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

Format modification

Browse source
This commit is contained in:
Jia Liu 2024-08-22 14:39:43 +08:00
parent 0451b1f9ef
commit 395ae48cb0
2 changed files with 2 additions and 1 deletions
Download patch file Download diff file Expand all files Collapse all files

1
examples/perplexity/perplexity.cpp
View file

@ -400,6 +400,7 @@ static results_perplexity perplexity_v2(llama_context * ctx, const gpt_params &
// clear the KV cache // clear the KV cache
llama_kv_cache_clear(ctx); llama_kv_cache_clear(ctx);
for (int j = 0; j < num_batches; ++j) { for (int j = 0; j < num_batches; ++j) {
const int batch_start = start + j * n_batch; const int batch_start = start + j * n_batch;
const int batch_size = std::min(end - batch_start, n_batch); const int batch_size = std::min(end - batch_start, n_batch);

2
include/llama.h
View file

@ -230,7 +230,7 @@ extern "C" {
llama_pos * pos; llama_pos * pos;
int32_t * n_seq_id; int32_t * n_seq_id;
llama_seq_id ** seq_id; llama_seq_id ** seq_id;
int8_t * output; // Previously named 'logits', renamed to 'output' now. int8_t * output; // Previously named "logits", renamed to "output" now.
// NOTE: helpers for smooth API transition - can be deprecated in the future // NOTE: helpers for smooth API transition - can be deprecated in the future
// for future-proof code, use the above fields instead and ignore everything below // for future-proof code, use the above fields instead and ignore everything below