vrr/koboldcpp
2
0
Fork 0
mirror of https://github.com/LostRuins/koboldcpp.git synced 2026-06-01 22:50:53 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions

Merge branch 'upstream' into concedo_experimental

Browse source 
# Conflicts:
#	.github/workflows/release.yml
#	.github/workflows/server.yml
#	.github/workflows/ui-build.yml
#	.github/workflows/ui-publish.yml
#	CMakeLists.txt
#	docs/autoparser.md
#	docs/backend/snapdragon/CMakeUserPresets.json
#	docs/backend/snapdragon/README.md
#	docs/backend/snapdragon/windows.md
#	docs/function-calling.md
#	examples/model-conversion/scripts/embedding/run-original-model.py
#	ggml/src/ggml-hexagon/htp/hmx-flash-attn-ops.c
#	ggml/src/ggml-opencl/ggml-opencl.cpp
#	ggml/src/ggml-opencl/kernels/cvt.cl
#	ggml/src/ggml-opencl/kernels/gemm_moe_mxfp4_f32_ns.cl
#	ggml/src/ggml-opencl/kernels/gemm_moe_q4_0_f32_ns.cl
#	ggml/src/ggml-opencl/kernels/gemm_moe_q4_1_f32_ns.cl
#	ggml/src/ggml-opencl/kernels/gemm_moe_q4_k_f32_ns.cl
#	ggml/src/ggml-opencl/kernels/gemm_moe_q5_0_f32_ns.cl
#	ggml/src/ggml-opencl/kernels/gemm_moe_q5_1_f32_ns.cl
#	ggml/src/ggml-opencl/kernels/gemm_moe_q5_k_f32_ns.cl
#	ggml/src/ggml-opencl/kernels/gemm_moe_q6_k_f32_ns.cl
#	ggml/src/ggml-opencl/kernels/gemv_moe_mxfp4_f32_ns.cl
#	ggml/src/ggml-opencl/kernels/gemv_moe_q4_0_f32_ns.cl
#	ggml/src/ggml-opencl/kernels/gemv_moe_q4_1_f32_ns.cl
#	ggml/src/ggml-opencl/kernels/gemv_moe_q4_k_f32_ns.cl
#	ggml/src/ggml-opencl/kernels/gemv_moe_q5_0_f32_ns.cl
#	ggml/src/ggml-opencl/kernels/gemv_moe_q5_1_f32_ns.cl
#	ggml/src/ggml-opencl/kernels/gemv_moe_q5_k_f32_ns.cl
#	ggml/src/ggml-opencl/kernels/gemv_moe_q6_k_f32_ns.cl
#	ggml/src/ggml-sycl/common.hpp
#	ggml/src/ggml-sycl/dmmv.cpp
#	ggml/src/ggml-sycl/gated_delta_net.cpp
#	ggml/src/ggml-sycl/ggml-sycl.cpp
#	ggml/src/ggml-vulkan/CMakeLists.txt
#	ggml/src/ggml-zendnn/CMakeLists.txt
#	ggml/src/ggml-zendnn/ggml-zendnn.cpp
#	requirements/requirements-convert_hf_to_gguf.txt
#	scripts/snapdragon/windows/setup-build.ps1
#	tools/perplexity/perplexity.cpp
This commit is contained in:
Concedo 2026-05-24 13:55:44 +08:00
commit 8ca4283f55
17 changed files with 323 additions and 203 deletions
Download patch file Download diff file Expand all files Collapse all files

12
src/llama-model.cpp
View file

@ -1465,6 +1465,12 @@ bool llama_model_base::load_tensors(llama_model_loader & ml) {
if (!layer.ssm_beta_s && layer.ssm_beta) {
layer.ssm_beta_s = create_tensor(tn(LLM_TENSOR_SSM_BETA, "scale", i), {1}, TENSOR_NOT_REQUIRED);
}
if (!layer.nextn.eh_proj_s && layer.nextn.eh_proj) {
layer.nextn.eh_proj_s = create_tensor(tn(LLM_TENSOR_NEXTN_EH_PROJ, "scale", i), {1}, TENSOR_NOT_REQUIRED);
}
if (!layer.nextn.shared_head_head_s && layer.nextn.shared_head_head) {
layer.nextn.shared_head_head_s = create_tensor(tn(LLM_TENSOR_NEXTN_SHARED_HEAD_HEAD, "scale", i), {1}, TENSOR_NOT_REQUIRED);
}
// input scales
if (!layer.wq_in_s && layer.wq) {
@ -1524,6 +1530,12 @@ bool llama_model_base::load_tensors(llama_model_loader & ml) {
if (!layer.ssm_beta_in_s && layer.ssm_beta) {
layer.ssm_beta_in_s = create_tensor(tn(LLM_TENSOR_SSM_BETA, "input_scale", i), {1}, TENSOR_NOT_REQUIRED);
}
if (!layer.nextn.eh_proj_in_s && layer.nextn.eh_proj) {
layer.nextn.eh_proj_in_s = create_tensor(tn(LLM_TENSOR_NEXTN_EH_PROJ, "input_scale", i), {1}, TENSOR_NOT_REQUIRED);
}
if (!layer.nextn.shared_head_head_in_s && layer.nextn.shared_head_head) {
layer.nextn.shared_head_head_in_s = create_tensor(tn(LLM_TENSOR_NEXTN_SHARED_HEAD_HEAD, "input_scale", i), {1}, TENSOR_NOT_REQUIRED);
}
}
// output scales
if (output && output->type == GGML_TYPE_NVFP4) {

16
src/llama-model.h
View file

@ -202,12 +202,16 @@ struct llama_layer_shortconv {
};
struct llama_layer_nextn {
struct ggml_tensor * eh_proj = nullptr;
struct ggml_tensor * embed_tokens = nullptr;
struct ggml_tensor * enorm = nullptr;
struct ggml_tensor * hnorm = nullptr;
struct ggml_tensor * shared_head_head = nullptr;
struct ggml_tensor * shared_head_norm = nullptr;
struct ggml_tensor * eh_proj = nullptr;
struct ggml_tensor * eh_proj_s = nullptr;
struct ggml_tensor * eh_proj_in_s = nullptr;
struct ggml_tensor * embed_tokens = nullptr;
struct ggml_tensor * enorm = nullptr;
struct ggml_tensor * hnorm = nullptr;
struct ggml_tensor * shared_head_head = nullptr;
struct ggml_tensor * shared_head_head_s = nullptr;
struct ggml_tensor * shared_head_head_in_s = nullptr;
struct ggml_tensor * shared_head_norm = nullptr;
};
struct llama_layer {

5
src/models/qwen35.cpp
View file

@ -538,7 +538,7 @@ llama_model_qwen35::graph_mtp::graph_mtp(const llama_model & model, const llm_gr
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);
ggml_tensor * cur = build_lora_mm(layer.nextn.eh_proj, concat, layer.nextn.eh_proj_s);
cb(cur, "mtp_eh_proj", il);
ggml_tensor * inpSA = cur;
@ -626,8 +626,9 @@ llama_model_qwen35::graph_mtp::graph_mtp(const llama_model & model, const llm_gr
cb(cur, "mtp_shared_head_norm", -1);
ggml_tensor * head_w = layer.nextn.shared_head_head ? layer.nextn.shared_head_head : model.output;
ggml_tensor * head_s = layer.nextn.shared_head_head ? layer.nextn.shared_head_head_s : model.output_s;
GGML_ASSERT(head_w && "QWEN35 MTP: missing LM head (nextn.shared_head_head or model.output)");
cur = build_lora_mm(head_w, cur);
cur = build_lora_mm(head_w, cur, head_s);
cb(cur, "result_output", -1);
res->t_logits = cur;

5
src/models/qwen35moe.cpp
View file

@ -602,7 +602,7 @@ llama_model_qwen35moe::graph_mtp::graph_mtp(const llama_model & model, const llm
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);
ggml_tensor * cur = build_lora_mm(layer.nextn.eh_proj, concat, layer.nextn.eh_proj_s);
cb(cur, "mtp_eh_proj", il);
ggml_tensor * inpSA = cur;
@ -722,8 +722,9 @@ llama_model_qwen35moe::graph_mtp::graph_mtp(const llama_model & model, const llm
cb(cur, "mtp_shared_head_norm", -1);
ggml_tensor * head_w = layer.nextn.shared_head_head ? layer.nextn.shared_head_head : model.output;
ggml_tensor * head_s = layer.nextn.shared_head_head ? layer.nextn.shared_head_head_s : model.output_s;
GGML_ASSERT(head_w && "QWEN35MOE MTP: missing LM head (nextn.shared_head_head or model.output)");
cur = build_lora_mm(head_w, cur);
cur = build_lora_mm(head_w, cur, head_s);
cb(cur, "result_output", -1);
res->t_logits = cur;