diff --git a/otherarch/gptj_v3.cpp b/otherarch/gptj_v3.cpp index b00bd6bd2..116c60dda 100644 --- a/otherarch/gptj_v3.cpp +++ b/otherarch/gptj_v3.cpp @@ -462,8 +462,8 @@ bool gptj_eval( // self-attention { - struct ggml_tensor * Qcur = ggml_rope_inplace(ctx0, ggml_reshape_3d(ctx0, ggml_mul_mat(ctx0, model.layers[il].c_attn_q_proj_w, cur), n_embd/n_head, n_head, N), n_past, n_rot, 0); - struct ggml_tensor * Kcur = ggml_rope_inplace(ctx0, ggml_reshape_3d(ctx0, ggml_mul_mat(ctx0, model.layers[il].c_attn_k_proj_w, cur), n_embd/n_head, n_head, N), n_past, n_rot, 0); + struct ggml_tensor * Qcur = ggml_rope_inplace(ctx0, ggml_reshape_3d(ctx0, ggml_mul_mat(ctx0, model.layers[il].c_attn_q_proj_w, cur), n_embd/n_head, n_head, N), n_past, n_rot, 0, 0); + struct ggml_tensor * Kcur = ggml_rope_inplace(ctx0, ggml_reshape_3d(ctx0, ggml_mul_mat(ctx0, model.layers[il].c_attn_k_proj_w, cur), n_embd/n_head, n_head, N), n_past, n_rot, 0, 0); // store key and value to memory { diff --git a/otherarch/neox_v3.cpp b/otherarch/neox_v3.cpp index 245d383d6..d0a124a59 100644 --- a/otherarch/neox_v3.cpp +++ b/otherarch/neox_v3.cpp @@ -492,8 +492,8 @@ bool gpt_neox_eval( struct ggml_tensor * Vcur = ggml_cont(ctx0, ggml_view_3d(ctx0, cur, n_embd/n_head, n_head, N, cur->nb[1]/n_head, cur->nb[1], 2*sizeof(float)*n_embd/n_head)); // using mode = 2 for GPT-NeoX mode - Qcur = ggml_rope_inplace(ctx0, Qcur, n_past, n_rot, 2); - Kcur = ggml_rope_inplace(ctx0, Kcur, n_past, n_rot, 2); + Qcur = ggml_rope_inplace(ctx0, Qcur, n_past, n_rot, 2, 0); + Kcur = ggml_rope_inplace(ctx0, Kcur, n_past, n_rot, 2, 0); // store key and value to memory {