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

server: fix crash in multimodal models with add_bos_token = false

Browse source
This commit is contained in:
Aidan Thornton 2024-01-12 23:38:56 +00:00
parent de473f5f8e
commit 3b36f2068e
1 changed files with 1 additions and 1 deletions
Download patch file Download diff file Expand all files Collapse all files

2
examples/server/server.cpp
View file

@ -1835,7 +1835,7 @@ struct llama_server_context
slot.cache_tokens = prompt_tokens; slot.cache_tokens = prompt_tokens;
if (slot.n_past == slot.num_prompt_tokens) if (slot.n_past == slot.num_prompt_tokens && slot.n_past > 0)
{ {
// we have to evaluate at least 1 token to generate logits. // we have to evaluate at least 1 token to generate logits.
LOG_TEE("slot %d : we have to evaluate at least 1 token to generate logits\n", slot.id); LOG_TEE("slot %d : we have to evaluate at least 1 token to generate logits\n", slot.id);