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

free batch before return

Browse source
This commit is contained in:
Xuan Son Nguyen 2024-10-12 22:47:27 +02:00
parent 6a9769a260
commit 0639ff16d0
2 changed files with 3 additions and 0 deletions
Download patch file Download diff file Expand all files Collapse all files

1
examples/imatrix/imatrix.cpp
View file

@ -518,6 +518,7 @@ static bool compute_imatrix(llama_context * ctx, const common_params & params) {
if (llama_decode(ctx, batch)) {
LOG_ERR("%s : failed to eval\n", __func__);
llama_batch_free(batch);
return false;
}

2
examples/perplexity/perplexity.cpp
View file

@ -423,6 +423,7 @@ static results_perplexity perplexity_v2(llama_context * ctx, const common_params
//LOG_DBG(" Batch %d: starts at %d, size is %d, n_past is %d\n",j,batch_start,batch_size,j * n_batch);
if (llama_decode(ctx, batch)) {
//LOG_ERR("%s : failed to eval\n", __func__);
llama_batch_free(batch);
return {tokens, -1, logit_history, prob_history};
}
@ -1821,6 +1822,7 @@ static void kl_divergence(llama_context * ctx, const common_params & params) {
if (llama_decode(ctx, batch)) {
LOG_ERR("%s : failed to eval\n", __func__);
llama_batch_free(batch);
return;
}