ikawrakow/ik_llama.cpp
1
0
Fork 0
mirror of https://github.com/ikawrakow/ik_llama.cpp.git synced 2026-01-26 17:20:01 +00:00
Code Issues Packages Projects Releases Wiki Activity

Split mode 'graph' fpr Qwen3-VL (#1107)

Browse Source 
Co-authored-by: Iwan Kawrakow <iwan.kawrakow@gmail.com>
This commit is contained in:
Kawrakow
2026-01-05 17:32:00 +02:00
committed by GitHub
parent 218dcc5727
commit 3c99284b67
3 changed files with 42 additions and 77 deletions
Download Patch File Download Diff File Expand all files Collapse all files

115
src/llama-build-context.cpp
View File

@@ -4077,9 +4077,10 @@ ggml_cgraph * llm_build_context::build_qwen3vl() {
int sections[4];
std::copy(std::begin(hparams.rope_sections), std::begin(hparams.rope_sections) + 4, sections);
std::vector<struct ggml_tensor *> deepstack_features(n_deepstack_layers, nullptr);
std::vector<struct ggml_tensor *> deepstack_features;
if (batch.embd) {
deepstack_features.resize(n_deepstack_layers, nullptr);
// Image input: split main embd and deepstack embds
struct ggml_tensor * inpL_main = ggml_view_2d(ctx0, inpL, n_embd, n_tokens, inpL->nb[1], 0);
for (size_t i = 0; i < n_deepstack_layers; i++) {
@@ -4095,93 +4096,33 @@ ggml_cgraph * llm_build_context::build_qwen3vl() {
struct ggml_tensor * KQ_mask = build_inp_KQ_mask();
for (int il = 0; il < n_layer; ++il) {
struct ggml_tensor * inpSA = inpL;
// norm
cur = llm_build_norm(ctx0, inpL, hparams,
model.layers[il].attn_norm, NULL,
LLM_NORM_RMS, cb, il);
cb(cur, "attn_norm", il);
// self-attention
{
auto [Qcur, Kcur, Vcur] = llm_build_mul_mat_qkv(gf, cur,
model.layers[il].wq, nullptr,
model.layers[il].wk, nullptr,
model.layers[il].wv, nullptr,
0, il);
Qcur = ggml_reshape_3d(ctx0, Qcur, n_embd_head, n_head, n_tokens);
Qcur = llm_build_norm(ctx0, Qcur, hparams, model.layers[il].attn_q_norm, NULL, LLM_NORM_RMS, cb, il);
cb(Qcur, "Qcur_normed", il);
Qcur = ggml_rope_multi(
ctx0, Qcur, inp_pos, nullptr,
n_rot, sections, rope_type, n_ctx_orig, freq_base, freq_scale,
ext_factor, attn_factor, beta_fast, beta_slow
);
cb(Qcur, "Qcur", il);
Kcur = ggml_reshape_3d(ctx0, Kcur, n_embd_head, n_head_kv, n_tokens);
Kcur = llm_build_norm(ctx0, Kcur, hparams, model.layers[il].attn_k_norm, NULL, LLM_NORM_RMS, cb, il);
cb(Kcur, "Kcur_normed", il);
Kcur = ggml_rope_multi(
ctx0, Kcur, inp_pos, nullptr,
n_rot, sections, rope_type, n_ctx_orig, freq_base, freq_scale,
ext_factor, attn_factor, beta_fast, beta_slow
);
cb(Kcur, "Kcur", il);
cb(Vcur, "Vcur", il);
cur = llm_build_kv(ctx0, lctx, kv_self, gf,
model.layers[il].wo, model.layers[il].bo,
Kcur, Vcur, Qcur, KQ_mask, n_tokens, kv_head, n_kv, 1.0f/sqrtf(float(n_embd_head)), cb, il);
}
cur = build_std_attention(gf, model.layers[il].attn_norm, inpL, inp_pos, nullptr, KQ_mask,
nullptr, nullptr, 1.0f/sqrtf(float(n_embd_head)), 0.0f, 0, il, true, false, true, false, true);
if (il == n_layer - 1) {
// skip computing output for unused tokens
struct ggml_tensor * inp_out_ids = build_inp_out_ids();
cur = ggml_get_rows(ctx0, cur, inp_out_ids);
inpSA = ggml_get_rows(ctx0, inpSA, inp_out_ids);
}
struct ggml_tensor * ffn_inp = ggml_add(ctx0, cur, inpSA);
cb(ffn_inp, "ffn_inp", il);
// feed-forward network
cur = llm_build_ffn(ctx0, lctx, model.layers[il].ffn_norm, ffn_inp,
cur = llm_build_ffn(ctx0, lctx, model.layers[il].ffn_norm, cur,
model.layers[il].ffn_up, NULL, NULL,
model.layers[il].ffn_gate, NULL, NULL,
model.layers[il].ffn_down, NULL, NULL,
NULL,
LLM_FFN_SILU, LLM_FFN_PAR, cb, il);
cb(cur, "ffn_out", il);
cur = ggml_add(ctx0, cur, ffn_inp);
LLM_FFN_SILU, LLM_FFN_PAR, cb, il, gf, true, false,
batch.embd && (size_t)il < n_deepstack_layers ? deepstack_features[il] : nullptr);
cur = lctx.cvec.apply_to(ctx0, cur, il);
cb(cur, "l_out", il);
if (batch.embd && (size_t)il < n_deepstack_layers) {
cur = ggml_add(ctx0, cur, deepstack_features[il]);
cb(cur, "deepstack_out", il);
}
// input for next layer
inpL = cur;
}
cur = inpL;
cur = llm_build_norm(ctx0, cur, hparams,
model.output_norm, NULL,
LLM_NORM_RMS, cb, -1);
cb(cur, "result_norm", -1);
// lm_head
cur = llm_build_lora_mm(lctx, ctx0, model.output, cur);
cur = build_output(lctx, ctx0, inpL, model.output, model.output_norm, cb);
cb(cur, "result_output", -1);
ggml_build_forward_expand(gf, cur);
@@ -9419,7 +9360,7 @@ ggml_cgraph * llm_build_context::llama_build_graph(
ggml_tensor * llm_build_context::build_std_attention(ggml_cgraph * gf, ggml_tensor * the_attn_norm,
ggml_tensor * input, ggml_tensor * inp_pos, ggml_tensor * rope_factors_in,
ggml_tensor * KQ_mask, ggml_tensor * sinks, ggml_tensor * inp_attn_scale, float KQ_scale, float f_attn_scale,
int n_swa, int il, bool do_rope, bool add_graph_split, bool add_input, bool is_norm) {
int n_swa, int il, bool do_rope, bool add_graph_split, bool add_input, bool is_norm, bool is_multi) {
float freq_base_l = n_swa > 0 ? hparams.rope_freq_base_train_swa : cparams.rope_freq_base;
float freq_scale_l = n_swa > 0 ? hparams.rope_freq_scale_train_swa : hparams.rope_freq_scale_train;
@@ -9492,10 +9433,21 @@ ggml_tensor * llm_build_context::build_std_attention(ggml_cgraph * gf, ggml_tens
rope_factors = extra->splits[id];
}
if (do_rope) {
Qcur = ggml_rope_ext(ctx0, Qcur, inp_pos, rope_factors, n_rot, rope_type, n_ctx_orig, freq_base_l, freq_scale_l,
ext_factor, attn_factor, beta_fast, beta_slow);
Kcur = ggml_rope_ext(ctx0, Kcur, inp_pos, rope_factors, n_rot, rope_type, n_ctx_orig, freq_base_l, freq_scale_l,
ext_factor, attn_factor, beta_fast, beta_slow);
if (is_multi) {
int sections[4];
std::copy(hparams.rope_sections.begin(), hparams.rope_sections.begin() + GGML_MROPE_SECTIONS, sections);
Qcur = ggml_rope_multi(ctx0, Qcur, inp_pos, rope_factors,
n_rot, sections, rope_type, n_ctx_orig, freq_base_l, freq_scale_l,
ext_factor, attn_factor, beta_fast, beta_slow);
Kcur = ggml_rope_multi(ctx0, Kcur, inp_pos, rope_factors,
n_rot, sections, rope_type, n_ctx_orig, freq_base_l, freq_scale_l,
ext_factor, attn_factor, beta_fast, beta_slow);
} else {
Qcur = ggml_rope_ext(ctx0, Qcur, inp_pos, rope_factors, n_rot, rope_type, n_ctx_orig, freq_base_l, freq_scale_l,
ext_factor, attn_factor, beta_fast, beta_slow);
Kcur = ggml_rope_ext(ctx0, Kcur, inp_pos, rope_factors, n_rot, rope_type, n_ctx_orig, freq_base_l, freq_scale_l,
ext_factor, attn_factor, beta_fast, beta_slow);
}
}
cb(Qcur, "Qcur", il_cb);
cb(Kcur, "Kcur", il_cb);
@@ -9634,10 +9586,21 @@ ggml_tensor * llm_build_context::build_std_attention(ggml_cgraph * gf, ggml_tens
model.layers[il].attn_q_norm, model.layers[il].attn_k_norm, f_attn_scale, il);
if (do_rope) {
Qcur = ggml_rope_ext(ctx0, Qcur, inp_pos, rope_factors_in, n_rot, rope_type, n_ctx_orig, freq_base_l, freq_scale_l,
ext_factor, attn_factor, beta_fast, beta_slow);
Kcur = ggml_rope_ext( ctx0, Kcur, inp_pos, rope_factors_in, n_rot, rope_type, n_ctx_orig, freq_base_l, freq_scale_l,
ext_factor, attn_factor, beta_fast, beta_slow);
if (is_multi) {
int sections[4];
std::copy(hparams.rope_sections.begin(), hparams.rope_sections.begin() + GGML_MROPE_SECTIONS, sections);
Qcur = ggml_rope_multi(ctx0, Qcur, inp_pos, rope_factors_in,
n_rot, sections, rope_type, n_ctx_orig, freq_base_l, freq_scale_l,
ext_factor, attn_factor, beta_fast, beta_slow);
Kcur = ggml_rope_multi(ctx0, Kcur, inp_pos, rope_factors_in,
n_rot, sections, rope_type, n_ctx_orig, freq_base_l, freq_scale_l,
ext_factor, attn_factor, beta_fast, beta_slow);
} else {
Qcur = ggml_rope_ext(ctx0, Qcur, inp_pos, rope_factors_in, n_rot, rope_type, n_ctx_orig, freq_base_l, freq_scale_l,
ext_factor, attn_factor, beta_fast, beta_slow);
Kcur = ggml_rope_ext( ctx0, Kcur, inp_pos, rope_factors_in, n_rot, rope_type, n_ctx_orig, freq_base_l, freq_scale_l,
ext_factor, attn_factor, beta_fast, beta_slow);
}
}
cb(Qcur, "Qcur", il);
cb(Kcur, "Kcur", il);

3
src/llama-build-context.h
View File

@@ -413,6 +413,7 @@ llm_expert_gating_func_type gating_op,
ggml_tensor * build_std_attention(ggml_cgraph * gf, ggml_tensor * attn_norm, ggml_tensor * cur, ggml_tensor * inp_pos, ggml_tensor * rope_factors,
ggml_tensor * KQ_mask, ggml_tensor * sinks, ggml_tensor * inp_attn_scale, float KQ_scale, float f_attn_scale,
int n_swa, int il, bool do_rope = true, bool add_graph_split = false, bool add_input = false, bool is_norm = false);
int n_swa, int il, bool do_rope = true, bool add_graph_split = false, bool add_input = false, bool is_norm = false,
bool is_multi = false);
};

1
src/llama.cpp
View File

@@ -1732,6 +1732,7 @@ static bool is_model_split_supported(const llama_model & model) {
LLM_ARCH_COHERE2,
LLM_ARCH_MIMO2,
LLM_ARCH_QWEN3,
LLM_ARCH_QWEN3VL,
};
auto it = k_supported.find(model.arch);
return it != k_supported.end();