sdgoij/llama.cpp
1
0
Fork 0
mirror of https://github.com/ggml-org/llama.cpp.git synced 2026-03-17 16:44:07 +00:00
Code Issues Packages Projects Releases Wiki Activity

llama-graph: replace cont with reshape for alpha in qwen35 (#20640)

Browse Source
This commit is contained in:
Aman Gupta
2026-03-16 22:07:13 +08:00
committed by GitHub
parent 67a2209fab
commit 3c8521c4f5
2 changed files with 2 additions and 2 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
src/models/qwen35.cpp
View File

@@ -224,7 +224,7 @@ ggml_tensor * llm_build_qwen35::build_layer_attn_linear(
beta = ggml_sigmoid(ctx0, beta);
ggml_tensor * alpha = build_lora_mm(model.layers[il].ssm_alpha, cur, model.layers[il].ssm_alpha_s);
alpha = ggml_cont_3d(ctx0, alpha, num_v_heads, n_seq_tokens, n_seqs);
alpha = ggml_reshape_3d(ctx0, alpha, num_v_heads, n_seq_tokens, n_seqs);
cb(alpha, "alpha", il);
ggml_tensor * alpha_biased = ggml_add(ctx0, alpha, model.layers[il].ssm_dt);

2
src/models/qwen35moe.cpp
View File

@@ -224,7 +224,7 @@ ggml_tensor * llm_build_qwen35moe ::build_layer_attn_linear(
beta = ggml_sigmoid(ctx0, beta);
ggml_tensor * alpha = build_lora_mm(model.layers[il].ssm_alpha, cur, model.layers[il].ssm_alpha_s);
alpha = ggml_cont_3d(ctx0, alpha, num_v_heads, n_seq_tokens, n_seqs);
alpha = ggml_reshape_3d(ctx0, alpha, num_v_heads, n_seq_tokens, n_seqs);
cb(alpha, "alpha", il);
ggml_tensor * alpha_biased = ggml_add(ctx0, alpha, model.layers[il].ssm_dt);