mirror of
https://github.com/ggml-org/llama.cpp.git
synced 2025-11-03 09:22:01 +00:00
llama : clear whitespaces
This commit is contained in:
@@ -6850,7 +6850,7 @@ struct llm_build_context {
|
||||
cur = build_rwkv_channel_mix(layer, ffn_norm, x_prev, LLM_ARCH_RWKV6);
|
||||
cur = ggml_add(ctx0, cur, ffn_inp);
|
||||
|
||||
token_shift = ggml_concat(ctx0,
|
||||
token_shift = ggml_concat(ctx0,
|
||||
ggml_view_3d(ctx0, att_norm, n_embd, 1, n_seqs, att_norm->nb[1], att_norm->nb[2], (n_seq_tokens-1)*n_embd*ggml_element_size(att_norm)),
|
||||
ggml_view_3d(ctx0, ffn_norm, n_embd, 1, n_seqs, ffn_norm->nb[1], ffn_norm->nb[2], (n_seq_tokens-1)*n_embd*ggml_element_size(ffn_norm)),
|
||||
1
|
||||
|
||||
Reference in New Issue
Block a user