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

fix: wrong parameter for flash attention in pickle __getstate__

Browse source
This commit is contained in:
Andrei Betlen 2024-04-30 09:32:47 -04:00
parent 22d77eefd2
commit 29b6e9a5c8
1 changed files with 1 additions and 1 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

2
llama_cpp/llama.py
View file

@ -1777,7 +1777,7 @@ class Llama:
logits_all=self.context_params.logits_all, logits_all=self.context_params.logits_all,
embedding=self.context_params.embeddings, embedding=self.context_params.embeddings,
offload_kqv=self.context_params.offload_kqv, offload_kqv=self.context_params.offload_kqv,
flash_offload=self.context_params.flash_offload, flash_attn=self.context_params.flash_attn,
# Sampling Params # Sampling Params
last_n_tokens_size=self.last_n_tokens_size, last_n_tokens_size=self.last_n_tokens_size,
# LoRA Params # LoRA Params