Skip to content

model : add TorConsMoe support #16072

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

Closed
wants to merge 1 commit into from
Closed

model : add TorConsMoe support #16072

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
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
18 changes: 18 additions & 0 deletions convert_hf_to_gguf.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -3749,6 +3749,24 @@ def set_vocab(self):
super().set_vocab()


@ModelBase.register("TorconsMoeForCausalLM")
class TorconsMoeModel(Qwen2MoeModel):
model_arch = gguf.MODEL_ARCH.TORCONSMOE

def __init__(self, *args, **kwargs):
super().__init__(*args, **kwargs)
hparams = ModelBase.load_hparams(self.dir_model, False)
self.origin_hf_arch = hparams.get('architectures', [None])[0]

def set_vocab(self):
# deal with intern-s1
if self.origin_hf_arch == 'InternS1ForConditionalGeneration':
self._set_vocab_interns1()
return

super().set_vocab()


@ModelBase.register("GPT2LMHeadModel")
class GPT2Model(TextModel):
model_arch = gguf.MODEL_ARCH.GPT2
Expand Down
19 changes: 19 additions & 0 deletions gguf-py/gguf/constants.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -335,6 +335,7 @@ class MODEL_ARCH(IntEnum):
QWEN2VL = auto()
QWEN3 = auto()
QWEN3MOE = auto()
TORCONSMOE = auto()
PHI2 = auto()
PHI3 = auto()
PHIMOE = auto()
Expand Down Expand Up @@ -671,6 +672,7 @@ class MODEL_TENSOR(IntEnum):
MODEL_ARCH.QWEN2VL: "qwen2vl",
MODEL_ARCH.QWEN3: "qwen3",
MODEL_ARCH.QWEN3MOE: "qwen3moe",
MODEL_ARCH.TORCONSMOE: "torconsmoe",
MODEL_ARCH.PHI2: "phi2",
MODEL_ARCH.PHI3: "phi3",
MODEL_ARCH.PHIMOE: "phimoe",
Expand Down Expand Up @@ -1462,6 +1464,23 @@ class MODEL_TENSOR(IntEnum):
MODEL_TENSOR.FFN_DOWN_EXP,
MODEL_TENSOR.FFN_UP_EXP,
],
MODEL_ARCH.TORCONSMOE: [
MODEL_TENSOR.TOKEN_EMBD,
MODEL_TENSOR.OUTPUT_NORM,
MODEL_TENSOR.OUTPUT,
MODEL_TENSOR.ATTN_NORM,
MODEL_TENSOR.ATTN_Q,
MODEL_TENSOR.ATTN_Q_NORM,
MODEL_TENSOR.ATTN_K,
MODEL_TENSOR.ATTN_K_NORM,
MODEL_TENSOR.ATTN_V,
MODEL_TENSOR.ATTN_OUT,
MODEL_TENSOR.FFN_NORM,
MODEL_TENSOR.FFN_GATE_INP,
MODEL_TENSOR.FFN_GATE_EXP,
MODEL_TENSOR.FFN_DOWN_EXP,
MODEL_TENSOR.FFN_UP_EXP,
],
MODEL_ARCH.PLAMO: [
MODEL_TENSOR.TOKEN_EMBD,
MODEL_TENSOR.OUTPUT_NORM,
Expand Down
21 changes: 21 additions & 0 deletions src/llama-arch.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -31,6 +31,7 @@ static const std::map<llm_arch, const char *> LLM_ARCH_NAMES = {
{ LLM_ARCH_QWEN2VL, "qwen2vl" },
{ LLM_ARCH_QWEN3, "qwen3" },
{ LLM_ARCH_QWEN3MOE, "qwen3moe" },
{ LLM_ARCH_TORCONSMOE, "torconsmoe" },
{ LLM_ARCH_PHI2, "phi2" },
{ LLM_ARCH_PHI3, "phi3" },
{ LLM_ARCH_PHIMOE, "phimoe" },
Expand Down Expand Up @@ -754,6 +755,26 @@ static const std::map<llm_arch, std::map<llm_tensor, const char *>> LLM_TENSOR_N
{ LLM_TENSOR_FFN_UP_EXPS, "blk.%d.ffn_up_exps" },
},
},
{
LLM_ARCH_TORCONSMOE,
{
{ LLM_TENSOR_TOKEN_EMBD, "token_embd" },
{ LLM_TENSOR_OUTPUT_NORM, "output_norm" },
{ LLM_TENSOR_OUTPUT, "output" },
{ LLM_TENSOR_ATTN_NORM, "blk.%d.attn_norm" },
{ LLM_TENSOR_ATTN_Q, "blk.%d.attn_q" },
{ LLM_TENSOR_ATTN_Q_NORM, "blk.%d.attn_q_norm" },
{ LLM_TENSOR_ATTN_K, "blk.%d.attn_k" },
{ LLM_TENSOR_ATTN_K_NORM, "blk.%d.attn_k_norm" },
{ LLM_TENSOR_ATTN_V, "blk.%d.attn_v" },
{ LLM_TENSOR_ATTN_OUT, "blk.%d.attn_output" },
{ LLM_TENSOR_FFN_NORM, "blk.%d.ffn_norm" },
{ LLM_TENSOR_FFN_GATE_INP, "blk.%d.ffn_gate_inp" },
{ LLM_TENSOR_FFN_GATE_EXPS, "blk.%d.ffn_gate_exps" },
{ LLM_TENSOR_FFN_DOWN_EXPS, "blk.%d.ffn_down_exps" },
{ LLM_TENSOR_FFN_UP_EXPS, "blk.%d.ffn_up_exps" },
},
},
{
LLM_ARCH_PHI2,
{
Expand Down
1 change: 1 addition & 0 deletions src/llama-arch.h
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -35,6 +35,7 @@ enum llm_arch {
LLM_ARCH_QWEN2VL,
LLM_ARCH_QWEN3,
LLM_ARCH_QWEN3MOE,
LLM_ARCH_TORCONSMOE,
LLM_ARCH_PHI2,
LLM_ARCH_PHI3,
LLM_ARCH_PHIMOE,
Expand Down
192 changes: 192 additions & 0 deletions src/llama-model.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -999,6 +999,17 @@ void llama_model::load_hparams(llama_model_loader & ml) {
default: type = LLM_TYPE_UNKNOWN;
}
} break;
case LLM_ARCH_TORCONSMOE:
{
ml.get_key(LLM_KV_EXPERT_FEED_FORWARD_LENGTH, hparams.n_ff_exp, false);

ml.get_key(LLM_KV_ATTENTION_LAYERNORM_RMS_EPS, hparams.f_norm_rms_eps);
switch (hparams.n_layer) {
case 48: type = LLM_TYPE_30B_A3B; break;
case 94: type = LLM_TYPE_235B_A22B; break;
default: type = LLM_TYPE_UNKNOWN;
}
} break;
case LLM_ARCH_PHI2:
{
ml.get_key(LLM_KV_ATTENTION_LAYERNORM_EPS, hparams.f_norm_eps);
Expand Down Expand Up @@ -3223,6 +3234,50 @@ bool llama_model::load_tensors(llama_model_loader & ml) {
// MoE branch
const int64_t n_ff_exp = hparams.n_ff_exp ? hparams.n_ff_exp : n_ff / n_expert_used;

layer.ffn_gate_exps = create_tensor(tn(LLM_TENSOR_FFN_GATE_EXPS, "weight", i), { n_embd, n_ff_exp, n_expert}, 0);
layer.ffn_down_exps = create_tensor(tn(LLM_TENSOR_FFN_DOWN_EXPS, "weight", i), {n_ff_exp, n_embd, n_expert}, 0);
layer.ffn_up_exps = create_tensor(tn(LLM_TENSOR_FFN_UP_EXPS, "weight", i), { n_embd, n_ff_exp, n_expert}, 0);
}
} break;
case LLM_ARCH_TORCONSMOE:
{
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);
}

for (int i = 0; i < n_layer; ++i) {
auto & layer = layers[i];

layer.attn_norm = create_tensor(tn(LLM_TENSOR_ATTN_NORM, "weight", i), {n_embd}, 0);

layer.wq = create_tensor(tn(LLM_TENSOR_ATTN_Q, "weight", i), {n_embd, n_embd_head_k * n_head}, 0);
layer.wk = create_tensor(tn(LLM_TENSOR_ATTN_K, "weight", i), {n_embd, n_embd_gqa}, 0);
layer.wv = create_tensor(tn(LLM_TENSOR_ATTN_V, "weight", i), {n_embd, n_embd_gqa}, 0);
layer.wo = create_tensor(tn(LLM_TENSOR_ATTN_OUT, "weight", i), {n_embd_head_k * n_head, n_embd}, 0);

layer.attn_k_norm = create_tensor(tn(LLM_TENSOR_ATTN_K_NORM, "weight", i), {n_embd_head_k}, 0);
layer.attn_q_norm = create_tensor(tn(LLM_TENSOR_ATTN_Q_NORM, "weight", i), {n_embd_head_k}, 0);

layer.ffn_norm = create_tensor(tn(LLM_TENSOR_FFN_NORM, "weight", i), {n_embd}, 0);

layer.ffn_gate_inp = create_tensor(tn(LLM_TENSOR_FFN_GATE_INP, "weight", i), {n_embd, n_expert}, 0);

if (n_expert == 0) {
throw std::runtime_error("n_expert must be > 0 for TORCONSMOE");
}
if (n_expert_used == 0) {
throw std::runtime_error("n_expert_used must be > 0 for TORCONSMOE");
}

// MoE branch
const int64_t n_ff_exp = hparams.n_ff_exp ? hparams.n_ff_exp : n_ff / n_expert_used;

layer.ffn_gate_exps = create_tensor(tn(LLM_TENSOR_FFN_GATE_EXPS, "weight", i), { n_embd, n_ff_exp, n_expert}, 0);
layer.ffn_down_exps = create_tensor(tn(LLM_TENSOR_FFN_DOWN_EXPS, "weight", i), {n_ff_exp, n_embd, n_expert}, 0);
layer.ffn_up_exps = create_tensor(tn(LLM_TENSOR_FFN_UP_EXPS, "weight", i), { n_embd, n_ff_exp, n_expert}, 0);
Expand Down Expand Up @@ -6143,6 +6198,10 @@ void llama_model::print_info() const {
LLAMA_LOG_INFO("%s: n_ff_exp = %d\n", __func__, hparams.n_ff_exp);
}

if (arch == LLM_ARCH_TORCONSMOE || arch == LLM_ARCH_OPENAI_MOE) {
LLAMA_LOG_INFO("%s: n_ff_exp = %d\n", __func__, hparams.n_ff_exp);
}

if (arch == LLM_ARCH_MINICPM ||
arch == LLM_ARCH_GRANITE ||
arch == LLM_ARCH_GRANITE_MOE ||
Expand Down Expand Up @@ -9276,6 +9335,134 @@ struct llm_build_qwen3moe : public llm_graph_context {
}
};

struct llm_build_torconsmoe : public llm_graph_context {
llm_build_torconsmoe(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 == hparams.n_rot);

ggml_tensor * cur;
ggml_tensor * inpL;

inpL = build_inp_embd(model.tok_embd);

// inp_pos - contains the positions
ggml_tensor * inp_pos = build_inp_pos();

auto * inp_attn = build_attn_inp_kv();

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, NULL,
LLM_NORM_RMS, il);
cb(cur, "attn_norm", il);

// self_attention
{
// compute Q and K and RoPE them
ggml_tensor * Qcur = build_lora_mm(model.layers[il].wq, cur);
cb(Qcur, "Qcur", il);

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

ggml_tensor * Vcur = build_lora_mm(model.layers[il].wv, 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);

Qcur = build_norm(Qcur, model.layers[il].attn_q_norm, NULL, LLM_NORM_RMS, il);
cb(Qcur, "Qcur_normed", il);

Qcur = ggml_rope_ext(
ctx0, Qcur, inp_pos, nullptr,
n_rot, rope_type, n_ctx_orig, freq_base, freq_scale,
ext_factor, attn_factor, beta_fast, beta_slow
);

Kcur = build_norm(Kcur, model.layers[il].attn_k_norm, NULL, LLM_NORM_RMS, il);
cb(Kcur, "Kcur_normed", il);

Kcur = ggml_rope_ext(
ctx0, Kcur, inp_pos, nullptr,
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,
model.layers[il].wo, model.layers[il].bo,
Qcur, Kcur, Vcur, nullptr, nullptr, nullptr, 1.0f/sqrtf(float(n_embd_head)), 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);

// MoE branch
cur = build_norm(ffn_inp,
model.layers[il].ffn_norm, NULL,
LLM_NORM_RMS, il);
cb(cur, "ffn_norm", il);

ggml_tensor * moe_out =
build_moe_ffn(cur,
model.layers[il].ffn_gate_inp,
model.layers[il].ffn_up_exps,
model.layers[il].ffn_gate_exps,
model.layers[il].ffn_down_exps,
nullptr,
n_expert, n_expert_used,
LLM_FFN_SILU, true,
false, 0.0,
LLAMA_EXPERT_GATING_FUNC_TYPE_SOFTMAX,
il);
cb(moe_out, "ffn_moe_out", il);
cur = moe_out;

cur = ggml_add(ctx0, cur, ffn_inp);

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

// input for next layer
inpL = cur;
}

cur = inpL;

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

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

// lm_head
cur = build_lora_mm(model.output, cur);

cb(cur, "result_output", -1);
res->t_logits = cur;

ggml_build_forward_expand(gf, cur);
}
};

struct llm_build_phi2 : public llm_graph_context {
llm_build_phi2(const llama_model & model, const llm_graph_params & params) : llm_graph_context(params) {
const int64_t n_embd_head = hparams.n_embd_head_v;
Expand Down Expand Up @@ -19098,6 +19285,10 @@ ggml_cgraph * llama_model::build_graph(const llm_graph_params & params) const {
{
llm = std::make_unique<llm_build_qwen3moe>(*this, params);
} break;
case LLM_ARCH_TORCONSMOE:
{
llm = std::make_unique<llm_build_torconsmoe>(*this, params);
} break;
case LLM_ARCH_PHI2:
{
llm = std::make_unique<llm_build_phi2>(*this, params);
Expand Down Expand Up @@ -19552,6 +19743,7 @@ llama_rope_type llama_model_rope_type(const llama_model * model) {
case LLM_ARCH_QWEN2MOE:
case LLM_ARCH_QWEN3:
case LLM_ARCH_QWEN3MOE:
case LLM_ARCH_TORCONSMOE:
case LLM_ARCH_LLADA_MOE:
case LLM_ARCH_OLMO2:
case LLM_ARCH_OLMOE:
Expand Down