Skip to content

model: using single llm_build per arch #21970

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
ngxson merged 5 commits into from
Apr 16, 2026
Merged

model: using single llm_build per arch #21970

Show file tree
Hide file tree
Changes from 1 commit
Commits
Show all changes
5 commits
Select commit Hold shift + click to select a range
be46a50
model: using single llm_build per arch
ngxson Apr 15, 2026
a2295b0
Merge branch 'master' into xsn/one_llm_build_per_arch
ngxson Apr 16, 2026
8c39fc8
fix merge
ngxson Apr 16, 2026
13a4d7a
Merge branch 'master' into xsn/one_llm_build_per_arch
ngxson Apr 16, 2026
e6125f8
nits
ngxson Apr 16, 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
6 changes: 3 additions & 3 deletions src/CMakeLists.txt
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -92,8 +92,8 @@ add_library(llama
models/lfm2.cpp
models/llada-moe.cpp
models/llada.cpp
models/llama-iswa.cpp
models/llama.cpp
models/llama4.cpp
models/maincoder.cpp
models/mamba-base.cpp
models/mamba.cpp
Expand Down Expand Up @@ -145,8 +145,8 @@ add_library(llama
models/starcoder.cpp
models/starcoder2.cpp
models/step35-iswa.cpp
models/t5-dec.cpp
models/t5-enc.cpp
models/t5.cpp
models/t5encoder.cpp
models/wavtokenizer-dec.cpp
models/xverse.cpp
)
Expand Down
31 changes: 12 additions & 19 deletions src/llama-model.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1274,8 +1274,7 @@ void llama_model::load_hparams(llama_model_loader & ml) {
}
// Set non-causal attention for diffusion models
hparams.causal_attn = false;
}
break;
} break;
case LLM_ARCH_LLADA:
{
ml.get_key(LLM_KV_ATTENTION_LAYERNORM_RMS_EPS, hparams.f_norm_rms_eps);
Expand All @@ -1289,8 +1288,7 @@ void llama_model::load_hparams(llama_model_loader & ml) {
}
// Set non-causal attention for diffusion models
hparams.causal_attn = false;
}
break;
} break;
case LLM_ARCH_LLADA_MOE:
{
ml.get_key(LLM_KV_EXPERT_FEED_FORWARD_LENGTH, hparams.n_ff_exp, false);
Expand Down Expand Up @@ -8769,9 +8767,9 @@ ggml_cgraph * llama_model::build_graph(const llm_graph_params & params) const {
case LLM_ARCH_LLAMA4:
{
if (hparams.swa_type == LLAMA_SWA_TYPE_NONE) {
llm = std::make_unique<llm_build_llama<false>>(*this, params);
llm = std::make_unique<llm_build_llama4<false>>(*this, params);
} else {
llm = std::make_unique<llm_build_llama_iswa>(*this, params);
llm = std::make_unique<llm_build_llama4<true>>(*this, params);
}
} break;
case LLM_ARCH_LLAMA_EMBED:
Expand Down Expand Up @@ -8849,23 +8847,19 @@ ggml_cgraph * llama_model::build_graph(const llm_graph_params & params) const {
case LLM_ARCH_DREAM:
{
llm = std::make_unique<llm_build_dream>(*this, params);
}
break;
} break;
case LLM_ARCH_LLADA:
{
llm = std::make_unique<llm_build_llada>(*this, params);
}
break;
} break;
case LLM_ARCH_LLADA_MOE:
{
llm = std::make_unique<llm_build_llada_moe>(*this, params);
}
break;
} break;
case LLM_ARCH_RND1:
{
llm = std::make_unique<llm_build_rnd1>(*this, params);
}
break;
} break;
case LLM_ARCH_QWEN2VL:
{
llm = std::make_unique<llm_build_qwen2vl>(*this, params);
Expand Down Expand Up @@ -9055,21 +9049,20 @@ ggml_cgraph * llama_model::build_graph(const llm_graph_params & params) const {
{
switch (params.gtype) {
case LLM_GRAPH_TYPE_ENCODER:
llm = std::make_unique<llm_build_t5_enc>(*this, params);
llm = std::make_unique<llm_build_t5<true>>(*this, params);
break;
case LLM_GRAPH_TYPE_DEFAULT:
case LLM_GRAPH_TYPE_DECODER:
llm = std::make_unique<llm_build_t5_dec>(*this, params);
llm = std::make_unique<llm_build_t5<false>>(*this, params);
break;
default:
GGML_ABORT("invalid graph type");
};
} break;
case LLM_ARCH_T5ENCODER:
{
llm = std::make_unique<llm_build_t5_enc>(*this, params);
}
break;
llm = std::make_unique<llm_build_t5encoder>(*this, params);
} break;
case LLM_ARCH_JAIS:
{
llm = std::make_unique<llm_build_jais>(*this, params);
Expand Down
16 changes: 14 additions & 2 deletions src/models/llama-iswa.cpp → src/models/llama4.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -1,6 +1,7 @@
#include "models.h"

llm_build_llama_iswa::llm_build_llama_iswa(const llama_model & model, const llm_graph_params & params) : llm_graph_context(params) {
template <bool iswa>
llm_build_llama4<iswa>::llm_build_llama4(const llama_model & model, const llm_graph_params & params) : llm_graph_context(params) {
const int64_t n_embd_head = hparams.n_embd_head_v();

GGML_ASSERT(n_embd_head == hparams.n_embd_head_k());
Expand All @@ -18,7 +19,14 @@ llm_build_llama_iswa::llm_build_llama_iswa(const llama_model & model, const llm_
ggml_tensor * inp_attn_scale = nullptr;
inp_attn_scale = build_inp_attn_scale();

auto * inp_attn = build_attn_inp_kv_iswa();
using inp_attn_type = std::conditional_t<iswa, llm_graph_input_attn_kv_iswa, llm_graph_input_attn_kv>;
inp_attn_type * inp_attn = nullptr;

if constexpr (iswa) {
inp_attn = build_attn_inp_kv_iswa();
} else {
inp_attn = build_attn_inp_kv();
}

const float kq_scale = hparams.f_attention_scale == 0.0f ? 1.0f/sqrtf(float(n_embd_head)) : hparams.f_attention_scale;

Expand Down Expand Up @@ -176,3 +184,7 @@ llm_build_llama_iswa::llm_build_llama_iswa(const llama_model & model, const llm_

ggml_build_forward_expand(gf, cur);
}

// Explicit template instantiations
template struct llm_build_llama4<false>;
template struct llm_build_llama4<true>;
14 changes: 8 additions & 6 deletions src/models/models.h
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -407,8 +407,9 @@ struct llm_build_llama : public llm_graph_context {
llm_build_llama(const llama_model & model, const llm_graph_params & params);
};

struct llm_build_llama_iswa : public llm_graph_context {
llm_build_llama_iswa(const llama_model & model, const llm_graph_params & params);
template <bool iswa>
struct llm_build_llama4 : public llm_graph_context {
llm_build_llama4(const llama_model & model, const llm_graph_params & params);
};

struct llm_build_maincoder : public llm_graph_context {
Expand Down Expand Up @@ -701,12 +702,13 @@ struct llm_build_step35_iswa : public llm_graph_context {
llm_build_step35_iswa(const llama_model & model, const llm_graph_params & params);
};

struct llm_build_t5_dec : public llm_graph_context {
llm_build_t5_dec(const llama_model & model, const llm_graph_params & params);
template <bool is_enc>
struct llm_build_t5 : public llm_graph_context {
llm_build_t5(const llama_model & model, const llm_graph_params & params);
};

struct llm_build_t5_enc : public llm_graph_context {
llm_build_t5_enc(const llama_model & model, const llm_graph_params & params);
struct llm_build_t5encoder : public llm_build_t5<true> {
llm_build_t5encoder(const llama_model & model, const llm_graph_params & params);
Comment on lines +710 to +711

@ggerganov ggerganov Apr 16, 2026

Copy link
Copy Markdown
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Do we need to keep this, instead of using llm_build_t5<true>?

@CISC CISC Apr 16, 2026

Copy link
Copy Markdown
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I think it's there to match the arch name (for the migration script)?

@ngxson ngxson Apr 16, 2026

Copy link
Copy Markdown
Collaborator Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

@ggerganov the t5encoder has its own tensor loader and hparams loader, so I think it make sense to have a dedicated graph builder. The dedicated class will then expanded into llama_model_t5encoder during the migration (via a script), and its hparams/tensors loader will be moved there

};

struct llm_build_wavtokenizer_dec : public llm_graph_context {
Expand Down
96 changes: 0 additions & 96 deletions src/models/t5-enc.cpp
View file
Open in desktop

This file was deleted.

99 changes: 98 additions & 1 deletion src/models/t5-dec.cpp → src/models/t5.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -1,6 +1,7 @@
#include "models.h"

llm_build_t5_dec::llm_build_t5_dec(const llama_model & model, const llm_graph_params & params) : llm_graph_context(params) {
template <>
llm_build_t5<false>::llm_build_t5(const llama_model & model, const llm_graph_params & params) : llm_graph_context(params) {
const int64_t n_embd_head = hparams.n_embd_head_v();
//const int64_t n_embd_gqa = hparams.n_embd_v_gqa();

Expand Down Expand Up @@ -164,3 +165,99 @@ llm_build_t5_dec::llm_build_t5_dec(const llama_model & model, const llm_graph_pa

ggml_build_forward_expand(gf, cur);
}

template <>
llm_build_t5<true>::llm_build_t5(const llama_model & model, const llm_graph_params & params) : llm_graph_context(params) {
const int64_t n_embd_head = hparams.n_embd_head_v();

GGML_ASSERT(n_embd_head == hparams.n_embd_head_k());

ggml_tensor * cur;
ggml_tensor * inpL;

inpL = build_inp_embd(model.tok_embd);

ggml_tensor * pos_bucket_enc = build_inp_pos_bucket_enc();

auto * inp_attn = build_attn_inp_no_cache();

ggml_tensor * inp_out_ids = build_inp_out_ids();

for (int il = 0; il < n_layer; ++il) {
ggml_tensor * inpSA = inpL;

// norm
cur = build_norm(inpL,
model.layers[il].attn_norm_enc, NULL,
LLM_NORM_RMS, il);
cb(cur, "attn_norm", il);

// self-attention
{
ggml_tensor * Qcur = build_lora_mm(model.layers[il].wq_enc, cur);
cb(Qcur, "Qcur", il);

ggml_tensor * Kcur = build_lora_mm(model.layers[il].wk_enc, cur);
cb(Kcur, "Kcur", il);

ggml_tensor * Vcur = build_lora_mm(model.layers[il].wv_enc, cur);
cb(Vcur, "Vcur", il);

Qcur = ggml_reshape_3d(ctx0, Qcur, n_embd_head, n_head, n_tokens);
Kcur = ggml_reshape_3d(ctx0, Kcur, n_embd_head, n_head_kv, n_tokens);
Vcur = ggml_reshape_3d(ctx0, Vcur, n_embd_head, n_head_kv, n_tokens);

ggml_tensor * attn_rel_b = model.layers[il].attn_rel_b_enc ? model.layers[il].attn_rel_b_enc : model.layers[0].attn_rel_b_enc;
ggml_tensor * kq_b = build_pos_bias(pos_bucket_enc, attn_rel_b);

cur = build_attn(inp_attn,
model.layers[il].wo_enc, nullptr,
Qcur, Kcur, Vcur, kq_b, nullptr, nullptr, 1.0f, il);
cb(cur, "kqv_out", il);
}
if (il == n_layer - 1 && inp_out_ids) {
cur = ggml_get_rows(ctx0, cur, inp_out_ids);
inpSA = ggml_get_rows(ctx0, inpSA, inp_out_ids);
}
ggml_tensor * ffn_inp = ggml_add(ctx0, cur, inpSA);
cb(ffn_inp, "ffn_inp", il);

// feed-forward network
{
cur = build_norm(ffn_inp,
model.layers[il].ffn_norm_enc, NULL,
LLM_NORM_RMS, il);
cb(cur, "ffn_norm", il);

// T5 uses relu, flan-T5 uses gelu-gated
cur = build_ffn(cur,
model.layers[il].ffn_up_enc, NULL, NULL,
model.layers[il].ffn_gate_enc, NULL, NULL,
model.layers[il].ffn_down_enc, NULL, NULL,
NULL,
model.layers[il].ffn_gate_enc ? LLM_FFN_GELU : LLM_FFN_RELU,
model.layers[il].ffn_gate_enc ? LLM_FFN_PAR : LLM_FFN_SEQ,
il);
cb(cur, "ffn_out", il);
}
cur = ggml_add(ctx0, cur, ffn_inp);
cb(cur, "ffn_out", il);

cur = build_cvec(cur, il);
cb(cur, "l_out", il);

// input for next layer
inpL = cur;
}
cur = inpL;
cb(cur, "result_embd", -1);

cur = build_norm(cur,
model.output_norm_enc, NULL,
LLM_NORM_RMS, -1);

cb(cur, "result_norm", -1);
res->t_embd = cur;

ggml_build_forward_expand(gf, cur);
}
3 changes: 3 additions & 0 deletions src/models/t5encoder.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -0,0 +1,3 @@
#include "models.h"

llm_build_t5encoder::llm_build_t5encoder(const llama_model & model, const llm_graph_params & params) : llm_build_t5<true>(model, params) {}