mirror of
https://github.com/ggerganov/llama.cpp.git
synced 2024-12-24 13:28:50 +01:00
llama : fix not enough space in buffer with Qwen (#5086)
This commit is contained in:
parent
6f9939d119
commit
011e8ec577
@ -4440,9 +4440,9 @@ static struct ggml_tensor * llm_build_kv(
|
||||
|
||||
// these nodes are added to the graph together so that they are not reordered
|
||||
// by doing so, the number of splits in the graph is reduced
|
||||
ggml_build_forward_expand(graph, q_cur);
|
||||
ggml_build_forward_expand(graph, k_cur);
|
||||
ggml_build_forward_expand(graph, v_cur);
|
||||
ggml_build_forward_expand(graph, q_cur);
|
||||
|
||||
llm_build_kv_store(ctx, hparams, kv, graph, k_cur, v_cur, n_ctx, n_tokens, kv_head, cb, il);
|
||||
|
||||
|
Loading…
Reference in New Issue
Block a user