Commit e6c16af
- .github/workflows/release.yml+3
- SECURITY.md+1-1
- app/llama.cpp+8-4
- common/arg.cpp+2-2
- common/chat.cpp+151
- common/common.cpp+47-47
- common/common.h+9-1
- common/fit.cpp+1-1
- common/jinja/runtime.cpp+46
- common/jinja/runtime.h+127
- common/jinja/value.cpp+44
- common/reasoning-budget.cpp+10-10
- common/speculative.cpp+371-65
- conversion/__init__.py+1
- conversion/qwen.py+52
- docs/speculative.md+28-1
- ggml/src/ggml-cuda/cpy.cu+45
- ggml/src/ggml-opencl/CMakeLists.txt+3
- ggml/src/ggml-opencl/fa_tune.h+91
- ggml/src/ggml-opencl/ggml-opencl.cpp+1.8k-254
- ggml/src/ggml-opencl/kernels/cvt.cl+152
- ggml/src/ggml-opencl/kernels/flash_attn_f16.cl+75-40
- ggml/src/ggml-opencl/kernels/flash_attn_f32.cl+73-38
- ggml/src/ggml-opencl/kernels/flash_attn_f32_f16.cl+684-81
- ggml/src/ggml-opencl/kernels/flash_attn_f32_q4_0.cl+1.0k
- ggml/src/ggml-opencl/kernels/flash_attn_f32_q8_0.cl+1.0k
- ggml/src/ggml-opencl/kernels/flash_attn_pre_f16.cl+156
- ggml/src/ggml-opencl/kernels/set_rows.cl+500
- ggml/src/ggml-openvino/ggml-openvino.cpp+4
- ggml/src/ggml-sycl/norm.cpp+103-49
- ggml/src/ggml-vulkan/vulkan-shaders/unary.comp+1-1
- gguf-py/gguf/constants.py+18
- models/templates/openbmb-MiniCPM5-1B.jinja+179
- src/llama-arch.cpp+1
- src/llama-arch.h+1
- src/llama-context.cpp+3-3
- src/llama-graph.cpp+6-1
- src/llama-model.cpp+5-1
- src/models/dflash.cpp+276
- src/models/models.h+16
- tests/CMakeLists.txt+4-4
- tests/test-backend-ops.cpp+29-8
- tests/test-chat-template.cpp+18-4
- tests/test-chat.cpp+78
- tests/test-export-graph-ops.cpp+1-1
- tests/test-jinja.cpp+30
- tests/test-llama-archs.cpp+2-2
- tools/rpc/CMakeLists.txt+1-1
- tools/rpc/README.md+17-17
- tools/server/server-context.cpp+44-42
- tools/server/server-http.cpp+8-8
- tools/server/server-schema.cpp+1-1
- tools/server/server-stream.cpp+5-6
- tools/server/server-task.cpp+6-6
- tools/server/server.cpp+4-8
0 commit comments