Public/koboldcpp
1
0
Fork 0
mirror of https://github.com/LostRuins/koboldcpp.git synced 2026-09-19 01:05:09 +02:00
Code Issues Packages Projects Releases Wiki Activity

Merge commit '8387ffb28d3467b81bc73727588c3fde772f8ebe' into concedo_experimental

Browse Source 
# Conflicts:
#	docs/backend/VirtGPU.md
#	docs/backend/ZenDNN.md
#	ggml/src/ggml-cpu/amx/amx.cpp
#	ggml/src/ggml-cpu/amx/mmq.cpp
#	ggml/src/ggml-sycl/add-id.cpp
#	ggml/src/ggml-virtgpu/backend/backend-dispatched-backend.cpp
#	ggml/src/ggml-virtgpu/backend/backend-dispatched-buffer-type.cpp
#	ggml/src/ggml-virtgpu/backend/backend-dispatched-buffer.cpp
#	ggml/src/ggml-virtgpu/backend/backend-dispatched.cpp
#	ggml/src/ggml-virtgpu/backend/backend-dispatched.gen.h
#	ggml/src/ggml-virtgpu/backend/backend-dispatched.h
#	ggml/src/ggml-virtgpu/backend/backend-virgl-apir.h
#	ggml/src/ggml-virtgpu/backend/backend.cpp
#	ggml/src/ggml-virtgpu/backend/shared/api_remoting.h
#	ggml/src/ggml-virtgpu/backend/shared/apir_backend.gen.h
#	ggml/src/ggml-virtgpu/backend/shared/apir_backend.h
#	ggml/src/ggml-virtgpu/backend/shared/apir_cs.h
#	ggml/src/ggml-virtgpu/backend/shared/apir_cs_ggml.h
#	ggml/src/ggml-virtgpu/backend/shared/apir_cs_rpc.h
#	ggml/src/ggml-virtgpu/ggml-backend-buffer-type.cpp
#	ggml/src/ggml-virtgpu/ggml-backend-device.cpp
#	ggml/src/ggml-virtgpu/ggml-backend-reg.cpp
#	ggml/src/ggml-virtgpu/ggml-backend.cpp
#	ggml/src/ggml-virtgpu/ggml-remoting.h
#	ggml/src/ggml-virtgpu/include/apir_hw.h
#	ggml/src/ggml-virtgpu/regenerate_remoting.py
#	ggml/src/ggml-virtgpu/virtgpu-forward-backend.cpp
#	ggml/src/ggml-virtgpu/virtgpu-forward-buffer-type.cpp
#	ggml/src/ggml-virtgpu/virtgpu-forward-buffer.cpp
#	ggml/src/ggml-virtgpu/virtgpu-forward-device.cpp
#	ggml/src/ggml-virtgpu/virtgpu-forward-impl.h
#	ggml/src/ggml-virtgpu/virtgpu-forward.gen.h
#	ggml/src/ggml-virtgpu/virtgpu.cpp
#	ggml/src/ggml-virtgpu/virtgpu.h
#	ggml/src/ggml-zendnn/CMakeLists.txt
#	ggml/src/ggml-zendnn/ggml-zendnn.cpp
#	src/CMakeLists.txt
#	tests/CMakeLists.txt
#	tests/test-tokenizer-0.sh
#	tools/cli/README.md
#	tools/completion/README.md
#	tools/imatrix/imatrix.cpp
#	tools/server/README.md
This commit is contained in:
Concedo
2026-02-28 12:45:16 +08:00
parent 5c40f07d4a 8387ffb28d
commit 4e358265a3
32 changed files with 521 additions and 80 deletions
Download Patch File Download Diff File Expand all files Collapse all files
src/llama-model.cpp
+52 -6
View File
@@ -53,6 +53,7 @@
#include "models/dream.cpp"
#include "models/ernie4-5-moe.cpp"
#include "models/ernie4-5.cpp"
#include "models/eurobert.cpp"
#include "models/exaone-moe.cpp"
#include "models/exaone.cpp"
#include "models/exaone4.cpp"
@@ -1092,6 +1093,16 @@ void llama_model::load_hparams(llama_model_loader & ml) {
type = LLM_TYPE_250M;
}
} break;
case LLM_ARCH_EUROBERT:
{
ml.get_key(LLM_KV_ATTENTION_LAYERNORM_RMS_EPS, hparams.f_norm_rms_eps);
ml.get_key(LLM_KV_ATTENTION_CAUSAL, hparams.causal_attn);
ml.get_key(LLM_KV_POOLING_TYPE, hparams.pooling_type);
if (hparams.n_layer == 12) {
type = LLM_TYPE_SMALL; // 0.2B
}
} break;
case LLM_ARCH_BLOOM:
{
ml.get_key(LLM_KV_ATTENTION_LAYERNORM_EPS, hparams.f_norm_eps);
@@ -3090,6 +3101,15 @@ bool llama_model::load_tensors(llama_model_loader & ml) {
// TODO: move to a separate function
const auto tn = LLM_TN(arch);
// helper: try merged gate_up_exps first, fall back to separate gate and up
auto create_tensor_gate_up_exps = [&](llama_layer & layer, int bid, int64_t n_embd_, int64_t n_ff_, int64_t n_expert_, int flags) {
layer.ffn_gate_up_exps = create_tensor(tn(LLM_TENSOR_FFN_GATE_UP_EXPS, "weight", bid), {n_embd_, n_ff_ * 2, n_expert_}, TENSOR_NOT_REQUIRED);
if (layer.ffn_gate_up_exps == nullptr) {
layer.ffn_gate_exps = create_tensor(tn(LLM_TENSOR_FFN_GATE_EXPS, "weight", bid), {n_embd_, n_ff_, n_expert_}, flags);
layer.ffn_up_exps = create_tensor(tn(LLM_TENSOR_FFN_UP_EXPS, "weight", bid), {n_embd_, n_ff_, n_expert_}, flags);
}
};
switch (arch) {
case LLM_ARCH_LLAMA:
case LLM_ARCH_REFACT:
@@ -3730,6 +3750,29 @@ bool llama_model::load_tensors(llama_model_loader & ml) {
layer.ffn_down = create_tensor(tn(LLM_TENSOR_FFN_DOWN, "weight", i), {n_ff, n_embd}, 0);
}
} break;
case LLM_ARCH_EUROBERT:
{
tok_embd = create_tensor(tn(LLM_TENSOR_TOKEN_EMBD, "weight"), {n_embd, n_vocab}, 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];
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}, 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, n_embd}, 0);
layer.ffn_norm = create_tensor(tn(LLM_TENSOR_FFN_NORM, "weight", i), {n_embd}, 0);
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);
}
} break;
case LLM_ARCH_JINA_BERT_V2:
{
tok_embd = create_tensor(tn(LLM_TENSOR_TOKEN_EMBD, "weight"), {n_embd, n_vocab}, 0); // word_embeddings
@@ -5348,9 +5391,8 @@ bool llama_model::load_tensors(llama_model_loader & ml) {
}
// MoE branch
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);
create_tensor_gate_up_exps(layer, i, n_embd, n_ff_exp, n_expert, 0);
// Shared expert branch
layer.ffn_gate_shexp = create_tensor(tn(LLM_TENSOR_FFN_GATE_SHEXP, "weight", i), {n_embd, n_ff_exp * n_expert_shared}, 0);
@@ -7552,9 +7594,8 @@ bool llama_model::load_tensors(llama_model_loader & ml) {
}
layer.ffn_gate_inp = create_tensor(tn(LLM_TENSOR_FFN_GATE_INP, "weight", i), { n_embd, n_expert }, 0);
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);
create_tensor_gate_up_exps(layer, i, n_embd, n_ff_exp, n_expert, 0);
// Shared experts
layer.ffn_gate_inp_shexp = create_tensor(tn(LLM_TENSOR_FFN_GATE_INP_SHEXP, "weight", i), { n_embd }, 0);
@@ -7618,9 +7659,8 @@ bool llama_model::load_tensors(llama_model_loader & ml) {
}
layer.ffn_gate_inp = create_tensor(tn(LLM_TENSOR_FFN_GATE_INP, "weight", i), { n_embd, n_expert }, 0);
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);
create_tensor_gate_up_exps(layer, i, n_embd, n_ff_exp, n_expert, 0);
// Shared experts
const int64_t n_ff_shexp = hparams.n_ff_shexp ? hparams.n_ff_shexp : n_ff;
@@ -8342,6 +8382,7 @@ llama_memory_i * llama_model::create_memory(const llama_memory_params & params,
case LLM_ARCH_NOMIC_BERT:
case LLM_ARCH_NOMIC_BERT_MOE:
case LLM_ARCH_NEO_BERT:
case LLM_ARCH_EUROBERT:
case LLM_ARCH_WAVTOKENIZER_DEC:
case LLM_ARCH_MODERN_BERT:
case LLM_ARCH_GEMMA_EMBEDDING:
@@ -8539,6 +8580,10 @@ ggml_cgraph * llama_model::build_graph(const llm_graph_params & params) const {
{
llm = std::make_unique<llm_build_neo_bert>(*this, params);
} break;
case LLM_ARCH_EUROBERT:
{
llm = std::make_unique<llm_build_eurobert>(*this, params);
} break;
case LLM_ARCH_BLOOM:
{
llm = std::make_unique<llm_build_bloom>(*this, params);
@@ -9165,6 +9210,7 @@ llama_rope_type llama_model_rope_type(const llama_model * model) {
case LLM_ARCH_MODERN_BERT:
case LLM_ARCH_NOMIC_BERT:
case LLM_ARCH_NOMIC_BERT_MOE:
case LLM_ARCH_EUROBERT:
case LLM_ARCH_STABLELM:
case LLM_ARCH_BITNET:
case LLM_ARCH_QWEN: