diff options
Diffstat (limited to 'src/llama.cpp')
| -rw-r--r-- | src/llama.cpp | 6 |
1 files changed, 3 insertions, 3 deletions
diff --git a/src/llama.cpp b/src/llama.cpp index 00a3c9b1..29926a94 100644 --- a/src/llama.cpp +++ b/src/llama.cpp @@ -13390,7 +13390,7 @@ struct llm_build_context { ggml_row_size(kv_pe_compresseed->type, kv_lora_rank)); cb(k_pe, "k_pe", il); - kv_compressed = ggml_cont(ctx0, kv_compressed); // TODO: the CUDA backend does not support non-contiguous norm + //kv_compressed = ggml_cont(ctx0, kv_compressed); // TODO: the CUDA backend does not support non-contiguous norm kv_compressed = llm_build_norm(ctx0, kv_compressed, hparams, model.layers[il].attn_kv_a_norm, NULL, LLM_NORM_RMS, cb, il); @@ -13422,7 +13422,7 @@ struct llm_build_context { 0); cb(v_states, "v_states", il); - q_pe = ggml_cont(ctx0, q_pe); // TODO: the CUDA backend does not support non-contiguous RoPE + //q_pe = ggml_cont(ctx0, q_pe); // TODO: the CUDA backend does not support non-contiguous RoPE q_pe = ggml_rope_ext( ctx0, q_pe, inp_pos, nullptr, n_rot, rope_type, n_ctx_orig, freq_base, freq_scale, @@ -13431,7 +13431,7 @@ struct llm_build_context { cb(q_pe, "q_pe", il); // shared RoPE key - k_pe = ggml_cont(ctx0, k_pe); // TODO: the CUDA backend does not support non-contiguous RoPE + //k_pe = ggml_cont(ctx0, k_pe); // TODO: the CUDA backend does not support non-contiguous RoPE k_pe = ggml_rope_ext( ctx0, k_pe, inp_pos, nullptr, n_rot, rope_type, n_ctx_orig, freq_base, freq_scale, |