Skip to content

models : fix qwen3.5 beta/gate shapes #19730

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
ggerganov merged 2 commits into from
Feb 19, 2026
Merged

models : fix qwen3.5 beta/gate shapes #19730

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
d3c3dc9
models : fix qwen3.5 beta/gate shapes
ggerganov Feb 19, 2026
a5cea28
cont : avoid extra reshapes
ggerganov Feb 19, 2026
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
12 changes: 5 additions & 7 deletions src/models/kimi-linear.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -149,17 +149,19 @@ llm_build_kimi_linear::llm_build_kimi_linear(const llama_model & model, const ll
g1 = ggml_mul(ctx0, g1, A);
cb(g1, "kda_g1", il);

g1 = ggml_reshape_4d(ctx0, g1, head_dim, n_head, n_seq_tokens, n_seqs);

// Compute beta (mixing coefficient)
ggml_tensor * beta = ggml_mul_mat(ctx0, layer.ssm_beta, cur);
beta = ggml_reshape_4d(ctx0, beta, n_head, 1, n_seq_tokens, n_seqs);
beta = ggml_reshape_4d(ctx0, beta, 1, n_head, n_seq_tokens, n_seqs);
cb(beta, "kda_beta", il);

beta = ggml_sigmoid(ctx0, beta);

// Reshape for KDA recurrence
// {n_embd, n_tokens} -> {n_embd, n_seq_tokens, n_seqs}
cur = ggml_reshape_3d(ctx0, cur, cur->ne[0], n_seq_tokens, n_seqs);

g1 = ggml_reshape_4d(ctx0, g1, head_dim, n_head, n_seq_tokens, n_seqs);

// Get SSM state and compute KDA recurrence using ggml_kda_scan
ggml_tensor * ssm_states_all = mctx_cur->get_s_l(il);
ggml_tensor * state = build_rs(inp_rs, ssm_states_all, hparams.n_embd_s(), n_seqs);
Expand All @@ -169,10 +171,6 @@ llm_build_kimi_linear::llm_build_kimi_linear(const llama_model & model, const ll

Qcur = ggml_l2_norm(ctx0, Qcur, eps_norm);
Kcur = ggml_l2_norm(ctx0, Kcur, eps_norm);
beta = ggml_sigmoid(ctx0, beta);

beta = ggml_reshape_4d(ctx0, beta, 1, n_head, n_seq_tokens, n_seqs);
g1 = ggml_reshape_4d(ctx0, g1, head_dim, n_head, n_seq_tokens, n_seqs);

// Choose between build_delta_net_chunking and build_delta_net_recurrent based on n_tokens
std::pair<ggml_tensor *, ggml_tensor *> attn_out = n_seq_tokens == 1 ?
Expand Down
4 changes: 3 additions & 1 deletion src/models/qwen35.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -216,7 +216,7 @@ ggml_tensor * llm_build_qwen35::build_layer_attn_linear(
ggml_tensor * z = qkvz.second;

ggml_tensor * beta = build_lora_mm(model.layers[il].ssm_beta, cur);
beta = ggml_reshape_4d(ctx0, beta, num_v_heads, 1, n_seq_tokens, n_seqs);
beta = ggml_reshape_4d(ctx0, beta, 1, num_v_heads, n_seq_tokens, n_seqs);
cb(beta, "beta", il);

beta = ggml_sigmoid(ctx0, beta);
Expand All @@ -232,6 +232,8 @@ ggml_tensor * llm_build_qwen35::build_layer_attn_linear(
ggml_tensor * gate = ggml_mul(ctx0, alpha_softplus, model.layers[il].ssm_a); // -A_log.exp() * softplus
cb(gate, "gate", il);

gate = ggml_reshape_4d(ctx0, gate, 1, num_v_heads, n_seq_tokens, n_seqs);

// Get convolution states from cache
ggml_tensor * conv_states_all = mctx_cur->get_r_l(il);
ggml_tensor * ssm_states_all = mctx_cur->get_s_l(il);
Expand Down
4 changes: 3 additions & 1 deletion src/models/qwen35moe.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -216,7 +216,7 @@ ggml_tensor * llm_build_qwen35moe ::build_layer_attn_linear(
ggml_tensor * z = qkvz.second;

ggml_tensor * beta = build_lora_mm(model.layers[il].ssm_beta, cur);
beta = ggml_reshape_4d(ctx0, beta, num_v_heads, 1, n_seq_tokens, n_seqs);
beta = ggml_reshape_4d(ctx0, beta, 1, num_v_heads, n_seq_tokens, n_seqs);
cb(beta, "beta", il);

beta = ggml_sigmoid(ctx0, beta);
Expand All @@ -232,6 +232,8 @@ ggml_tensor * llm_build_qwen35moe ::build_layer_attn_linear(
ggml_tensor * gate = ggml_mul(ctx0, alpha_softplus, model.layers[il].ssm_a); // -A_log.exp() * softplus
cb(gate, "gate", il);

gate = ggml_reshape_4d(ctx0, gate, 1, num_v_heads, n_seq_tokens, n_seqs);

// Get convolution states from cache
ggml_tensor * conv_states_all = mctx_cur->get_r_l(il);
ggml_tensor * ssm_states_all = mctx_cur->get_s_l(il);
Expand Down
Loading