extern/whisper.cpp
1
0
Fork 1
mirror of https://github.com/ggerganov/whisper.cpp.git synced 2025-02-05 04:50:18 +01:00
Code Issues Packages Projects Releases Wiki Activity

talk-llama : use llama_decode instead of llama_eval

Browse Source
This commit is contained in:
Georgi Gerganov 2024-03-08 12:04:43 +02:00
parent 8e409d1113
commit 2f5a5a66dd
No known key found for this signature in database
GPG Key ID: 449E073F9DC10735
1 changed files with 32 additions and 4 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

36
examples/talk-llama/talk-llama.cpp
View File

@ -391,6 +391,8 @@ int main(int argc, char ** argv) {
prompt_llama = ::replace(prompt_llama, "{4}", chat_symb);
llama_batch batch = llama_batch_init(llama_n_ctx(ctx_llama), 0, 1);
// init session
std::string path_session = params.path_session;
std::vector<llama_token> session_tokens;
@ -426,8 +428,21 @@ int main(int argc, char ** argv) {
printf("\n");
printf("%s : initializing - please wait ...\n", __func__);
if (llama_eval(ctx_llama, embd_inp.data(), embd_inp.size(), 0)) {
fprintf(stderr, "%s : failed to eval\n", __func__);
// prepare batch
{
batch.n_tokens = embd_inp.size();
for (int i = 0; i < batch.n_tokens; i++) {
batch.token[i] = embd_inp[i];
batch.pos[i] = i;
batch.n_seq_id[i] = 1;
batch.seq_id[i][0] = 0;
batch.logits[i] = i == batch.n_tokens - 1;
}
}
if (llama_decode(ctx_llama, batch)) {
fprintf(stderr, "%s : failed to decode\n", __func__);
return 1;
}
@ -647,8 +662,21 @@ int main(int argc, char ** argv) {
n_session_consumed = session_tokens.size();
}
if (llama_eval(ctx_llama, embd.data(), embd.size(), n_past)) {
fprintf(stderr, "%s : failed to eval\n", __func__);
// prepare batch
{
batch.n_tokens = embd.size();
for (int i = 0; i < batch.n_tokens; i++) {
batch.token[i] = embd[i];
batch.pos[i] = n_past + i;
batch.n_seq_id[i] = 1;
batch.seq_id[i][0] = 0;
batch.logits[i] = i == batch.n_tokens - 1;
}
}
if (llama_decode(ctx_llama, batch)) {
fprintf(stderr, "%s : failed to decode\n", __func__);
return 1;
}
}