Files
llama.cpp/src/models/cohere2moe.cpp
T
Yaniss Amazouz c61b98b875 model: add NVIDIA Nemotron-3-Puzzle-75B-A9B (NemotronHPuzzle) support (#25444)
* hparams: add per-layer n_ff_exp/n_expert_used arrays with scalar-or-array loading

G1/G2 infrastructure for variable-per-layer expert FFN size and top-k routing
(required for Puzzle-75B which has 5 distinct n_ff_exp values and 7 top-k values
across its 40 MoE layers).

Design: rename scalar members to _impl suffix (following existing convention),
add LLAMA_MAX_LAYERS arrays, add n_ff_exp(il)/n_expert_used(il) accessors with
scalar fallback. No new GGUF keys: reuses existing expert_feed_forward_length and
expert_used_count keys via get_key_or_arr (scalar -> broadcast, array -> per-layer).

- llama-hparams.h: n_ff_exp -> n_ff_exp_impl, n_expert_used -> n_expert_used_impl;
  add n_ff_exp_arr / n_expert_used_arr arrays; add per-layer accessor declarations.
- llama-hparams.cpp: implement n_ff_exp(il) and n_expert_used(il); out-of-range
  il returns impl safely (shared code, no abort).
- llama-model.cpp: central n_expert_used load changed to get_key_or_arr; derive
  impl as max-of-array for validations and backward compat; zero both new arrays;
  HunyuanVL override also zeroes n_expert_used_arr.
- llama-graph.cpp: aggregation loop in build_moe_ffn uses hparams.n_expert_used(il)
  so per-layer top-k bounds the ggml_view loop correctly.
- All other files: mechanical rename hparams.n_{ff_exp,expert_used} -> *_impl.
  Scalar arches are unaffected (broadcast fills all array slots with the single value).

(cherry picked from commit 269a81e03d66e1c353e1a203a0c03a03eb2b1a4e)

* nemotron-h: use per-layer n_ff_exp(il) and n_expert_used(il) at MoE call-sites

Load n_ff_exp via get_key_or_arr into hparams.n_ff_exp_arr in load_arch_hparams;
derive impl as max for existing uniform GGUFs.

In load_arch_tensors, compute n_ff_exp_i = hparams.n_ff_exp(i) with fallback to
n_ff(i)/n_expert_used(i) for GGUFs that omit expert_feed_forward_length.

In build_ffn_layer, pass hparams.n_expert_used(il) to build_moe_ffn so per-layer
top-k is used for expert routing selection.

All other nemotron-h behaviour (mamba2, attention, shared-exp, latent projection,
routed_scaling_factor, expert_weights_norm, sigmoid gating) is unchanged.

(cherry picked from commit b1878a101793cd4e59868ac72635a86ea694987c)

* arch/*.cpp + gguf-py: mechanical rename n_ff_exp->n_ff_exp_impl, n_expert_used->n_expert_used_impl

All non-nemotron arch files continue using the scalar impl member directly.
Behaviour is identical: the impl value is the broadcast value from the GGUF scalar.

gguf_writer: add_expert_feed_forward_length and add_expert_used_count now accept
int | Sequence[int], mirroring add_feed_forward_length, so converters can write
per-layer arrays with the same existing GGUF keys.

(cherry picked from commit 8f009f54bea5ef9a6a354123bd25e9d5ea2d5e03)

* convert: support NemotronHPuzzleForCausalLM (per-block MoE config)

Parse block_configs/mtp_block_configs into per-layer arrays (scalar-or-array
keys), append the MTP [attention, moe] sub-blocks as blk.88/blk.89 with
nextn tensors, accept the backbone.* prefix, and register the arch.
Also fix a pre-existing undeclared _experts attribute on NemotronHModel.

(cherry picked from commit d1a592f278336e78457454eb6c96bca917135f10)

* nemotron-h: distinguish Nemotron 3 Puzzle (75B.A9B) from Super (120B.A12B)

Both have 88 layers; the per-layer expert_used_count array (heterogeneous
for Puzzle, broadcast-uniform for Super) is the discriminator.

(cherry picked from commit f824e09dc812169589cf5662c92d149a4c18c30a)

* convert: accept the official Puzzle BF16 checkpoint's tensor naming

The officially distributed BF16 checkpoint (NVIDIA-Nemotron-Labs-3-Puzzle-
75B-A9B-BF16) names the trunk model.* (model.layers.*, model.embeddings,
model.norm_f) where the original release used the NemotronH-style
backbone.*, and spells the router bias e_score_correction_bias instead of
e_score_correction.bias. Normalize both at the top of
NemotronHPuzzleModel.modify_tensors so either checkpoint converts; every
tensor name in the official index (42683 keys, MTP head included) resolves
through the tensor map after normalization.

(cherry picked from commit 189b67fc2c9d50970416c94b3317a6e7baa49b03)

* laguna: use n_ff_exp_impl for the uniform-MoE FFN size

Laguna landed after this branch was cut and reads hparams.n_ff_exp as a
scalar. This series turns it into a per-layer array with an n_ff_exp(il)
accessor, so the three scalar reads no longer compile. Laguna is a
uniform MoE, so point them at the scalar fallback n_ff_exp_impl, same as
deepseek2/qwen3moe/gemma4 in this series. No behaviour change.

(cherry picked from commit dbedc9e19c50dca0acdfb402362e2707bee424ae)

* arch: extend the n_ff_exp/n_expert_used rename to archs added upstream

kimi-k3, dflash, bailingmoe3, deepseek4, granite-swa and the nemotron-h MTP
block still referenced the scalar fields by their old names. n_ff_exp and
n_expert_used are accessors now, so those reads no longer compile; point the
non-per-layer archs at the _impl scalars and use the indexed form where the
call site is per-layer.

* convert: keep Puzzle opted out of the NemotronH MTP export path

#26725 added MTP export to NemotronHModel, keyed on num_nextn_predict_layers.
Puzzle's config carries that key, but NemotronHPuzzleModel bypasses
NemotronHModel.__init__ (its per-block config needs a different setup), so
_mtp_bid was never assigned and modify_tensors raised AttributeError on any
mtp.* tensor. Puzzle's head is also laid out by mtp_block_configs, not the
mtp.layers.* form the base maps.

Set _mtp_bid to None, drop mtp.* in filter_tensors, and declare
supports_mtp_export = False so --mtp / --no-mtp fail at the CLI.

* llama: replace n_ff_exp/n_expert_used scalars with per-layer accessors

Follow-up to review feedback: the previous revision kept the scalar
hparams fields alongside the new per-layer arrays, which duplicated
state that get_key_or_arr already handles by broadcasting a scalar
value over every layer.

Drop both scalars and expose n_ff_exp(il) / n_expert_used(il) built
exactly like the existing n_head_kv(il) and n_ff(il) accessors: they
index the array and GGML_ABORT out of range, with il defaulting to 0
so genuinely uniform call sites stay a plain n_ff_exp().

Arch loaders now read both keys through get_key_or_arr over
n_layer_all, and the n_expert_used validation checks the maximum
across layers instead of a single field.

* llama: restore per-key required flags on the expert hparam reads

The scalar-to-array conversion passed required=false at every call site,
which silently made mandatory keys optional. Each read now carries the
same required flag it had before the conversion.
2026-09-03 08:53:08 +02:00

445 lines
19 KiB
C++

#include "models.h"
void llama_model_cohere2moe::load_arch_hparams(llama_model_loader & ml) {
const bool found_norm = ml.get_key(LLM_KV_ATTENTION_LAYERNORM_EPS, hparams.f_norm_eps, false);
const bool found_norm_rms = ml.get_key(LLM_KV_ATTENTION_LAYERNORM_RMS_EPS, hparams.f_norm_rms_eps, false);
if (!found_norm && !found_norm_rms) {
throw std::runtime_error("missing Cohere2 MoE norm epsilon");
}
if (!found_norm_rms) {
hparams.f_norm_rms_eps = 0.0f;
}
ml.get_key(LLM_KV_ATTENTION_SLIDING_WINDOW, hparams.n_swa);
ml.get_key(LLM_KV_LOGIT_SCALE, hparams.f_logit_scale);
ml.get_key(LLM_KV_LEADING_DENSE_BLOCK_COUNT, hparams.n_layer_dense_lead);
ml.get_key_or_arr(LLM_KV_EXPERT_FEED_FORWARD_LENGTH, hparams.n_ff_exp_arr, hparams.n_layer_all);
ml.get_key(LLM_KV_EXPERT_SHARED_FEED_FORWARD_LENGTH, hparams.n_ff_shexp, false);
ml.get_key(LLM_KV_EXPERT_SHARED_COUNT, hparams.n_expert_shared, false);
ml.get_key(LLM_KV_EXPERT_WEIGHTS_NORM, hparams.expert_weights_norm, false);
ml.get_key(LLM_KV_EXPERT_WEIGHTS_SCALE, hparams.expert_weights_scale, false);
ml.get_key(LLM_KV_EXPERT_GATING_FUNC, hparams.expert_gating_func, false);
if (hparams.expert_gating_func == LLAMA_EXPERT_GATING_FUNC_TYPE_NONE) {
hparams.expert_gating_func = LLAMA_EXPERT_GATING_FUNC_TYPE_SIGMOID;
}
hparams.swa_type = LLAMA_SWA_TYPE_STANDARD;
uint32_t swa_period = 4;
if (ml.get_key_or_arr(LLM_KV_ATTENTION_SLIDING_WINDOW_PATTERN, swa_period, false)) {
hparams.set_swa_pattern(swa_period, true);
} else {
ml.get_key_or_arr(LLM_KV_ATTENTION_SLIDING_WINDOW_PATTERN, hparams.is_swa_impl, hparams.n_layer());
}
hparams.rope_freq_base_train_swa = hparams.rope_freq_base_train;
hparams.rope_freq_scale_train_swa = hparams.rope_freq_scale_train;
ml.get_key(LLM_KV_ROPE_FREQ_BASE_SWA, hparams.rope_freq_base_train_swa, false);
switch (hparams.n_layer()) {
case 49: type = LLM_TYPE_30B_A3B; break;
default: type = LLM_TYPE_UNKNOWN;
}
}
void llama_model_cohere2moe::load_arch_tensors(llama_model_loader & ml) {
LLAMA_LOAD_LOCALS;
const bool mtp_only = (hparams.n_layer_nextn > 0) && (ml.get_weight("blk.0.attn_norm.weight") == nullptr);
// Trunk-only: the GGUF declares MTP layers in metadata but the actual MTP
// tensors live in a separate file. Mark MTP tensors NOT_REQUIRED so the
// trunk loads cleanly.
const std::string mtp_probe = "blk." + std::to_string(n_layer) + ".nextn.eh_proj.weight";
const bool trunk_only = (hparams.n_layer_nextn > 0) && (ml.get_weight(mtp_probe.c_str()) == nullptr);
const int trunk_flags = mtp_only ? TENSOR_NOT_REQUIRED : 0;
int mtp_flags = trunk_only ? TENSOR_NOT_REQUIRED : 0;
if (!ml.load_mtp) {
mtp_flags |= TENSOR_SKIP;
}
tok_embd = create_tensor(tn(LLM_TENSOR_TOKEN_EMBD, "weight"), { n_embd, n_vocab }, 0);
// output
output_norm = create_tensor(tn(LLM_TENSOR_OUTPUT_NORM, "weight"), { n_embd }, 0);
output = create_tensor(tn(LLM_TENSOR_OUTPUT, "weight"), { n_embd, n_vocab }, TENSOR_NOT_REQUIRED);
// if output is NULL, init from the input tok embed
if (output == NULL) {
output = create_tensor(tn(LLM_TENSOR_TOKEN_EMBD, "weight"), { n_embd, n_vocab }, TENSOR_DUPLICATED);
}
if (n_expert == 0) {
throw std::runtime_error("n_expert must be > 0 for Cohere2Moe");
}
if (n_expert_used == 0) {
throw std::runtime_error("n_expert_used must be > 0 for Cohere2Moe");
}
auto load_block_trunk = [&](int i, int flags) {
auto & layer = layers[i];
layer.attn_norm = create_tensor(tn(LLM_TENSOR_ATTN_NORM, "weight", i), { n_embd }, flags);
create_tensor_qkv(layer, i, n_embd, n_embd_head_k * n_head, n_embd_gqa, n_embd_gqa, flags);
layer.wo = create_tensor(tn(LLM_TENSOR_ATTN_OUT, "weight", i), { n_embd_head_k * n_head, n_embd }, flags);
if (static_cast<uint32_t>(i) < hparams.n_layer_dense_lead) {
layer.ffn_gate = create_tensor(tn(LLM_TENSOR_FFN_GATE, "weight", i), { n_embd, n_ff }, flags);
layer.ffn_down = create_tensor(tn(LLM_TENSOR_FFN_DOWN, "weight", i), { n_ff, n_embd }, flags);
layer.ffn_up = create_tensor(tn(LLM_TENSOR_FFN_UP, "weight", i), { n_embd, n_ff }, flags);
} else {
const int64_t n_ff_exp = hparams.n_ff_exp() ? hparams.n_ff_exp() : n_ff;
layer.ffn_gate_inp = create_tensor(tn(LLM_TENSOR_FFN_GATE_INP, "weight", i), { n_embd, n_expert }, flags);
layer.ffn_down_exps = create_tensor(tn(LLM_TENSOR_FFN_DOWN_EXPS, "weight", i), { n_ff_exp, n_embd, n_expert }, flags);
create_tensor_gate_up_exps(layer, i, n_embd, n_ff_exp, n_expert, flags);
if (hparams.n_expert_shared > 0) {
const int64_t n_ff_shexp = hparams.n_ff_shexp ? hparams.n_ff_shexp : n_ff_exp * hparams.n_expert_shared;
layer.ffn_gate_shexp = create_tensor(tn(LLM_TENSOR_FFN_GATE_SHEXP, "weight", i), { n_embd, n_ff_shexp }, flags);
layer.ffn_down_shexp = create_tensor(tn(LLM_TENSOR_FFN_DOWN_SHEXP, "weight", i), { n_ff_shexp, n_embd }, flags);
layer.ffn_up_shexp = create_tensor(tn(LLM_TENSOR_FFN_UP_SHEXP, "weight", i), { n_embd, n_ff_shexp }, flags);
}
}
};
auto load_block_mtp = [&](int i, int flags) {
auto & layer = layers[i];
// MTP block looks like a full-attention Cohere2 MoE decoder block.
layer.attn_norm = create_tensor(tn(LLM_TENSOR_ATTN_NORM, "weight", i), { n_embd }, flags);
create_tensor_qkv(layer, i, n_embd, n_embd_head_k * n_head, n_embd_gqa, n_embd_gqa, flags);
layer.wo = create_tensor(tn(LLM_TENSOR_ATTN_OUT, "weight", i), { n_embd_head_k * n_head, n_embd }, flags);
const int64_t n_ff_exp = hparams.n_ff_exp() ? hparams.n_ff_exp() : n_ff;
// Routed experts
layer.ffn_gate_inp = create_tensor(tn(LLM_TENSOR_FFN_GATE_INP, "weight", i), { n_embd, n_expert }, flags);
layer.ffn_down_exps = create_tensor(tn(LLM_TENSOR_FFN_DOWN_EXPS, "weight", i), { n_ff_exp, n_embd, n_expert }, flags);
create_tensor_gate_up_exps(layer, i, n_embd, n_ff_exp, n_expert, flags);
if (hparams.n_expert_shared > 0) {
const int64_t n_ff_shexp = hparams.n_ff_shexp ? hparams.n_ff_shexp : n_ff_exp * hparams.n_expert_shared;
// Shared experts
layer.ffn_gate_shexp = create_tensor(tn(LLM_TENSOR_FFN_GATE_SHEXP, "weight", i), { n_embd, n_ff_shexp }, flags);
layer.ffn_down_shexp = create_tensor(tn(LLM_TENSOR_FFN_DOWN_SHEXP, "weight", i), { n_ff_shexp, n_embd }, flags);
layer.ffn_up_shexp = create_tensor(tn(LLM_TENSOR_FFN_UP_SHEXP, "weight", i), { n_embd, n_ff_shexp }, flags);
}
// NextN-specific tensors that define the MTP block.
layer.nextn.eh_proj = create_tensor(tn(LLM_TENSOR_NEXTN_EH_PROJ, "weight", i), { 2 * n_embd, n_embd }, flags);
layer.nextn.enorm = create_tensor(tn(LLM_TENSOR_NEXTN_ENORM, "weight", i), { n_embd }, flags);
layer.nextn.hnorm = create_tensor(tn(LLM_TENSOR_NEXTN_HNORM, "weight", i), { n_embd }, flags);
layer.nextn.embed_tokens = create_tensor(tn(LLM_TENSOR_NEXTN_EMBED_TOKENS, "weight", i), { n_embd, n_vocab }, TENSOR_NOT_REQUIRED);
layer.nextn.shared_head_head = create_tensor(tn(LLM_TENSOR_NEXTN_SHARED_HEAD_HEAD, "weight", i), { n_embd, n_vocab }, TENSOR_NOT_REQUIRED);
layer.nextn.shared_head_norm = create_tensor(tn(LLM_TENSOR_NEXTN_SHARED_HEAD_NORM, "weight", i), { n_embd }, TENSOR_NOT_REQUIRED);
};
for (int i = 0; i < n_layer; ++i) {
load_block_trunk(i, trunk_flags);
}
// MTP/NextN layers are loaded as extra decoder blocks.
for (int i = n_layer; i < n_layer_all; ++i) {
load_block_mtp(i, mtp_flags);
}
}
std::unique_ptr<llm_graph_context> llama_model_cohere2moe::build_arch_graph(const llm_graph_params & params) const {
if (params.gtype == LLM_GRAPH_TYPE_DECODER_MTP) {
return std::make_unique<graph_mtp>(*this, params);
}
return std::make_unique<graph>(*this, params);
}
llama_model_cohere2moe::graph::graph(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_ASSERT(n_embd_head == n_rot);
const llm_norm_type cohere2moe_norm_type = hparams.f_norm_rms_eps == 0.0f ? LLM_NORM : LLM_NORM_RMS;
const float f_logit_scale = hparams.f_logit_scale;
ggml_tensor * cur;
ggml_tensor * inpL = build_inp_embd(model.tok_embd);
ggml_tensor * inp_pos = build_inp_pos();
auto * inp_attn = build_attn_inp_kv_iswa();
ggml_tensor * inp_out_ids = build_inp_out_ids();
// MTP/NextN layers are loaded as extra decoder blocks but not executed in the main pass.
for (int il = 0; il < n_layer; ++il) {
const bool is_swa = hparams.is_swa(il);
// Dense-prefix full-attention layers use RoPE; later layers follow the SWA pattern.
const bool force_rope = static_cast<uint32_t>(il) < hparams.n_layer_dense_lead;
cur = build_norm(inpL, model.layers[il].attn_norm, nullptr, cohere2moe_norm_type, il);
cb(cur, "attn_norm", il);
ggml_tensor * ffn_inp = cur;
{
const auto & layer = model.layers[il];
auto [Qcur, Kcur, Vcur] = build_qkv(layer, cur,
n_embd_head, n_head, n_head_kv, il);
if (is_swa || force_rope) {
ggml_tensor * rope_factors = model.get_rope_factors(cparams, il);
Qcur = ggml_rope_ext(
ctx0, Qcur, inp_pos, rope_factors,
n_rot, rope_type, n_ctx_orig, freq_base, freq_scale,
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, freq_scale,
ext_factor, attn_factor, beta_fast, beta_slow);
}
cb(Qcur, "Qcur", il);
cb(Kcur, "Kcur", il);
cb(Vcur, "Vcur", il);
cur = build_attn(inp_attn,
layer.wo, layer.wo_b, layer.wo_s,
Qcur, Kcur, Vcur, nullptr, nullptr, nullptr,
1.0f / sqrtf(float(n_embd_head)), il);
}
if (il == n_layer - 1 && inp_out_ids && cparams.embeddings_nextn_masked) {
cur = ggml_get_rows(ctx0, cur, inp_out_ids);
inpL = ggml_get_rows(ctx0, inpL, inp_out_ids);
ffn_inp = ggml_get_rows(ctx0, ffn_inp, inp_out_ids);
}
ggml_tensor * attn_out = cur;
const auto & layer = model.layers[il];
if (layer.ffn_gate_inp == nullptr) {
cur = build_ffn(ffn_inp,
layer.ffn_up, nullptr, layer.ffn_up_s,
layer.ffn_gate, nullptr, layer.ffn_gate_s,
layer.ffn_down, nullptr, layer.ffn_down_s,
nullptr, LLM_FFN_SILU, LLM_FFN_PAR, il);
cb(cur, "ffn_out", il);
} else {
cur = build_moe_ffn(ffn_inp,
layer.ffn_gate_inp,
layer.ffn_up_exps,
layer.ffn_gate_exps,
layer.ffn_down_exps,
nullptr,
n_expert, n_expert_used,
LLM_FFN_SILU, hparams.expert_weights_norm,
hparams.expert_weights_scale,
(llama_expert_gating_func_type) hparams.expert_gating_func,
il,
nullptr, layer.ffn_gate_up_exps,
layer.ffn_up_exps_s,
layer.ffn_gate_exps_s,
layer.ffn_down_exps_s);
cb(cur, "ffn_moe_out", il);
if (layer.ffn_up_shexp) {
ggml_tensor * ffn_shexp = build_ffn(ffn_inp,
layer.ffn_up_shexp, nullptr, layer.ffn_up_shexp_s,
layer.ffn_gate_shexp, nullptr, layer.ffn_gate_shexp_s,
layer.ffn_down_shexp, nullptr, layer.ffn_down_shexp_s,
nullptr, LLM_FFN_SILU, LLM_FFN_PAR, il);
cb(ffn_shexp, "ffn_shexp", il);
cur = ggml_add(ctx0, cur, ffn_shexp);
cur = ggml_scale(ctx0, cur, 0.5f);
cb(cur, "ffn_out", il);
}
}
cur = ggml_add(ctx0, cur, inpL);
cur = ggml_add(ctx0, cur, attn_out);
cur = build_cvec(cur, il);
cb(cur, "l_out", il);
inpL = cur;
}
cur = inpL;
cur = build_norm(cur, model.output_norm, nullptr, cohere2moe_norm_type, -1);
cb(cur, "h_nextn", -1);
res->t_h_nextn = cur;
if (!cparams.embeddings_nextn_masked && inp_out_ids) {
cur = ggml_get_rows(ctx0, cur, inp_out_ids);
}
cb(cur, "result_norm", -1);
res->t_embd = cur;
cur = build_lora_mm(model.output, cur);
if (f_logit_scale) {
cur = ggml_scale(ctx0, cur, f_logit_scale);
}
cb(cur, "result_output", -1);
res->t_logits = cur;
ggml_build_forward_expand(gf, cur);
}
llama_model_cohere2moe::graph_mtp::graph_mtp(const llama_model & model, const llm_graph_params & params) : llm_graph_context(params) {
GGML_ASSERT(hparams.n_layer_nextn > 0 && "COHERE2MOE MTP requires n_layer_nextn > 0");
GGML_ASSERT(hparams.n_layer_nextn == 1 && "COHERE2MOE MTP currently only supports a single MTP block");
const int64_t n_embd_head = hparams.n_embd_head_v();
GGML_ASSERT(n_embd_head == hparams.n_embd_head_k());
GGML_ASSERT(n_embd_head == n_rot);
const int il = hparams.n_layer();
const auto & layer = model.layers[il];
GGML_ASSERT(layer.nextn.eh_proj && "MTP block missing nextn.eh_proj");
GGML_ASSERT(layer.nextn.enorm && "MTP block missing nextn.enorm");
GGML_ASSERT(layer.nextn.hnorm && "MTP block missing nextn.hnorm");
GGML_ASSERT(layer.ffn_gate_inp && "MTP block missing ffn_gate_inp");
const llm_norm_type cohere2moe_norm_type = hparams.f_norm_rms_eps == 0.0f ? LLM_NORM : LLM_NORM_RMS;
// TODO: extract in a common llm_graph_context::build_inp_embd_h()
auto inp = std::make_unique<llm_graph_input_embd_h>(hparams.n_embd);
inp->tokens = ggml_new_tensor_1d(ctx0, GGML_TYPE_I32, n_tokens);
ggml_set_input(inp->tokens);
inp->embd = ggml_new_tensor_2d(ctx0, GGML_TYPE_F32, hparams.n_embd_inp(), n_tokens);
ggml_set_input(inp->embd);
// TODO: make static using `ggml_build_forward_select()`
// see llm_graph_context::build_inp_embd() for reference
ggml_tensor * tok_embd;
if (ubatch.token) {
ggml_tensor * tok_embd_w = layer.nextn.embed_tokens ? layer.nextn.embed_tokens : model.tok_embd;
tok_embd = ggml_get_rows(ctx0, tok_embd_w, inp->tokens);
} else {
tok_embd = inp->embd;
}
cb(tok_embd, "mtp_tok_embd", il);
inp->h = ggml_new_tensor_2d(ctx0, GGML_TYPE_F32, hparams.n_embd, n_tokens);
ggml_set_input(inp->h);
ggml_set_name(inp->h, "mtp_h_input");
ggml_tensor * h_embd = inp->h;
res->add_input(std::move(inp));
ggml_tensor * inp_pos = build_inp_pos();
ggml_tensor * inp_out_ids = build_inp_out_ids();
auto * inp_attn = build_attn_inp_kv_iswa();
ggml_tensor * h_norm = build_norm(h_embd, layer.nextn.hnorm, nullptr, cohere2moe_norm_type, il);
cb(h_norm, "mtp_hnorm", il);
ggml_tensor * e_norm = build_norm(tok_embd, layer.nextn.enorm, nullptr, cohere2moe_norm_type, il);
cb(e_norm, "mtp_enorm", il);
ggml_tensor * concat = ggml_concat(ctx0, e_norm, h_norm, /*dim=*/ 0);
cb(concat, "mtp_concat", il);
ggml_tensor * cur = build_lora_mm(layer.nextn.eh_proj, concat, layer.nextn.eh_proj_s);
cb(cur, "mtp_eh_proj", il);
ggml_tensor * inpL = cur;
cur = build_norm(cur, layer.attn_norm, nullptr, cohere2moe_norm_type, il);
cb(cur, "mtp_attn_norm", il);
ggml_tensor * ffn_inp = cur;
auto [Qcur, Kcur, Vcur] = build_qkv(layer, cur, n_embd_head, n_head, n_head_kv, il);
ggml_tensor * rope_factors = model.get_rope_factors(cparams, il);
Qcur = ggml_rope_ext(
ctx0, Qcur, inp_pos, rope_factors,
n_rot, rope_type, n_ctx_orig, freq_base, freq_scale,
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, freq_scale,
ext_factor, attn_factor, beta_fast, beta_slow);
cb(Qcur, "mtp_Qcur", il);
cb(Kcur, "mtp_Kcur", il);
cb(Vcur, "mtp_Vcur", il);
cur = build_attn(inp_attn,
layer.wo, layer.wo_b, layer.wo_s,
Qcur, Kcur, Vcur, nullptr, nullptr, nullptr,
1.0f / sqrtf(float(n_embd_head)), il);
cb(cur, "mtp_attn_out", il);
ggml_tensor * attn_out = cur;
cur = build_moe_ffn(ffn_inp,
layer.ffn_gate_inp,
layer.ffn_up_exps,
layer.ffn_gate_exps,
layer.ffn_down_exps,
nullptr,
n_expert, n_expert_used,
LLM_FFN_SILU, hparams.expert_weights_norm,
hparams.expert_weights_scale,
(llama_expert_gating_func_type) hparams.expert_gating_func,
il,
nullptr, layer.ffn_gate_up_exps,
layer.ffn_up_exps_s,
layer.ffn_gate_exps_s,
layer.ffn_down_exps_s);
cb(cur, "mtp_ffn_moe_out", il);
if (layer.ffn_up_shexp) {
ggml_tensor * ffn_shexp = build_ffn(ffn_inp,
layer.ffn_up_shexp, nullptr, layer.ffn_up_shexp_s,
layer.ffn_gate_shexp, nullptr, layer.ffn_gate_shexp_s,
layer.ffn_down_shexp, nullptr, layer.ffn_down_shexp_s,
nullptr, LLM_FFN_SILU, LLM_FFN_PAR, il);
cb(ffn_shexp, "mtp_ffn_shexp", il);
cur = ggml_add(ctx0, cur, ffn_shexp);
cur = ggml_scale(ctx0, cur, 0.5f);
cb(cur, "mtp_ffn_out", il);
}
cur = ggml_add(ctx0, cur, inpL);
cur = ggml_add(ctx0, cur, attn_out);
cb(cur, "mtp_post_ffn", il);
ggml_tensor * head_norm_w = layer.nextn.shared_head_norm
? layer.nextn.shared_head_norm
: model.output_norm;
GGML_ASSERT(head_norm_w && "COHERE2MOE MTP: missing both nextn.shared_head_norm and output_norm");
cur = build_norm(cur, head_norm_w, nullptr, cohere2moe_norm_type, -1);
cb(cur, "h_nextn", -1);
res->t_h_nextn = cur;
cur = ggml_get_rows(ctx0, cur, inp_out_ids);
cb(cur, "mtp_shared_head_norm", -1);
ggml_tensor * head_w = layer.nextn.shared_head_head ? layer.nextn.shared_head_head : model.output;
GGML_ASSERT(head_w && "COHERE2MOE MTP: missing LM head (nextn.shared_head_head or model.output)");
cur = build_lora_mm(head_w, cur, layer.nextn.shared_head_head ? layer.nextn.shared_head_head_s : nullptr);
if (hparams.f_logit_scale) {
cur = ggml_scale(ctx0, cur, hparams.f_logit_scale);
}
cb(cur, "result_output", -1);
res->t_logits = cur;
ggml_build_forward_expand(gf, cur);
}