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

Update llama.cpp

Browse source
This commit is contained in:
Andrei Betlen 2023-12-13 21:43:16 -05:00
parent e9bc4c4baf
commit c0fc0a1e82
2 changed files with 3 additions and 3 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

4
llama_cpp/llama_cpp.py
View file

@ -405,7 +405,7 @@ class llama_model_params(Structure):
# // Keep the booleans together to avoid misalignment during copy-by-value.
# bool mul_mat_q; // if true, use experimental mul_mat_q kernels (DEPRECATED - always true)
# bool logits_all; // the llama_eval() call computes all logits, not just the last one
# bool logits_all; // the llama_eval() call computes all logits, not just the last one (DEPRECATED - set llama_batch.logits instead)
# bool embedding; // embedding mode only
# bool offload_kqv; // whether to offload the KQV ops (including the KV cache) to GPU
# };
@ -430,7 +430,7 @@ class llama_context_params(Structure):
type_v (int): data type for V cache
mul_mat_q (bool): if true, use experimental mul_mat_q kernels (DEPRECATED - always true)
f16_kv (bool): use fp16 for KV cache, fp32 otherwise
logits_all (bool): the llama_eval() call computes all logits, not just the last one
logits_all (bool): the llama_eval() call computes all logits, not just the last one (DEPRECATED - set llama_batch.logits instead)
embedding (bool): embedding mode only"""
_fields_ = [
("seed", c_uint32),

2
vendor/llama.cpp vendored

@ -1 +1 @@
Subproject commit 8a7b2fa528f130631a5f43648481596ab320ed5a
Subproject commit 948ff137ec37f1ec74c02905917fa0afc9b97514