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

remove metal check

Browse source
This commit is contained in:
Johannes Gäßler 2024-08-24 19:10:12 +02:00
parent 832c6ee394
commit 6e4080450d
1 changed files with 0 additions and 8 deletions
Download patch file Download diff file Expand all files Collapse all files

8
src/llama.cpp
View file

@ -17534,14 +17534,6 @@ struct llama_context * llama_new_context_with_model(
params.flash_attn = false; params.flash_attn = false;
} }
#ifdef GGML_USE_METAL
if (params.flash_attn && model->hparams.attn_soft_cap) {
LLAMA_LOG_WARN("%s: flash_attn is not compatible with attn_soft_cap - forcing off\n", __func__);
params.flash_attn = false;
}
#endif // GGML_USE_METAL
if (params.flash_attn && model->hparams.n_embd_head_k != model->hparams.n_embd_head_v) { if (params.flash_attn && model->hparams.n_embd_head_k != model->hparams.n_embd_head_v) {
LLAMA_LOG_WARN("%s: flash_attn requires n_embd_head_k == n_embd_head_v - forcing off\n", __func__); LLAMA_LOG_WARN("%s: flash_attn requires n_embd_head_k == n_embd_head_v - forcing off\n", __func__);
params.flash_attn = false; params.flash_attn = false;