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

changed token functions to use new model variants

Browse source
This commit is contained in:
Marcus Dunn 2023-10-23 09:29:11 -07:00
parent 38cdb82235
commit 2df3801706
1 changed files with 2 additions and 2 deletions
Download patch file Download diff file Expand all files Collapse all files

4
examples/perplexity/perplexity.cpp
View file

@ -227,7 +227,7 @@ static results_perplexity perplexity_v2(llama_context * ctx, const gpt_params &
// add BOS token for the first batch of each chunk
if (add_bos && j == 0) {
tokens[batch_start] = llama_token_bos(ctx);
tokens[batch_start] = llama_token_bos(llama_get_model(ctx));
}
const auto batch_logits = llama_get_logits(ctx);
@ -350,7 +350,7 @@ static results_perplexity perplexity(llama_context * ctx, const gpt_params & par
// add BOS token for the first batch of each chunk
if (add_bos && j == 0) {
tokens[batch_start] = llama_token_bos(ctx);
tokens[batch_start] = llama_token_bos(llama_get_model(ctx));
}
if (llama_decode(ctx, llama_batch_get_one(tokens.data() + batch_start, batch_size, j * n_batch, 0))) {