Merge branch 'master' into pr/8836
This commit is contained in:
Nexesenex
commit
8c10533409
5 changed files with 20 additions and 20 deletions
4
.github/workflows/bench.yml
vendored
4
.github/workflows/bench.yml
vendored
|
|
@ -129,6 +129,8 @@ jobs:
|
|||
|
||||
- name: Server bench
|
||||
id: server_bench
|
||||
env:
|
||||
HEAD_REF: ${{ github.head_ref || github.ref_name }}
|
||||
run: |
|
||||
set -eux
|
||||
|
||||
|
|
@ -137,7 +139,7 @@ jobs:
|
|||
python bench.py \
|
||||
--runner-label ${{ env.RUNNER_LABEL }} \
|
||||
--name ${{ github.job }} \
|
||||
--branch ${{ github.head_ref || github.ref_name }} \
|
||||
--branch $HEAD_REF \
|
||||
--commit ${{ github.event.inputs.sha || github.event.pull_request.head.sha || github.sha }} \
|
||||
--scenario script.js \
|
||||
--duration ${{ github.event.inputs.duration || env.DURATION }} \
|
||||
|
|
|
|||
22
.github/workflows/build.yml
vendored
22
.github/workflows/build.yml
vendored
|
|
@ -47,7 +47,7 @@ jobs:
|
|||
sysctl -a
|
||||
mkdir build
|
||||
cd build
|
||||
cmake -DLLAMA_FATAL_WARNINGS=ON -DGGML_METAL_EMBED_LIBRARY=ON -DLLAMA_CURL=ON -DBUILD_SHARED_LIBS=OFF ..
|
||||
cmake -DLLAMA_FATAL_WARNINGS=ON -DGGML_METAL_EMBED_LIBRARY=ON -DLLAMA_CURL=ON -DGGML_RPC=ON -DBUILD_SHARED_LIBS=OFF ..
|
||||
cmake --build . --config Release -j $(sysctl -n hw.logicalcpu)
|
||||
|
||||
- name: Test
|
||||
|
|
@ -105,7 +105,7 @@ jobs:
|
|||
sysctl -a
|
||||
# Metal is disabled due to intermittent failures with Github runners not having a GPU:
|
||||
# https://github.com/ggerganov/llama.cpp/actions/runs/8635935781/job/23674807267#step:5:2313
|
||||
cmake -B build -DLLAMA_FATAL_WARNINGS=ON -DGGML_METAL=OFF -DLLAMA_CURL=ON -DBUILD_SHARED_LIBS=OFF
|
||||
cmake -B build -DLLAMA_FATAL_WARNINGS=ON -DGGML_METAL=OFF -DLLAMA_CURL=ON -DGGML_RPC=ON -DBUILD_SHARED_LIBS=OFF
|
||||
cmake --build build --config Release -j $(sysctl -n hw.logicalcpu)
|
||||
|
||||
- name: Test
|
||||
|
|
@ -222,7 +222,7 @@ jobs:
|
|||
run: |
|
||||
mkdir build
|
||||
cd build
|
||||
cmake .. -DLLAMA_FATAL_WARNINGS=ON -DLLAMA_CURL=ON -DBUILD_SHARED_LIBS=OFF
|
||||
cmake .. -DLLAMA_FATAL_WARNINGS=ON -DLLAMA_CURL=ON -DGGML_RPC=ON -DBUILD_SHARED_LIBS=OFF
|
||||
cmake --build . --config Release -j $(nproc)
|
||||
|
||||
- name: Test
|
||||
|
|
@ -696,22 +696,20 @@ jobs:
|
|||
strategy:
|
||||
matrix:
|
||||
include:
|
||||
- build: 'rpc-x64'
|
||||
defines: '-DGGML_NATIVE=OFF -DLLAMA_BUILD_SERVER=ON -DGGML_RPC=ON -DBUILD_SHARED_LIBS=ON'
|
||||
- build: 'noavx-x64'
|
||||
defines: '-DGGML_NATIVE=OFF -DLLAMA_BUILD_SERVER=ON -DGGML_AVX=OFF -DGGML_AVX2=OFF -DGGML_FMA=OFF -DBUILD_SHARED_LIBS=ON'
|
||||
defines: '-DGGML_NATIVE=OFF -DLLAMA_BUILD_SERVER=ON -DGGML_RPC=ON -DGGML_AVX=OFF -DGGML_AVX2=OFF -DGGML_FMA=OFF -DBUILD_SHARED_LIBS=ON'
|
||||
- build: 'avx2-x64'
|
||||
defines: '-DGGML_NATIVE=OFF -DLLAMA_BUILD_SERVER=ON -DBUILD_SHARED_LIBS=ON'
|
||||
defines: '-DGGML_NATIVE=OFF -DLLAMA_BUILD_SERVER=ON -DGGML_RPC=ON -DBUILD_SHARED_LIBS=ON'
|
||||
- build: 'avx-x64'
|
||||
defines: '-DGGML_NATIVE=OFF -DLLAMA_BUILD_SERVER=ON -DGGML_AVX2=OFF -DBUILD_SHARED_LIBS=ON'
|
||||
defines: '-DGGML_NATIVE=OFF -DLLAMA_BUILD_SERVER=ON -DGGML_RPC=ON -DGGML_AVX2=OFF -DBUILD_SHARED_LIBS=ON'
|
||||
- build: 'avx512-x64'
|
||||
defines: '-DGGML_NATIVE=OFF -DLLAMA_BUILD_SERVER=ON -DGGML_AVX512=ON -DBUILD_SHARED_LIBS=ON'
|
||||
defines: '-DGGML_NATIVE=OFF -DLLAMA_BUILD_SERVER=ON -DGGML_RPC=ON -DGGML_AVX512=ON -DBUILD_SHARED_LIBS=ON'
|
||||
- build: 'openblas-x64'
|
||||
defines: '-DGGML_NATIVE=OFF -DLLAMA_BUILD_SERVER=ON -DGGML_BLAS=ON -DBUILD_SHARED_LIBS=ON -DGGML_BLAS_VENDOR=OpenBLAS -DBLAS_INCLUDE_DIRS="$env:RUNNER_TEMP/openblas/include" -DBLAS_LIBRARIES="$env:RUNNER_TEMP/openblas/lib/openblas.lib"'
|
||||
defines: '-DGGML_NATIVE=OFF -DLLAMA_BUILD_SERVER=ON -DGGML_RPC=ON -DGGML_BLAS=ON -DBUILD_SHARED_LIBS=ON -DGGML_BLAS_VENDOR=OpenBLAS -DBLAS_INCLUDE_DIRS="$env:RUNNER_TEMP/openblas/include" -DBLAS_LIBRARIES="$env:RUNNER_TEMP/openblas/lib/openblas.lib"'
|
||||
- build: 'kompute-x64'
|
||||
defines: '-DGGML_NATIVE=OFF -DLLAMA_BUILD_SERVER=ON -DGGML_KOMPUTE=ON -DKOMPUTE_OPT_DISABLE_VULKAN_VERSION_CHECK=ON -DBUILD_SHARED_LIBS=ON'
|
||||
defines: '-DGGML_NATIVE=OFF -DLLAMA_BUILD_SERVER=ON -DGGML_RPC=ON -DGGML_KOMPUTE=ON -DKOMPUTE_OPT_DISABLE_VULKAN_VERSION_CHECK=ON -DBUILD_SHARED_LIBS=ON'
|
||||
- build: 'vulkan-x64'
|
||||
defines: '-DGGML_NATIVE=OFF -DLLAMA_BUILD_SERVER=ON -DGGML_VULKAN=ON -DBUILD_SHARED_LIBS=ON'
|
||||
defines: '-DGGML_NATIVE=OFF -DLLAMA_BUILD_SERVER=ON -DGGML_RPC=ON -DGGML_VULKAN=ON -DBUILD_SHARED_LIBS=ON'
|
||||
- build: 'llvm-arm64'
|
||||
defines: '-G "Ninja Multi-Config" -D CMAKE_TOOLCHAIN_FILE=cmake/arm64-windows-llvm.cmake -DGGML_NATIVE=OFF -DLLAMA_BUILD_SERVER=ON -DBUILD_SHARED_LIBS=ON'
|
||||
- build: 'msvc-arm64'
|
||||
|
|
|
|||
|
|
@ -186,10 +186,12 @@ Unless otherwise noted these projects are open-source with permissive licensing:
|
|||
|
||||
- [akx/ggify](https://github.com/akx/ggify) – download PyTorch models from HuggingFace Hub and convert them to GGML
|
||||
- [crashr/gppm](https://github.com/crashr/gppm) – launch llama.cpp instances utilizing NVIDIA Tesla P40 or P100 GPUs with reduced idle power consumption
|
||||
- [gpustack/gguf-parser](https://github.com/gpustack/gguf-parser-go/tree/main/cmd/gguf-parser) - review/check the GGUF file and estimate the memory usage
|
||||
|
||||
**Infrastructure:**
|
||||
|
||||
- [Paddler](https://github.com/distantmagic/paddler) - Stateful load balancer custom-tailored for llama.cpp
|
||||
- [GPUStack](https://github.com/gpustack/gpustack) - Manage GPU clusters for running LLMs
|
||||
|
||||
**Games:**
|
||||
- [Lucy's Labyrinth](https://github.com/MorganRO8/Lucys_Labyrinth) - A simple maze game where agents controlled by an AI model will try to trick you.
|
||||
|
|
|
|||
|
|
@ -369,6 +369,9 @@ namespace grammar_parser {
|
|||
}
|
||||
// Validate the state to ensure that all rules are defined
|
||||
for (const auto & rule : state.rules) {
|
||||
if (rule.empty()) {
|
||||
throw std::runtime_error("Undefined rule");
|
||||
}
|
||||
for (const auto & elem : rule) {
|
||||
if (elem.type == LLAMA_GRETYPE_RULE_REF) {
|
||||
// Ensure that the rule at that location exists
|
||||
|
|
|
|||
|
|
@ -3575,13 +3575,8 @@ namespace GGUFMeta {
|
|||
|
||||
using llama_buf_map = std::unordered_map<uint32_t, ggml_backend_buffer_t>;
|
||||
|
||||
// TODO: update when needed or think of some clever automatic way to do this
|
||||
static size_t llama_model_max_nodes(const llama_model & /*model*/) {
|
||||
//if (model.arch == LLM_ARCH_LLAMA && model.hparams.n_layer > ??) { // llama-3 405B
|
||||
// return 32768;
|
||||
//}
|
||||
|
||||
return 8192;
|
||||
static size_t llama_model_max_nodes(const llama_model & model) {
|
||||
return std::max<size_t>(8192, model.tensors_by_name.size()*5);
|
||||
}
|
||||
|
||||
struct llama_model_loader {
|
||||
|
|
|
|||
Loading…
Add table
Add a link
Reference in a new issue