root/llama.cpp
1
0
Fork 0
mirror of https://github.com/ggerganov/llama.cpp.git synced 2024-12-28 12:24:35 +00:00
Code Issues Actions 2 Packages Projects Releases Wiki Activity

gguf-llama.cpp : fix n_head_kv

Browse Source
This commit is contained in:
klosax 2023-08-14 23:18:41 +02:00 committed by GitHub
parent ab2cbd03ca
commit ca4758290c
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
1 changed files with 3 additions and 2 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

5
gguf-llama.cpp
View File

@ -630,8 +630,9 @@ struct gguf_file_loader {
hparams.n_rot = read_u32("llama.rope.dimension_count"); hparams.n_rot = read_u32("llama.rope.dimension_count");
hparams.f_rms_norm_eps = read_f32("llama.attention.layer_norm_rms_epsilon"); hparams.f_rms_norm_eps = read_f32("llama.attention.layer_norm_rms_epsilon");
// LLaMAv2 // n_head_kv default to n_head
// hparams.n_head_kv = read_u32("llama.attention.head_count_kv"); hparams.n_head_kv = gguf_find_key(gguf_ctx, "llama.attention.head_count_kv") == -1 ? hparams.n_head : read_u32("llama.attention.head_count_kv");
} }
void read_vocab() { void read_vocab() {