;
batch.seq_id[i][0] = 0;
batch.logits[i] = false;
for (size_t i = 0; i < tokens_list.size(); ++i) {
llama_batch_add(batch, tokens_list[i], i, { 0 }, false);
}
GGML_ASSERT(batch.n_tokens == (int) tokens_list.size());
// llama_decode will output logits only for the last token of the prompt
batch.logits[batch.n_tokens - 1] = true;
@ -147,7 +141,7 @@ int main(int argc, char ** argv) {
while (n_cur <= n_len) {
// prepare the next batch
batch.n_tokens = 0;
llama_batch_clear(batch);
// sample the next token for each parallel sequence / stream
for (int32_t i = 0; i < n_parallel; ++i) {
@ -199,16 +193,10 @@ int main(int argc, char ** argv) {
streams[i] += llama_token_to_piece(ctx, new_token_id);
// push this new token for next evaluation
batch.token [batch.n_tokens] = new_token_id;
batch.pos [batch.n_tokens] = n_cur;
batch.n_seq_id[batch.n_tokens] = 1;
batch.seq_id [batch.n_tokens][0] = i;
batch.logits [batch.n_tokens] = true;
i_batch[i] = batch.n_tokens;
batch.n_tokens += 1;
// push this new token for next evaluation
llama_batch_add(batch, new_token_id, n_cur, { i }, true);
n_decode += 1;
}