happyz
/
llama.cpp
mirror of https://github.com/ggerganov/llama.cpp.git
1
0
Fork 0
Code Issues Packages Projects Releases Wiki Activity

Merge a1523db199 into 58062860af

This commit is contained in:
Ryan Mangeno 2025-12-17 05:51:06 +02:00 committed by GitHub
parent 58062860af a1523db199
commit 0c92c48b3a
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
15 changed files with 306 additions and 1 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

43
convert_hf_to_gguf.py
View File

@ -1200,6 +1200,9 @@ class TextModel(ModelBase):
if chkhsh == "a1e163ecab2e718a4c829d1148b6e86824ec36163bb71941c3dca9cd5ac25756":
# ref: https://huggingface.co/JetBrains/Mellum-4b-base
res = "mellum"
if chkhsh == "a0b64b4385f123663873756336c085744376d015ff328bb1d901598f63c44152":
# ref: https://huggingface.co/answerdotai/ModernBERT-base
res = "modern-bert"
if chkhsh == "49fc0303c9e0d2c2c565c510f64b2d9b271276acdcdadff733249eda9f7d59df":
# ref: https://huggingface.co/arcee-ai/Trinity-Tokenizer
res = "afmoe"
@ -9906,6 +9909,46 @@ class SmallThinkerModel(TextModel):
raise ValueError(f"Unprocessed experts: {experts}")
@ModelBase.register("ModernBertModel", "ModernBertForMaskedLM", "ModernBertForSequenceClassification")
class ModernBertModel(BertModel):
model_arch = gguf.MODEL_ARCH.MODERN_BERT
def set_vocab(self):
self.gguf_writer.add_add_bos_token(True)
self.gguf_writer.add_add_eos_token(True)
self.gguf_writer.add_add_sep_token(True)
self._set_vocab_gpt2()
def set_gguf_parameters(self):
super().set_gguf_parameters()
self.gguf_writer.add_dense_every_n_layers(self.hparams["global_attn_every_n_layers"])
self.gguf_writer.add_sliding_window(self.hparams["local_attention"])
self.gguf_writer.add_rope_freq_base(self.hparams["global_rope_theta"])
self.gguf_writer.add_rope_freq_base_swa(self.hparams["local_rope_theta"])
self.gguf_writer.add_rope_scaling_type(gguf.RopeScalingType.NONE)
self.gguf_writer.add_vocab_size(self.hparams["vocab_size"])
def modify_tensors(self, data_torch: Tensor, name: str, bid: int | None) -> Iterable[tuple[str, Tensor]]:
# rename custom "head" layers to standard bert "cls.predictions" names for compatibility
if name == "head.norm.weight":
name = "cls.predictions.transform.LayerNorm.weight"
elif name == "head.norm.bias":
name = "cls.predictions.transform.LayerNorm.bias"
elif name == "head.dense.weight":
name = "cls.predictions.transform.dense.weight"
elif name == "head.dense.bias":
name = "cls.predictions.transform.dense.bias"
# These layers act as MLM head, so we don't need them
if name.startswith("decoder."):
return []
if name.startswith("model."):
name = name[6:]
return super().modify_tensors(data_torch, name, bid)
@ModelBase.register("ApertusForCausalLM")
class ApertusModel(LlamaModel):
model_arch = gguf.MODEL_ARCH.APERTUS

1
convert_hf_to_gguf_update.py
View File

@ -139,6 +139,7 @@ models = [
{"name": "lfm2", "tokt": TOKENIZER_TYPE.BPE, "repo": "https://huggingface.co/LiquidAI/LFM2-Tokenizer"},
{"name": "exaone4", "tokt": TOKENIZER_TYPE.BPE, "repo": "https://huggingface.co/LGAI-EXAONE/EXAONE-4.0-32B", },
{"name": "mellum", "tokt": TOKENIZER_TYPE.BPE, "repo": "https://huggingface.co/JetBrains/Mellum-4b-base", },
{"name": "modern-bert", "tokt": TOKENIZER_TYPE.BPE, "repo": "https://huggingface.co/answerdotai/ModernBERT-base", },
{"name": "afmoe", "tokt": TOKENIZER_TYPE.BPE, "repo": "https://huggingface.co/arcee-ai/Trinity-Tokenizer", },
{"name": "bailingmoe2", "tokt": TOKENIZER_TYPE.BPE, "repo": "https://huggingface.co/inclusionAI/Ling-mini-base-2.0", },
{"name": "granite-docling", "tokt": TOKENIZER_TYPE.BPE, "repo": "https://huggingface.co/ibm-granite/granite-docling-258M", },

18
gguf-py/gguf/constants.py
View File

@ -176,11 +176,13 @@ class Keys:
SHARED_KV_LAYERS = "{arch}.attention.shared_kv_layers"
SLIDING_WINDOW_PATTERN = "{arch}.attention.sliding_window_pattern"
TEMPERATURE_SCALE = "{arch}.attention.temperature_scale"
DENSE_EVERY_N_LAYERS = "{arch}.attention.dense_every_n_layers"
class Rope:
DIMENSION_COUNT = "{arch}.rope.dimension_count"
DIMENSION_SECTIONS = "{arch}.rope.dimension_sections"
FREQ_BASE = "{arch}.rope.freq_base"
FREQ_BASE_SWA = "{arch}.rope.freq_base_swa"
SCALING_TYPE = "{arch}.rope.scaling.type"
SCALING_FACTOR = "{arch}.rope.scaling.factor"
SCALING_ATTN_FACTOR = "{arch}.rope.scaling.attn_factor"
@ -354,6 +356,7 @@ class MODEL_ARCH(IntEnum):
STARCODER = auto()
REFACT = auto()
BERT = auto()
MODERN_BERT = auto()
NOMIC_BERT = auto()
NOMIC_BERT_MOE = auto()
NEO_BERT = auto()
@ -730,6 +733,7 @@ MODEL_ARCH_NAMES: dict[MODEL_ARCH, str] = {
MODEL_ARCH.STARCODER: "starcoder",
MODEL_ARCH.REFACT: "refact",
MODEL_ARCH.BERT: "bert",
MODEL_ARCH.MODERN_BERT: "modern-bert",
MODEL_ARCH.NOMIC_BERT: "nomic-bert",
MODEL_ARCH.NOMIC_BERT_MOE: "nomic-bert-moe",
MODEL_ARCH.NEO_BERT: "neo-bert",
@ -1316,6 +1320,20 @@ MODEL_TENSORS: dict[MODEL_ARCH, list[MODEL_TENSOR]] = {
MODEL_TENSOR.CLS,
MODEL_TENSOR.CLS_OUT,
],
MODEL_ARCH.MODERN_BERT: [
MODEL_TENSOR.TOKEN_EMBD,
MODEL_TENSOR.TOKEN_EMBD_NORM,
MODEL_TENSOR.OUTPUT_NORM,
MODEL_TENSOR.ATTN_NORM,
MODEL_TENSOR.ATTN_OUT,
MODEL_TENSOR.ATTN_QKV,
MODEL_TENSOR.POS_EMBD,
MODEL_TENSOR.FFN_UP,
MODEL_TENSOR.FFN_DOWN,
MODEL_TENSOR.FFN_NORM,
MODEL_TENSOR.CLS,
MODEL_TENSOR.CLS_OUT,
],
MODEL_ARCH.NOMIC_BERT: [
MODEL_TENSOR.TOKEN_EMBD,
MODEL_TENSOR.TOKEN_EMBD_NORM,

6
gguf-py/gguf/gguf_writer.py
View File

@ -777,6 +777,9 @@ class GGUFWriter:
def add_sliding_window_pattern(self, value: Sequence[bool]) -> None:
self.add_array(Keys.Attention.SLIDING_WINDOW_PATTERN.format(arch=self.arch), value)
def add_dense_every_n_layers(self, value: int) -> None:
self.add_uint32(Keys.Attention.DENSE_EVERY_N_LAYERS.format(arch=self.arch), value)
def add_dense_features_dims(self, dense:str, in_f:int, out_f:int) -> None:
self.add_uint32(Keys.LLM.DENSE_FEAT_IN_SIZE.format(arch=self.arch, dense=dense), in_f)
self.add_uint32(Keys.LLM.DENSE_FEAT_OUT_SIZE.format(arch=self.arch, dense=dense), out_f)
@ -886,6 +889,9 @@ class GGUFWriter:
def add_value_residual_mix_lora_rank(self, length: int) -> None:
self.add_uint32(Keys.Attention.VALUE_RESIDUAL_MIX_LORA_RANK.format(arch=self.arch), length)
def add_rope_freq_base_swa(self, value: float) -> None:
self.add_float32(Keys.Rope.FREQ_BASE_SWA.format(arch=self.arch), value)
def add_gate_lora_rank(self, length: int) -> None:
self.add_uint32(Keys.Attention.GATE_LORA_RANK.format(arch=self.arch), length)

9
gguf-py/gguf/tensor_mapping.py
View File

@ -17,6 +17,7 @@ class TensorNameMap:
"embed_tokens", # embeddinggemma
"tok_embeddings", # llama-pth
"embeddings.word_embeddings", # bert nomic-bert
"embeddings.tok_embeddings", # modern-bert
"language_model.embedding.word_embeddings", # persimmon
"wte", # gpt2
"transformer.embd.wte", # phi2
@ -46,6 +47,7 @@ class TensorNameMap:
MODEL_TENSOR.TOKEN_EMBD_NORM: (
"word_embeddings_layernorm", # bloom
"embeddings.LayerNorm", # bert
"embeddings.norm", # modern-bert
"emb_ln", # nomic-bert
"transformer.norm", # openelm
"rwkv.blocks.0.pre_ln", # rwkv
@ -104,6 +106,7 @@ class TensorNameMap:
"backbone.final_layer_norm", # wavtokenizer
"model.norm", # llama4
"model.transformer.ln_f", # llada
"final_norm", # modern-bert
"model.norm", # cogvlm
),
@ -151,6 +154,7 @@ class TensorNameMap:
"model.layers.{bid}.input_layernorm", # llama4
"layers.{bid}.input_layernorm", # embeddinggemma
"transformer_encoder.{bid}.attention_norm", # neobert
"layers.{bid}.attn_norm", # modern-bert
"model.layers.{bid}.operator_norm", # lfm2
"model.transformer.blocks.{bid}.attn_norm", # llada
"layers.{bid}.input_layernorm", # qwen3-embedding
@ -187,6 +191,7 @@ class TensorNameMap:
"encoder.layers.{bid}.self_attention.query_key_value", # chatglm
"transformer.layers.{bid}.attn.qkv_proj", # openelm
"transformer_encoder.{bid}.qkv", # neobert
"layers.{bid}.attn.Wqkv", # modern-bert
"model.layers.{bid}.self_attn.language_expert_query_key_value", # cogvlm
),
@ -261,6 +266,7 @@ class TensorNameMap:
"model.layers.{bid}.self_attn.linear_attn", # deci
"layers.{bid}.attention.wo", # llama-pth
"encoder.layer.{bid}.attention.output.dense", # bert
"layers.{bid}.attn.Wo", # modern-bert
"transformer.layer.{bid}.attention.out_lin", # distillbert
"transformer.h.{bid}.attn.out_proj", # gpt-j
"language_model.encoder.layers.{bid}.self_attention.dense", # persimmon
@ -344,6 +350,7 @@ class TensorNameMap:
"layers.{bid}.post_attention_layernorm", # qwen3-embedding
"model.layers.{bid}.feedforward_layernorm", # apertus
"model.layers.{bid}.pre_mlp_layernorm", # kormo
"layers.{bid}.mlp_norm" # modern-bert
),
# Pre feed-forward norm
@ -407,6 +414,7 @@ class TensorNameMap:
"layers.{bid}.mlp.up_proj", # embeddinggemma
"layers.{bid}.feed_forward.w3", # llama-pth
"encoder.layer.{bid}.intermediate.dense", # bert
"layers.{bid}.mlp.Wi", # modern-bert
"transformer.layer.{bid}.ffn.lin1", # distillbert
"transformer.h.{bid}.mlp.fc_in", # gpt-j
"transformer.h.{bid}.mlp.linear_3", # refact
@ -521,6 +529,7 @@ class TensorNameMap:
"layers.{bid}.mlp.down_proj", # embeddinggemma
"layers.{bid}.feed_forward.w2", # llama-pth
"encoder.layer.{bid}.output.dense", # bert
"layers.{bid}.mlp.Wo", # modern-bert
"transformer.layer.{bid}.ffn.lin2", # distillbert
"transformer.h.{bid}.mlp.fc_out", # gpt-j
"language_model.encoder.layers.{bid}.mlp.dense_4h_to_h", # persimmon

1
src/CMakeLists.txt
View File

@ -90,6 +90,7 @@ add_library(llama
models/mamba.cpp
models/minicpm3.cpp
models/minimax-m2.cpp
models/modern-bert.cpp
models/mpt.cpp
models/nemotron-h.cpp
models/nemotron.cpp

18
src/llama-arch.cpp
View File

@ -20,6 +20,7 @@ static const std::map<llm_arch, const char *> LLM_ARCH_NAMES = {
{ LLM_ARCH_STARCODER, "starcoder" },
{ LLM_ARCH_REFACT, "refact" },
{ LLM_ARCH_BERT, "bert" },
{ LLM_ARCH_MODERN_BERT, "modern-bert" },
{ LLM_ARCH_NOMIC_BERT, "nomic-bert" },
{ LLM_ARCH_NOMIC_BERT_MOE, "nomic-bert-moe" },
{ LLM_ARCH_NEO_BERT, "neo-bert" },
@ -204,6 +205,7 @@ static const std::map<llm_kv, const char *> LLM_KV_NAMES = {
{ LLM_KV_ATTENTION_GATE_LORA_RANK, "%s.attention.gate_lora_rank" },
{ LLM_KV_ATTENTION_RELATIVE_BUCKETS_COUNT, "%s.attention.relative_buckets_count" },
{ LLM_KV_ATTENTION_SLIDING_WINDOW, "%s.attention.sliding_window" },
{ LLM_KV_ATTENTION_DENSE_EVERY_N_LAYERS, "%s.attention.dense_every_n_layers" },
{ LLM_KV_ATTENTION_SCALE, "%s.attention.scale" },
{ LLM_KV_ATTENTION_OUTPUT_SCALE, "%s.attention.output_scale" },
{ LLM_KV_ATTENTION_TEMPERATURE_LENGTH, "%s.attention.temperature_length" },
@ -214,6 +216,7 @@ static const std::map<llm_kv, const char *> LLM_KV_NAMES = {
{ LLM_KV_ROPE_DIMENSION_COUNT, "%s.rope.dimension_count" },
{ LLM_KV_ROPE_DIMENSION_SECTIONS, "%s.rope.dimension_sections" },
{ LLM_KV_ROPE_FREQ_BASE, "%s.rope.freq_base" },
{ LLM_KV_ROPE_FREQ_BASE_SWA, "%s.rope.freq_base_swa" },
{ LLM_KV_ROPE_SCALE_LINEAR, "%s.rope.scale_linear" },
{ LLM_KV_ROPE_SCALING_TYPE, "%s.rope.scaling.type" },
{ LLM_KV_ROPE_SCALING_FACTOR, "%s.rope.scaling.factor" },
@ -778,6 +781,21 @@ static std::set<llm_tensor> llm_get_tensor_names(llm_arch arch) {
LLM_TENSOR_CLS,
LLM_TENSOR_CLS_OUT,
};
case LLM_ARCH_MODERN_BERT:
return {
LLM_TENSOR_TOKEN_EMBD,
LLM_TENSOR_TOKEN_EMBD_NORM,
LLM_TENSOR_OUTPUT_NORM,
LLM_TENSOR_ATTN_NORM,
LLM_TENSOR_ATTN_OUT,
LLM_TENSOR_ATTN_QKV,
LLM_TENSOR_ATTN_ROT_EMBD,
LLM_TENSOR_FFN_DOWN,
LLM_TENSOR_FFN_UP,
LLM_TENSOR_FFN_NORM,
LLM_TENSOR_CLS,
LLM_TENSOR_CLS_OUT,
};
case LLM_ARCH_JINA_BERT_V2:
return {
LLM_TENSOR_TOKEN_EMBD,

4
src/llama-arch.h
View File

@ -24,6 +24,7 @@ enum llm_arch {
LLM_ARCH_STARCODER,
LLM_ARCH_REFACT,
LLM_ARCH_BERT,
LLM_ARCH_MODERN_BERT,
LLM_ARCH_NOMIC_BERT,
LLM_ARCH_NOMIC_BERT_MOE,
LLM_ARCH_NEO_BERT,
@ -188,6 +189,7 @@ enum llm_kv {
LLM_KV_EMBEDDING_SCALE,
LLM_KV_TOKEN_SHIFT_COUNT,
LLM_KV_INTERLEAVE_MOE_LAYER_STEP,
LLM_KV_DENSE_EVERY_N_LAYERS,
LLM_KV_ATTENTION_HEAD_COUNT,
LLM_KV_ATTENTION_HEAD_COUNT_KV,
@ -208,6 +210,7 @@ enum llm_kv {
LLM_KV_ATTENTION_GATE_LORA_RANK,
LLM_KV_ATTENTION_RELATIVE_BUCKETS_COUNT,
LLM_KV_ATTENTION_SLIDING_WINDOW,
LLM_KV_ATTENTION_DENSE_EVERY_N_LAYERS,
LLM_KV_ATTENTION_SCALE,
LLM_KV_ATTENTION_OUTPUT_SCALE,
LLM_KV_ATTENTION_TEMPERATURE_LENGTH,
@ -218,6 +221,7 @@ enum llm_kv {
LLM_KV_ROPE_DIMENSION_COUNT,
LLM_KV_ROPE_DIMENSION_SECTIONS,
LLM_KV_ROPE_FREQ_BASE,
LLM_KV_ROPE_FREQ_BASE_SWA,
LLM_KV_ROPE_SCALE_LINEAR,
LLM_KV_ROPE_SCALING_TYPE,
LLM_KV_ROPE_SCALING_FACTOR,

1
src/llama-hparams.h
View File

@ -123,6 +123,7 @@ struct llama_hparams {
llama_swa_type swa_type = LLAMA_SWA_TYPE_NONE;
// the size of the sliding window (0 - no SWA)
uint32_t n_swa = 0;
uint32_t n_swa_pattern = 1;
// if swa_layers[il] == true, then layer il is SWA
// if swa_layers[il] == false, then layer il is dense (i.e. non-SWA)
// by default, all layers are dense

1
src/llama-model-saver.cpp
View File

@ -182,6 +182,7 @@ void llama_model_saver::add_kv_from_model() {
add_kv(LLM_KV_ATTENTION_KV_LORA_RANK, hparams.n_lora_kv);
add_kv(LLM_KV_ATTENTION_RELATIVE_BUCKETS_COUNT, hparams.n_rel_attn_bkts);
add_kv(LLM_KV_ATTENTION_SLIDING_WINDOW, hparams.n_swa);
add_kv(LLM_KV_ATTENTION_DENSE_EVERY_N_LAYERS, hparams.n_swa_pattern);
add_kv(LLM_KV_ATTENTION_SCALE, hparams.f_attention_scale);
const float rope_scaling_factor = hparams.rope_freq_scale_train == 1.0f ? 0.0f : 1.0f/hparams.rope_freq_scale_train;

61
src/llama-model.cpp
View File

@ -875,6 +875,30 @@ void llama_model::load_hparams(llama_model_loader & ml) {
default: type = LLM_TYPE_UNKNOWN;
}
} break;
case LLM_ARCH_MODERN_BERT:
{
hparams.swa_type = LLAMA_SWA_TYPE_SYMMETRIC;
ml.get_key(LLM_KV_ROPE_FREQ_BASE_SWA, hparams.rope_freq_base_train_swa);
ml.get_key(LLM_KV_ATTENTION_SLIDING_WINDOW, hparams.n_swa);
ml.get_key(LLM_KV_ATTENTION_DENSE_EVERY_N_LAYERS, hparams.n_swa_pattern);
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, false);
hparams.set_swa_pattern(hparams.n_swa_pattern);
switch (hparams.n_layer) {
case 12:
type = LLM_TYPE_47M; break; // granite-embedding-small
case 22:
type = LLM_TYPE_149M; break; // modern-bert-base
case 28:
type = LLM_TYPE_395M; break; // modern-bert-large
default: type = LLM_TYPE_UNKNOWN;
}
} break;
case LLM_ARCH_JINA_BERT_V2:
{
ml.get_key(LLM_KV_ATTENTION_LAYERNORM_EPS, hparams.f_norm_eps);
@ -3155,6 +3179,37 @@ bool llama_model::load_tensors(llama_model_loader & ml) {
layer.layer_out_norm_b = create_tensor(tn(LLM_TENSOR_LAYER_OUT_NORM, "bias", i), {n_embd}, 0);
}
} break;
case LLM_ARCH_MODERN_BERT:
{
tok_embd = create_tensor(tn(LLM_TENSOR_TOKEN_EMBD, "weight"), {n_embd, n_vocab}, 0);
tok_norm = create_tensor(tn(LLM_TENSOR_TOKEN_EMBD_NORM, "weight"), {n_embd}, 0);
output_norm = create_tensor(tn(LLM_TENSOR_OUTPUT_NORM, "weight"), {n_embd}, 0);
for(int i = 0; i < n_layer; ++i) {
auto& layer = layers[i];
if ( i != 0 ) {
layer.attn_norm = create_tensor(tn(LLM_TENSOR_ATTN_NORM, "weight", i), {n_embd}, 0);
} else{
// layer 0 uses identity
layer.attn_norm = create_tensor(tn(LLM_TENSOR_ATTN_NORM, "weight", i), {n_embd}, TENSOR_NOT_REQUIRED);
}
layer.wqkv = create_tensor(tn(LLM_TENSOR_ATTN_QKV, "weight", i), {n_embd, 3 * n_embd }, 0);
layer.wo = create_tensor(tn(LLM_TENSOR_ATTN_OUT, "weight", i), {n_embd, n_embd}, 0);
layer.ffn_up = create_tensor(tn(LLM_TENSOR_FFN_UP, "weight", i), {n_embd, 2 * n_ff}, 0);
layer.ffn_down = create_tensor(tn(LLM_TENSOR_FFN_DOWN, "weight", i), {n_ff, n_embd}, 0);
layer.ffn_norm = create_tensor(tn(LLM_TENSOR_FFN_NORM, "weight", i), {n_embd}, 0);
}
cls = create_tensor(tn(LLM_TENSOR_CLS, "weight"), {n_embd, n_embd}, TENSOR_NOT_REQUIRED);
cls_out = create_tensor(tn(LLM_TENSOR_CLS_OUT, "weight"), {n_embd, hparams.n_cls_out}, TENSOR_NOT_REQUIRED);
cls_out_b = create_tensor(tn(LLM_TENSOR_CLS_OUT, "bias"), {hparams.n_cls_out}, TENSOR_NOT_REQUIRED);
} break;
case LLM_ARCH_NEO_BERT:
{
tok_embd = create_tensor(tn(LLM_TENSOR_TOKEN_EMBD, "weight"), {n_embd, n_vocab}, 0);
@ -7087,6 +7142,7 @@ llama_memory_i * llama_model::create_memory(const llama_memory_params & params,
case LLM_ARCH_NOMIC_BERT_MOE:
case LLM_ARCH_NEO_BERT:
case LLM_ARCH_WAVTOKENIZER_DEC:
case LLM_ARCH_MODERN_BERT:
case LLM_ARCH_GEMMA_EMBEDDING:
case LLM_ARCH_DREAM:
case LLM_ARCH_LLADA:
@ -7246,6 +7302,10 @@ ggml_cgraph * llama_model::build_graph(const llm_graph_params & params) const {
{
llm = std::make_unique<llm_build_bert>(*this, params);
} break;
case LLM_ARCH_MODERN_BERT:
{
llm = std::make_unique<llm_build_modern_bert<true>>(*this, params);
} break;
case LLM_ARCH_NEO_BERT:
{
llm = std::make_unique<llm_build_neo_bert>(*this, params);
@ -7814,6 +7874,7 @@ llama_rope_type llama_model_rope_type(const llama_model * model) {
case LLM_ARCH_DBRX:
case LLM_ARCH_BERT:
case LLM_ARCH_JINA_BERT_V3:
case LLM_ARCH_MODERN_BERT:
case LLM_ARCH_NOMIC_BERT:
case LLM_ARCH_NOMIC_BERT_MOE:
case LLM_ARCH_STABLELM:

3
src/llama-model.h
View File

@ -24,12 +24,14 @@ enum llm_type {
LLM_TYPE_17M,
LLM_TYPE_22M,
LLM_TYPE_33M,
LLM_TYPE_47M,
LLM_TYPE_60M,
LLM_TYPE_70M,
LLM_TYPE_80M,
LLM_TYPE_109M,
LLM_TYPE_137M,
LLM_TYPE_140M,
LLM_TYPE_149M,
LLM_TYPE_160M,
LLM_TYPE_190M,
LLM_TYPE_220M,
@ -39,6 +41,7 @@ enum llm_type {
LLM_TYPE_335M,
LLM_TYPE_350M,
LLM_TYPE_360M,
LLM_TYPE_395M,
LLM_TYPE_410M,
LLM_TYPE_450M,
LLM_TYPE_475M,

10
src/llama-vocab.cpp
View File

@ -1878,7 +1878,8 @@ void llama_vocab::impl::load(llama_model_loader & ml, const LLM_KV & kv) {
tokenizer_pre == "jina-v2-es" ||
tokenizer_pre == "jina-v2-de" ||
tokenizer_pre == "a.x-4.0" ||
tokenizer_pre == "mellum") {
tokenizer_pre == "mellum" ||
tokenizer_pre == "modern-bert" ) {
pre_type = LLAMA_VOCAB_PRE_TYPE_GPT2;
} else if (
tokenizer_pre == "jina-v1-en" ||
@ -2528,6 +2529,13 @@ void llama_vocab::impl::load(llama_model_loader & ml, const LLM_KV & kv) {
for (const auto * token : {"<unk>", "<s>", "<|endoftext|>"}) {
_set_token_attr(token, LLAMA_TOKEN_ATTR_RSTRIP, false);
}
} else if (_contains_any(model_name, {"modern-bert"})) {
if (token_to_id.count("[MASK]") == 0 ) {
LLAMA_LOG_WARN("%s: Mask token missing in vocab!\n", __func__);
}
else {
_set_token_attr("[MASK]", LLAMA_TOKEN_ATTR_LSTRIP, true);
}
}
}
}

5
src/models/models.h
View File

@ -327,6 +327,11 @@ struct llm_build_mistral3 : public llm_graph_context {
llm_build_mistral3(const llama_model & model, const llm_graph_params & params);
};
template <bool iswa>
struct llm_build_modern_bert : public llm_graph_context {
llm_build_modern_bert(const llama_model & model, const llm_graph_params & params);
};
struct llm_build_mpt : public llm_graph_context {
llm_build_mpt(const llama_model & model, const llm_graph_params & params);
};

126
src/models/modern-bert.cpp Normal file
View File

@ -0,0 +1,126 @@
#include "models.h"
template <bool iswa>
llm_build_modern_bert<iswa>::llm_build_modern_bert(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();
const float rope_theta_global = hparams.rope_freq_base_train;
const float rope_theta_local = hparams.rope_freq_base_train_swa;
const uint32_t n_swa_pattern = hparams.n_swa_pattern;
GGML_ASSERT(n_embd_head == hparams.n_embd_head_k);
ggml_tensor * cur = nullptr;
ggml_tensor * inpL = nullptr;
ggml_tensor * inp_pos = build_inp_pos();
// construct input embeddings (token, type, position)
inpL = build_inp_embd(model.tok_embd);
cb(inpL, "inp_embd", -1);
// embed layer norm
inpL = build_norm(inpL, model.tok_norm, nullptr, LLM_NORM, -1);
cb(inpL, "inp_norm", -1);
ggml_tensor * inp_out_ids = build_inp_out_ids();
auto * inp_attn = build_attn_inp_no_cache();
for (int il = 0; il < n_layer; ++il) {
ggml_tensor * cur = inpL;
ggml_tensor * Qcur = nullptr;
ggml_tensor * Kcur = nullptr;
ggml_tensor * Vcur = nullptr;
const float rope_theta = (il % n_swa_pattern == 0) ? rope_theta_global : rope_theta_local;
// attention layer norm
if (model.layers[il].attn_norm) {
cur = build_norm(inpL,
model.layers[il].attn_norm, NULL,
LLM_NORM, il);
cb(cur, "attn_norm", il);
}
// self attention
cur = build_lora_mm(model.layers[il].wqkv, cur);
cb(cur, "wqkv", il);
const size_t type_size = ggml_type_size(cur->type);
Qcur = ggml_view_3d(ctx0, cur, n_embd_head, n_head, n_tokens, n_embd_head*type_size, cur->nb[1], 0*type_size*(n_embd));
Kcur = ggml_view_3d(ctx0, cur, n_embd_head, n_head_kv, n_tokens, n_embd_head*type_size, cur->nb[1], 1*type_size*(n_embd));
Vcur = ggml_view_3d(ctx0, cur, n_embd_head, n_head_kv, n_tokens, n_embd_head*type_size, cur->nb[1], 1*type_size*(n_embd + n_embd_gqa));
// RoPE
Qcur = ggml_rope_ext(
ctx0, Qcur, inp_pos, nullptr,
n_rot, rope_type, n_ctx_orig, rope_theta, freq_scale,
ext_factor, attn_factor, beta_fast, beta_slow
);
Kcur = ggml_rope_ext(
ctx0, Kcur, inp_pos, nullptr,
n_rot, rope_type, n_ctx_orig, rope_theta, 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, nullptr,
Qcur, Kcur, Vcur, nullptr, nullptr, nullptr, 1.0f/sqrtf(float(n_embd_head)), il);
cb(cur, "kqv_out", il);
if (il == n_layer - 1 && inp_out_ids) {
cur = ggml_get_rows(ctx0, cur, inp_out_ids);
inpL = ggml_get_rows(ctx0, inpL, inp_out_ids);
}
// re-add the layer input
cur = ggml_add(ctx0, cur, inpL);
ggml_tensor * ffn_inp = cur;
// attention layer norm
cur = build_norm(cur, model.layers[il].ffn_norm, nullptr, LLM_NORM, il);
cb(ffn_inp, "ffn_inp", il);
cur = build_ffn(cur,
model.layers[il].ffn_up,
NULL, NULL, NULL, NULL, NULL,
model.layers[il].ffn_down,
NULL, NULL, NULL,
LLM_FFN_GEGLU, LLM_FFN_SEQ, il);
// attentions bypass the intermediate layer
cur = ggml_add(ctx0, cur, ffn_inp);
// input for next layer
inpL = cur;
}
cur = inpL;
cur = build_norm(cur,
model.output_norm, NULL,
LLM_NORM, -1);
cb(cur, "final_norm_out", -1);
if (hparams.pooling_type == LLAMA_POOLING_TYPE_CLS) {
// extracting cls token
cur = ggml_view_1d(ctx0, cur, hparams.n_embd, 0);
cb(cur, "cls_pooled_embd", -1);
}
cb(cur, "res_embd", -1);
res->t_embd = cur;
ggml_build_forward_expand(gf, cur);
}
// Explicit template instantiations
template struct llm_build_modern_bert<false>;
template struct llm_build_modern_bert<true>;