llama : place d2t by its set_rows use so it can stay on the GPU (#30912)

This commit is contained in:
Aman Gupta authored and GitHub committed 2026-10-21 16:32:29 +08:00
1 parent ea9e3446b5
commit eb182d85c0
2 files changed
+8 -1

No files matched your search

+1 -1
View File
@@ -1110,7 +1110,7 @@ static const std::map<llm_tensor, llm_tensor_info> LLM_TENSOR_INFOS = {
{LLM_TENSOR_MASKED_EMBD_ORDERING, {LLM_TENSOR_LAYER_INPUT, GGML_OP_NONE}},
// eagle3
{LLM_TENSOR_FC, {LLM_TENSOR_LAYER_OUTPUT, GGML_OP_MUL_MAT}},
{LLM_TENSOR_D2T, {LLM_TENSOR_LAYER_OUTPUT, GGML_OP_GET_ROWS}},
{LLM_TENSOR_D2T, {LLM_TENSOR_LAYER_OUTPUT, GGML_OP_SET_ROWS}},
// dspark
{LLM_TENSOR_DSPARK_MARKOV_W1, {LLM_TENSOR_LAYER_OUTPUT, GGML_OP_GET_ROWS}},
{LLM_TENSOR_DSPARK_MARKOV_W2, {LLM_TENSOR_LAYER_OUTPUT, GGML_OP_MUL_MAT}},