llama : fix vocab_only logic when GPU is enabled
This commit is contained in:
parent
99161230c4
commit
323a9d3b8c
1 changed files with 24 additions and 24 deletions
|
@ -5547,7 +5547,6 @@ struct llama_context * llama_new_context_with_model(
|
|||
}
|
||||
#endif
|
||||
}
|
||||
}
|
||||
|
||||
#ifdef GGML_USE_METAL
|
||||
if (params.n_gpu_layers > 0) {
|
||||
|
@ -5584,6 +5583,7 @@ struct llama_context * llama_new_context_with_model(
|
|||
#undef LLAMA_METAL_CHECK_BUF
|
||||
}
|
||||
#endif
|
||||
}
|
||||
|
||||
#ifdef GGML_USE_MPI
|
||||
ctx->ctx_mpi = ggml_mpi_init();
|
||||
|
|
Loading…
Add table
Add a link
Reference in a new issue