mirror of
https://github.com/ggml-org/llama.cpp.git
synced 2025-06-27 20:05:20 +00:00
model : Nomic Embed Text V2 with Mixture-of-Experts (MoE) architecture (#12466)
* Nomic Embed Text V2 with Mixture-of-Experts (MoE) architecture - Adds MoE-based embedding model supporting multilingual embeddings. - Selects architecture variant based on hyperparameter detection (MoE layers). - Removes unnecessary subclass initialization checks for clarity. https://www.nomic.ai/blog/posts/nomic-embed-text-v2 Co-authored-by: Jared Van Bortel <jared@nomic.ai> * fix tokenizer * don't rename this tensor --------- Co-authored-by: Jared Van Bortel <jared@nomic.ai>
This commit is contained in:
AT
GitHub
@ -695,10 +695,12 @@ void llama_model::load_hparams(llama_model_loader & ml) {
|
||||
}
|
||||
} break;
|
||||
case LLM_ARCH_NOMIC_BERT:
|
||||
case LLM_ARCH_NOMIC_BERT_MOE:
|
||||
{
|
||||
ml.get_key(LLM_KV_ATTENTION_LAYERNORM_EPS, hparams.f_norm_eps);
|
||||
ml.get_key(LLM_KV_ATTENTION_CAUSAL, hparams.causal_attn);
|
||||
ml.get_key(LLM_KV_POOLING_TYPE, hparams.pooling_type);
|
||||
ml.get_key(LLM_KV_MOE_EVERY_N_LAYERS, hparams.moe_every_n_layers, 0);
|
||||
|
||||
if (hparams.n_layer == 12 && hparams.n_embd == 768) {
|
||||
type = LLM_TYPE_137M;
|
||||
@ -2057,6 +2059,7 @@ bool llama_model::load_tensors(llama_model_loader & ml) {
|
||||
} break;
|
||||
case LLM_ARCH_BERT:
|
||||
case LLM_ARCH_NOMIC_BERT:
|
||||
case LLM_ARCH_NOMIC_BERT_MOE:
|
||||
{
|
||||
tok_embd = create_tensor(tn(LLM_TENSOR_TOKEN_EMBD, "weight"), {n_embd, n_vocab}, 0);
|
||||
type_embd = create_tensor(tn(LLM_TENSOR_TOKEN_TYPES, "weight"), {n_embd, n_token_types}, 0);
|
||||
@ -2090,20 +2093,31 @@ bool llama_model::load_tensors(llama_model_loader & ml) {
|
||||
layer.wqkv = create_tensor(tn(LLM_TENSOR_ATTN_QKV, "weight", i), {n_embd, n_embd + 2*n_embd_gqa}, 0);
|
||||
}
|
||||
|
||||
if (arch == LLM_ARCH_NOMIC_BERT_MOE) {
|
||||
layer.bqkv = create_tensor(tn(LLM_TENSOR_ATTN_QKV, "bias", i), {n_embd + 2*n_embd_gqa}, 0);
|
||||
}
|
||||
|
||||
layer.wo = create_tensor(tn(LLM_TENSOR_ATTN_OUT, "weight", i), {n_embd, n_embd}, 0);
|
||||
|
||||
layer.attn_out_norm = create_tensor(tn(LLM_TENSOR_ATTN_OUT_NORM, "weight", i), {n_embd}, 0);
|
||||
layer.attn_out_norm_b = create_tensor(tn(LLM_TENSOR_ATTN_OUT_NORM, "bias", i), {n_embd}, 0);
|
||||
|
||||
layer.ffn_up = create_tensor(tn(LLM_TENSOR_FFN_UP, "weight", i), {n_embd, n_ff}, 0);
|
||||
layer.ffn_down = create_tensor(tn(LLM_TENSOR_FFN_DOWN, "weight", i), {n_ff, n_embd}, 0);
|
||||
|
||||
if (arch == LLM_ARCH_BERT) {
|
||||
if (hparams.moe_every_n_layers > 0 && i % hparams.moe_every_n_layers == 1) {
|
||||
layer.bo = create_tensor(tn(LLM_TENSOR_ATTN_OUT, "bias", i), {n_embd}, 0);
|
||||
layer.ffn_up_b = create_tensor(tn(LLM_TENSOR_FFN_UP, "bias", i), {n_ff}, 0);
|
||||
layer.ffn_down_b = create_tensor(tn(LLM_TENSOR_FFN_DOWN, "bias", i), {n_embd}, 0);
|
||||
layer.ffn_up_exps = create_tensor(tn(LLM_TENSOR_FFN_UP_EXPS, "weight", i), { n_embd, n_ff, n_expert}, 0);
|
||||
layer.ffn_down_exps = create_tensor(tn(LLM_TENSOR_FFN_DOWN_EXPS, "weight", i), { n_ff, n_embd, n_expert}, 0);
|
||||
layer.ffn_gate_inp = create_tensor(tn(LLM_TENSOR_FFN_GATE_INP, "weight", i), {n_embd, n_expert}, 0);
|
||||
} else {
|
||||
layer.ffn_gate = create_tensor(tn(LLM_TENSOR_FFN_GATE, "weight", i), {n_embd, n_ff}, 0);
|
||||
layer.ffn_up = create_tensor(tn(LLM_TENSOR_FFN_UP, "weight", i), {n_embd, n_ff}, 0);
|
||||
layer.ffn_down = create_tensor(tn(LLM_TENSOR_FFN_DOWN, "weight", i), {n_ff, n_embd}, 0);
|
||||
|
||||
if (arch == LLM_ARCH_BERT || arch == LLM_ARCH_NOMIC_BERT_MOE) {
|
||||
layer.bo = create_tensor(tn(LLM_TENSOR_ATTN_OUT, "bias", i), {n_embd}, 0);
|
||||
layer.ffn_up_b = create_tensor(tn(LLM_TENSOR_FFN_UP, "bias", i), {n_ff}, 0);
|
||||
layer.ffn_down_b = create_tensor(tn(LLM_TENSOR_FFN_DOWN, "bias", i), {n_embd}, 0);
|
||||
} else {
|
||||
layer.ffn_gate = create_tensor(tn(LLM_TENSOR_FFN_GATE, "weight", i), {n_embd, n_ff}, 0);
|
||||
}
|
||||
}
|
||||
|
||||
layer.layer_out_norm = create_tensor(tn(LLM_TENSOR_LAYER_OUT_NORM, "weight", i), {n_embd}, 0);
|
||||
@ -5730,6 +5744,11 @@ struct llm_build_bert : public llm_graph_context {
|
||||
cur = build_lora_mm(model.layers[il].wqkv, cur);
|
||||
cb(cur, "wqkv", il);
|
||||
|
||||
if (model.arch == LLM_ARCH_NOMIC_BERT_MOE) {
|
||||
cur = ggml_add(ctx0, cur, model.layers[il].bqkv);
|
||||
cb(cur, "bqkv", il);
|
||||
}
|
||||
|
||||
Qcur = ggml_cont(ctx0, ggml_view_2d(ctx0, cur, n_embd, n_tokens, cur->nb[1], 0*sizeof(float)*(n_embd)));
|
||||
Kcur = ggml_cont(ctx0, ggml_view_2d(ctx0, cur, n_embd_gqa, n_tokens, cur->nb[1], 1*sizeof(float)*(n_embd)));
|
||||
Vcur = ggml_cont(ctx0, ggml_view_2d(ctx0, cur, n_embd_gqa, n_tokens, cur->nb[1], 1*sizeof(float)*(n_embd + n_embd_gqa)));
|
||||
@ -5782,13 +5801,29 @@ struct llm_build_bert : public llm_graph_context {
|
||||
cb(ffn_inp, "ffn_inp", il);
|
||||
|
||||
// feed-forward network
|
||||
if (model.arch == LLM_ARCH_BERT) {
|
||||
if (hparams.moe_every_n_layers > 0 && il % hparams.moe_every_n_layers == 1) {
|
||||
// MoE branch
|
||||
cur = build_moe_ffn(cur,
|
||||
model.layers[il].ffn_gate_inp,
|
||||
model.layers[il].ffn_up_exps,
|
||||
nullptr,
|
||||
model.layers[il].ffn_down_exps,
|
||||
nullptr,
|
||||
hparams.n_expert,
|
||||
hparams.n_expert_used,
|
||||
LLM_FFN_GELU,
|
||||
false, false,
|
||||
0.0f,
|
||||
LLAMA_EXPERT_GATING_FUNC_TYPE_SOFTMAX, il);
|
||||
cb(cur, "ffn_moe_out", il);
|
||||
} else if (model.arch == LLM_ARCH_BERT || model.arch == LLM_ARCH_NOMIC_BERT_MOE) {
|
||||
cur = build_ffn(cur,
|
||||
model.layers[il].ffn_up, model.layers[il].ffn_up_b, NULL,
|
||||
NULL, NULL, NULL,
|
||||
model.layers[il].ffn_down, model.layers[il].ffn_down_b, NULL,
|
||||
NULL,
|
||||
LLM_FFN_GELU, LLM_FFN_SEQ, il);
|
||||
cb(cur, "ffn_out", il);
|
||||
} else if (model.arch == LLM_ARCH_JINA_BERT_V2) {
|
||||
cur = build_ffn(cur,
|
||||
model.layers[il].ffn_up, NULL, NULL,
|
||||
@ -5796,6 +5831,7 @@ struct llm_build_bert : public llm_graph_context {
|
||||
model.layers[il].ffn_down, model.layers[il].ffn_down_b, NULL,
|
||||
NULL,
|
||||
LLM_FFN_GELU, LLM_FFN_PAR, il);
|
||||
cb(cur, "ffn_out", il);
|
||||
} else {
|
||||
cur = build_ffn(cur,
|
||||
model.layers[il].ffn_up, NULL, NULL,
|
||||
@ -5803,8 +5839,8 @@ struct llm_build_bert : public llm_graph_context {
|
||||
model.layers[il].ffn_down, NULL, NULL,
|
||||
NULL,
|
||||
LLM_FFN_SILU, LLM_FFN_PAR, il);
|
||||
cb(cur, "ffn_out", il);
|
||||
}
|
||||
cb(cur, "ffn_out", il);
|
||||
|
||||
// attentions bypass the intermediate layer
|
||||
cur = ggml_add(ctx0, cur, ffn_inp);
|
||||
@ -12843,6 +12879,7 @@ llm_graph_result_ptr llama_model::build_graph(
|
||||
case LLM_ARCH_BERT:
|
||||
case LLM_ARCH_JINA_BERT_V2:
|
||||
case LLM_ARCH_NOMIC_BERT:
|
||||
case LLM_ARCH_NOMIC_BERT_MOE:
|
||||
{
|
||||
llm = std::make_unique<llm_build_bert>(*this, params, gf);
|
||||
} break;
|
||||
@ -13201,6 +13238,7 @@ llama_rope_type llama_model_rope_type(const llama_model * model) {
|
||||
case LLM_ARCH_DBRX:
|
||||
case LLM_ARCH_BERT:
|
||||
case LLM_ARCH_NOMIC_BERT:
|
||||
case LLM_ARCH_NOMIC_BERT_MOE:
|
||||
case LLM_ARCH_STABLELM:
|
||||
case LLM_ARCH_BITNET:
|
||||
case LLM_ARCH_QWEN:
|
||||
|
||||
Reference in New Issue
Block a user