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

Removed usage of output bias tensor since it's not present in DeepSeek-V2 models.

Browse source
This commit is contained in:
Stanisław Szymczyk 2024-05-21 17:53:37 +02:00
parent 842ff3fed1
commit c033958d7c
1 changed files with 1 additions and 1 deletions
Download patch file Download diff file Expand all files Collapse all files

2
llama.cpp
View file

@ -11080,7 +11080,7 @@ struct llm_build_context {
key_states = ggml_set_inplace(ctx0, key_states, k_pe, key_states->nb[1], key_states->nb[2], key_states->nb[3], ggml_element_size(key_states) * qk_nope_head_dim);
cur = llm_build_kv(ctx0, model, hparams, cparams, kv_self, gf,
model.layers[il].wo, model.layers[il].bo,
model.layers[il].wo, NULL,
key_states, value_states, query_states, KQ_mask, n_tokens, kv_head, n_kv, kq_scale, cb, il);
}