Mirrors/llama.cpp
1
0
Fork 0
mirror of https://github.com/ggerganov/llama.cpp.git synced 2024-12-24 13:28:50 +01:00
Code Issues Packages Projects Releases Wiki Activity

embedding : evaluate prompt in batches (#2713)

Browse Source
This commit is contained in:
slaren 2023-08-22 16:03:12 +02:00 committed by GitHub
parent 1123f7fbdf
commit 519c981f8b
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
1 changed files with 22 additions and 15 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

15
examples/embedding/embedding.cpp
View File

@ -72,12 +72,20 @@ int main(int argc, char ** argv) {
fprintf(stderr, "\n");
}
if (params.embedding){
if (embd_inp.size() > 0) {
if (llama_eval(ctx, embd_inp.data(), embd_inp.size(), n_past, params.n_threads)) {
if (embd_inp.size() > (size_t)params.n_ctx) {
fprintf(stderr, "%s: error: prompt is longer than the context window (%zu tokens, n_ctx = %d)\n",
__func__, embd_inp.size(), params.n_ctx);
return 1;
}
while (!embd_inp.empty()) {
int n_tokens = std::min(params.n_batch, (int) embd_inp.size());
if (llama_eval(ctx, embd_inp.data(), n_tokens, n_past, params.n_threads)) {
fprintf(stderr, "%s : failed to eval\n", __func__);
return 1;
}
n_past += n_tokens;
embd_inp.erase(embd_inp.begin(), embd_inp.begin() + n_tokens);
}
const int n_embd = llama_n_embd(ctx);
@ -87,7 +95,6 @@ int main(int argc, char ** argv) {
printf("%f ", embeddings[i]);
}
printf("\n");
}
llama_print_timings(ctx);
llama_free(ctx);