From be20e7f49d9e5c6d9e8d9b4871eeba3df7a1639d Mon Sep 17 00:00:00 2001 From: Xuan Son Nguyen Date: Fri, 5 Jul 2024 18:08:32 +0200 Subject: [PATCH 1/3] Reorganize documentation pages (#8325) * re-organize docs * add link among docs * add link to build docs * fix style * de-duplicate sections --- .gitignore | 1 + README.md | 672 ++---------------- docs/android.md | 56 ++ docs/{ => backend}/BLIS.md | 0 README-sycl.md => docs/backend/SYCL.md | 0 docs/build.md | 288 ++++++++ docs/{ => development}/HOWTO-add-model.md | 2 +- docs/{ => development}/debugging-tests.md | 0 .../llama-star/idea-arch.key | Bin .../llama-star/idea-arch.pdf | Bin .../token_generation_performance_tips.md | 0 docs/docker.md | 86 +++ docs/install.md | 39 + examples/quantize/README.md | 89 ++- 14 files changed, 628 insertions(+), 605 deletions(-) create mode 100644 docs/android.md rename docs/{ => backend}/BLIS.md (100%) rename README-sycl.md => docs/backend/SYCL.md (100%) create mode 100644 docs/build.md rename docs/{ => development}/HOWTO-add-model.md (99%) rename docs/{ => development}/debugging-tests.md (100%) rename docs/{ => development}/llama-star/idea-arch.key (100%) rename docs/{ => development}/llama-star/idea-arch.pdf (100%) rename docs/{ => development}/token_generation_performance_tips.md (100%) create mode 100644 docs/docker.md create mode 100644 docs/install.md diff --git a/.gitignore b/.gitignore index 6125e350c..4866f6122 100644 --- a/.gitignore +++ b/.gitignore @@ -47,6 +47,7 @@ build* !build-info.cpp.in !build-info.sh !build.zig +!docs/build.md /libllama.so /llama-* android-ndk-* diff --git a/README.md b/README.md index 875fed437..a4bbf00bd 100644 --- a/README.md +++ b/README.md @@ -13,7 +13,7 @@ Inference of Meta's [LLaMA](https://arxiv.org/abs/2302.13971) model (and others) > [!IMPORTANT] [2024 Jun 12] Binaries have been renamed w/ a `llama-` prefix. `main` is now `llama-cli`, `server` is `llama-server`, etc (https://github.com/ggerganov/llama.cpp/pull/7809) -### Recent API changes +## Recent API changes - [2024 Jun 26] The source code and CMake build scripts have been restructured https://github.com/ggerganov/llama.cpp/pull/8006 - [2024 Apr 21] `llama_token_to_piece` can now optionally render special tokens https://github.com/ggerganov/llama.cpp/pull/6807 @@ -24,7 +24,7 @@ Inference of Meta's [LLaMA](https://arxiv.org/abs/2302.13971) model (and others) - [2024 Mar 4] Embeddings API updated https://github.com/ggerganov/llama.cpp/pull/5796 - [2024 Mar 3] `struct llama_context_params` https://github.com/ggerganov/llama.cpp/pull/5849 -### Hot topics +## Hot topics - **`convert.py` has been deprecated and moved to `examples/convert_legacy_llama.py`, please use `convert_hf_to_gguf.py`** https://github.com/ggerganov/llama.cpp/pull/7430 - Initial Flash-Attention support: https://github.com/ggerganov/llama.cpp/pull/5021 @@ -39,37 +39,6 @@ Inference of Meta's [LLaMA](https://arxiv.org/abs/2302.13971) model (and others) ---- -
- Table of Contents -
    -
  1. - Description -
    2. -
  2. - Usage - -
    3. -
  3. Contributing
    4. -
  4. Coding guidelines
    5. -
  5. Docs
    6. -
-
- ## Description The main goal of `llama.cpp` is to enable LLM inference with minimal setup and state-of-the-art performance on a wide @@ -87,14 +56,6 @@ Since its [inception](https://github.com/ggerganov/llama.cpp/issues/33#issuecomm improved significantly thanks to many contributions. It is the main playground for developing new features for the [ggml](https://github.com/ggerganov/ggml) library. -**Supported platforms:** - -- [X] Mac OS -- [X] Linux -- [X] Windows (via CMake) -- [X] Docker -- [X] FreeBSD - **Supported models:** Typically finetunes of the base models below are supported as well. @@ -150,12 +111,6 @@ Typically finetunes of the base models below are supported as well. - [x] [Moondream](https://huggingface.co/vikhyatk/moondream2) - [x] [Bunny](https://github.com/BAAI-DCAI/Bunny) -**HTTP server** - -[llama.cpp web server](./examples/server) is a lightweight [OpenAI API](https://github.com/openai/openai-openapi) compatible HTTP server that can be used to serve local models and easily connect them to existing clients. - -[simplechat](./examples/server/public_simplechat) is a simple chat client, which can be used to chat with the model exposed using above web server (use --path to point to simplechat), from a local web browser. - **Bindings:** - Python: [abetlen/llama-cpp-python](https://github.com/abetlen/llama-cpp-python) @@ -224,9 +179,10 @@ Unless otherwise noted these projects are open-source with permissive licensing: - [Paddler](https://github.com/distantmagic/paddler) - Stateful load balancer custom-tailored for llama.cpp ---- +## Demo -Here is a typical run using LLaMA v2 13B on M2 Ultra: +
+Typical run using LLaMA v2 13B on M2 Ultra ``` $ make -j && ./llama-cli -m models/llama-13b-v2/ggml-model-q4_0.gguf -p "Building a website can be done in 10 simple steps:\nStep 1:" -n 400 -e @@ -306,454 +262,85 @@ llama_print_timings: eval time = 24513.59 ms / 399 runs ( 61.44 ms llama_print_timings: total time = 25431.49 ms ``` +
+ +
+Demo of running both LLaMA-7B and whisper.cpp on a single M1 Pro MacBook + And here is another demo of running both LLaMA-7B and [whisper.cpp](https://github.com/ggerganov/whisper.cpp) on a single M1 Pro MacBook: https://user-images.githubusercontent.com/1991296/224442907-7693d4be-acaa-4e01-8b4f-add84093ffff.mp4 +
+ ## Usage Here are the end-to-end binary build and model conversion steps for most supported models. -### Get the Code +### Basic usage + +Firstly, you need to get the binary. There are different methods that you can follow: +- Method 1: Clone this repository and build locally, see [how to build](./docs/build.md) +- Method 2: If you are using MacOS or Linux, you can install llama.cpp via [brew, flox or nix](./docs/install.md) +- Method 3: Use a Docker image, see [documentation for Docker](./docs/docker.md) +- Method 4: Download pre-built binary from [releases](https://github.com/ggerganov/llama.cpp/releases) + +You can run a basic completion using this command: ```bash -git clone https://github.com/ggerganov/llama.cpp -cd llama.cpp +llama-cli -m your_model.gguf -p "I believe the meaning of life is" -n 128 + +# Output: +# I believe the meaning of life is to find your own truth and to live in accordance with it. For me, this means being true to myself and following my passions, even if they don't align with societal expectations. I think that's what I love about yoga – it's not just a physical practice, but a spiritual one too. It's about connecting with yourself, listening to your inner voice, and honoring your own unique journey. ``` -### Build +See [this page](./examples/main/README.md) for a full list of parameters. -In order to build llama.cpp you have four different options. +### Conversation mode -- Using `make`: - - On Linux or MacOS: - - ```bash - make - ``` - - - On Windows: - - 1. Download the latest fortran version of [w64devkit](https://github.com/skeeto/w64devkit/releases). - 2. Extract `w64devkit` on your pc. - 3. Run `w64devkit.exe`. - 4. Use the `cd` command to reach the `llama.cpp` folder. - 5. From here you can run: - ```bash - make - ``` - - - Notes: - - For faster compilation, add the `-j` argument to run multiple jobs in parallel. For example, `make -j 8` will run 8 jobs in parallel. - - For faster repeated compilation, install [ccache](https://ccache.dev/). - - For debug builds, run `make LLAMA_DEBUG=1` - -- Using `CMake`: - - ```bash - cmake -B build - cmake --build build --config Release - ``` - - **Notes**: - - - For faster compilation, add the `-j` argument to run multiple jobs in parallel. For example, `cmake --build build --config Release -j 8` will run 8 jobs in parallel. - - For faster repeated compilation, install [ccache](https://ccache.dev/). - - For debug builds, there are two cases: - - 1. Single-config generators (e.g. default = `Unix Makefiles`; note that they just ignore the `--config` flag): - - ```bash - cmake -B build -DCMAKE_BUILD_TYPE=Debug - cmake --build build - ``` - - 2. Multi-config generators (`-G` param set to Visual Studio, XCode...): - - ```bash - cmake -B build -G "Xcode" - cmake --build build --config Debug - ``` - -- Using `gmake` (FreeBSD): - - 1. Install and activate [DRM in FreeBSD](https://wiki.freebsd.org/Graphics) - 2. Add your user to **video** group - 3. Install compilation dependencies. - - ```bash - sudo pkg install gmake automake autoconf pkgconf llvm15 openblas - - gmake CC=/usr/local/bin/clang15 CXX=/usr/local/bin/clang++15 -j4 - ``` - -### Homebrew - -On Mac and Linux, the homebrew package manager can be used via -``` -brew install llama.cpp -``` -The formula is automatically updated with new `llama.cpp` releases. More info: https://github.com/ggerganov/llama.cpp/discussions/7668 - -### Nix - -On Mac and Linux, the Nix package manager can be used via -``` -nix profile install nixpkgs#llama-cpp -``` -For flake enabled installs. - -Or -``` -nix-env --file '' --install --attr llama-cpp -``` -For non-flake enabled installs. - -This expression is automatically updated within the [nixpkgs repo](https://github.com/NixOS/nixpkgs/blob/nixos-24.05/pkgs/by-name/ll/llama-cpp/package.nix#L164). - -#### Flox - -On Mac and Linux, Flox can be used to install llama.cpp within a Flox environment via -``` -flox install llama-cpp -``` -Flox follows the nixpkgs build of llama.cpp. - -### Metal Build - -On MacOS, Metal is enabled by default. Using Metal makes the computation run on the GPU. -To disable the Metal build at compile time use the `GGML_NO_METAL=1` flag or the `GGML_METAL=OFF` cmake option. - -When built with Metal support, you can explicitly disable GPU inference with the `--n-gpu-layers|-ngl 0` command-line -argument. - -### BLAS Build - -Building the program with BLAS support may lead to some performance improvements in prompt processing using batch sizes higher than 32 (the default is 512). Support with CPU-only BLAS implementations doesn't affect the normal generation performance. We may see generation performance improvements with GPU-involved BLAS implementations, e.g. cuBLAS, hipBLAS. There are currently several different BLAS implementations available for build and use: - -- #### Accelerate Framework: - - This is only available on Mac PCs and it's enabled by default. You can just build using the normal instructions. - -- #### OpenBLAS: - - This provides BLAS acceleration using only the CPU. Make sure to have OpenBLAS installed on your machine. - - - Using `make`: - - On Linux: - ```bash - make GGML_OPENBLAS=1 - ``` - - - On Windows: - - 1. Download the latest fortran version of [w64devkit](https://github.com/skeeto/w64devkit/releases). - 2. Download the latest version of [OpenBLAS for Windows](https://github.com/xianyi/OpenBLAS/releases). - 3. Extract `w64devkit` on your pc. - 4. From the OpenBLAS zip that you just downloaded copy `libopenblas.a`, located inside the `lib` folder, inside `w64devkit\x86_64-w64-mingw32\lib`. - 5. From the same OpenBLAS zip copy the content of the `include` folder inside `w64devkit\x86_64-w64-mingw32\include`. - 6. Run `w64devkit.exe`. - 7. Use the `cd` command to reach the `llama.cpp` folder. - 8. From here you can run: - - ```bash - make GGML_OPENBLAS=1 - ``` - - - Using `CMake` on Linux: - - ```bash - cmake -B build -DGGML_BLAS=ON -DGGML_BLAS_VENDOR=OpenBLAS - cmake --build build --config Release - ``` - -- #### BLIS - - Check [BLIS.md](docs/BLIS.md) for more information. - -- #### SYCL - SYCL is a higher-level programming model to improve programming productivity on various hardware accelerators. - - llama.cpp based on SYCL is used to **support Intel GPU** (Data Center Max series, Flex series, Arc series, Built-in GPU and iGPU). - - For detailed info, please refer to [llama.cpp for SYCL](README-sycl.md). - -- #### Intel oneMKL - Building through oneAPI compilers will make avx_vnni instruction set available for intel processors that do not support avx512 and avx512_vnni. Please note that this build config **does not support Intel GPU**. For Intel GPU support, please refer to [llama.cpp for SYCL](./README-sycl.md). - - - Using manual oneAPI installation: - By default, `GGML_BLAS_VENDOR` is set to `Generic`, so if you already sourced intel environment script and assign `-DGGML_BLAS=ON` in cmake, the mkl version of Blas will automatically been selected. Otherwise please install oneAPI and follow the below steps: - ```bash - source /opt/intel/oneapi/setvars.sh # You can skip this step if in oneapi-basekit docker image, only required for manual installation - cmake -B build -DGGML_BLAS=ON -DGGML_BLAS_VENDOR=Intel10_64lp -DCMAKE_C_COMPILER=icx -DCMAKE_CXX_COMPILER=icpx -DGGML_NATIVE=ON - cmake --build build --config Release - ``` - - - Using oneAPI docker image: - If you do not want to source the environment vars and install oneAPI manually, you can also build the code using intel docker container: [oneAPI-basekit](https://hub.docker.com/r/intel/oneapi-basekit). Then, you can use the commands given above. - - Check [Optimizing and Running LLaMA2 on Intel® CPU](https://www.intel.com/content/www/us/en/content-details/791610/optimizing-and-running-llama2-on-intel-cpu.html) for more information. - -- #### CUDA - - This provides GPU acceleration using the CUDA cores of your Nvidia GPU. Make sure to have the CUDA toolkit installed. You can download it from your Linux distro's package manager (e.g. `apt install nvidia-cuda-toolkit`) or from here: [CUDA Toolkit](https://developer.nvidia.com/cuda-downloads). - - For Jetson user, if you have Jetson Orin, you can try this: [Offical Support](https://www.jetson-ai-lab.com/tutorial_text-generation.html). If you are using an old model(nano/TX2), need some additional operations before compiling. - - - Using `make`: - ```bash - make GGML_CUDA=1 - ``` - - Using `CMake`: - - ```bash - cmake -B build -DGGML_CUDA=ON - cmake --build build --config Release - ``` - - The environment variable [`CUDA_VISIBLE_DEVICES`](https://docs.nvidia.com/cuda/cuda-c-programming-guide/index.html#env-vars) can be used to specify which GPU(s) will be used. The following compilation options are also available to tweak performance: - - | Option | Legal values | Default | Description | - |-------------------------------|------------------------|---------|-----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------| - | GGML_CUDA_FORCE_DMMV | Boolean | false | Force the use of dequantization + matrix vector multiplication kernels instead of using kernels that do matrix vector multiplication on quantized data. By default the decision is made based on compute capability (MMVQ for 6.1/Pascal/GTX 1000 or higher). Does not affect k-quants. | - | GGML_CUDA_DMMV_X | Positive integer >= 32 | 32 | Number of values in x direction processed by the CUDA dequantization + matrix vector multiplication kernel per iteration. Increasing this value can improve performance on fast GPUs. Power of 2 heavily recommended. Does not affect k-quants. | - | GGML_CUDA_MMV_Y | Positive integer | 1 | Block size in y direction for the CUDA mul mat vec kernels. Increasing this value can improve performance on fast GPUs. Power of 2 recommended. | - | GGML_CUDA_FORCE_MMQ | Boolean | false | Force the use of custom matrix multiplication kernels for quantized models instead of FP16 cuBLAS even if there is no int8 tensor core implementation available (affects V100, RDNA3). MMQ kernels are enabled by default on GPUs with int8 tensor core support. With MMQ force enabled, speed for large batch sizes will be worse but VRAM consumption will be lower. | - | GGML_CUDA_FORCE_CUBLAS | Boolean | false | Force the use of FP16 cuBLAS instead of custom matrix multiplication kernels for quantized models | - | GGML_CUDA_F16 | Boolean | false | If enabled, use half-precision floating point arithmetic for the CUDA dequantization + mul mat vec kernels and for the q4_1 and q5_1 matrix matrix multiplication kernels. Can improve performance on relatively recent GPUs. | - | GGML_CUDA_KQUANTS_ITER | 1 or 2 | 2 | Number of values processed per iteration and per CUDA thread for Q2_K and Q6_K quantization formats. Setting this value to 1 can improve performance for slow GPUs. | - | GGML_CUDA_PEER_MAX_BATCH_SIZE | Positive integer | 128 | Maximum batch size for which to enable peer access between multiple GPUs. Peer access requires either Linux or NVLink. When using NVLink enabling peer access for larger batch sizes is potentially beneficial. | - | GGML_CUDA_FA_ALL_QUANTS | Boolean | false | Compile support for all KV cache quantization type (combinations) for the FlashAttention CUDA kernels. More fine-grained control over KV cache size but compilation takes much longer. | - -- #### hipBLAS - - This provides BLAS acceleration on HIP-supported AMD GPUs. - Make sure to have ROCm installed. - You can download it from your Linux distro's package manager or from here: [ROCm Quick Start (Linux)](https://rocm.docs.amd.com/projects/install-on-linux/en/latest/tutorial/quick-start.html#rocm-install-quick). - - - Using `make`: - ```bash - make GGML_HIPBLAS=1 - ``` - - Using `CMake` for Linux (assuming a gfx1030-compatible AMD GPU): - ```bash - HIPCXX="$(hipconfig -l)/clang" HIP_PATH="$(hipconfig -R)" \ - cmake -S . -B build -DGGML_HIPBLAS=ON -DAMDGPU_TARGETS=gfx1030 -DCMAKE_BUILD_TYPE=Release \ - && cmake --build build --config Release -- -j 16 - ``` - On Linux it is also possible to use unified memory architecture (UMA) to share main memory between the CPU and integrated GPU by setting `-DGGML_HIP_UMA=ON`. - However, this hurts performance for non-integrated GPUs (but enables working with integrated GPUs). - - Note that if you get the following error: - ``` - clang: error: cannot find ROCm device library; provide its path via '--rocm-path' or '--rocm-device-lib-path', or pass '-nogpulib' to build without ROCm device library - ``` - Try searching for a directory under `HIP_PATH` that contains the file - `oclc_abi_version_400.bc`. Then, add the following to the start of the - command: `HIP_DEVICE_LIB_PATH=`, so something - like: - ```bash - HIPCXX="$(hipconfig -l)/clang" HIP_PATH="$(hipconfig -p)" \ - HIP_DEVICE_LIB_PATH= \ - cmake -S . -B build -DGGML_HIPBLAS=ON -DAMDGPU_TARGETS=gfx1030 -DCMAKE_BUILD_TYPE=Release \ - && cmake --build build -- -j 16 - ``` - - - Using `make` (example for target gfx1030, build with 16 CPU threads): - ```bash - make -j16 GGML_HIPBLAS=1 GGML_HIP_UMA=1 AMDGPU_TARGETS=gfx1030 - ``` - - - Using `CMake` for Windows (using x64 Native Tools Command Prompt for VS, and assuming a gfx1100-compatible AMD GPU): - ```bash - set PATH=%HIP_PATH%\bin;%PATH% - cmake -S . -B build -G Ninja -DAMDGPU_TARGETS=gfx1100 -DGGML_HIPBLAS=ON -DCMAKE_C_COMPILER=clang -DCMAKE_CXX_COMPILER=clang++ -DCMAKE_BUILD_TYPE=Release - cmake --build build - ``` - Make sure that `AMDGPU_TARGETS` is set to the GPU arch you want to compile for. The above example uses `gfx1100` that corresponds to Radeon RX 7900XTX/XT/GRE. You can find a list of targets [here](https://llvm.org/docs/AMDGPUUsage.html#processors) - Find your gpu version string by matching the most significant version information from `rocminfo | grep gfx | head -1 | awk '{print $2}'` with the list of processors, e.g. `gfx1035` maps to `gfx1030`. - - - The environment variable [`HIP_VISIBLE_DEVICES`](https://rocm.docs.amd.com/en/latest/understand/gpu_isolation.html#hip-visible-devices) can be used to specify which GPU(s) will be used. - If your GPU is not officially supported you can use the environment variable [`HSA_OVERRIDE_GFX_VERSION`] set to a similar GPU, for example 10.3.0 on RDNA2 (e.g. gfx1030, gfx1031, or gfx1035) or 11.0.0 on RDNA3. - The following compilation options are also available to tweak performance (yes, they refer to CUDA, not HIP, because it uses the same code as the cuBLAS version above): - - | Option | Legal values | Default | Description | - |------------------------|------------------------|---------|------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------| - | GGML_CUDA_DMMV_X | Positive integer >= 32 | 32 | Number of values in x direction processed by the HIP dequantization + matrix vector multiplication kernel per iteration. Increasing this value can improve performance on fast GPUs. Power of 2 heavily recommended. Does not affect k-quants. | - | GGML_CUDA_MMV_Y | Positive integer | 1 | Block size in y direction for the HIP mul mat vec kernels. Increasing this value can improve performance on fast GPUs. Power of 2 recommended. Does not affect k-quants. | - | GGML_CUDA_KQUANTS_ITER | 1 or 2 | 2 | Number of values processed per iteration and per HIP thread for Q2_K and Q6_K quantization formats. Setting this value to 1 can improve performance for slow GPUs. | - -- #### Vulkan - - **With docker**: - - You don't need to install Vulkan SDK. It will be installed inside the container. - - ```sh - # Build the image - docker build -t llama-cpp-vulkan -f .devops/llama-cli-vulkan.Dockerfile . - - # Then, use it: - docker run -it --rm -v "$(pwd):/app:Z" --device /dev/dri/renderD128:/dev/dri/renderD128 --device /dev/dri/card1:/dev/dri/card1 llama-cpp-vulkan -m "/app/models/YOUR_MODEL_FILE" -p "Building a website can be done in 10 simple steps:" -n 400 -e -ngl 33 - ``` - - **Without docker**: - - Firstly, you need to make sure you have installed [Vulkan SDK](https://vulkan.lunarg.com/doc/view/latest/linux/getting_started_ubuntu.html) - - For example, on Ubuntu 22.04 (jammy), use the command below: - - ```bash - wget -qO - https://packages.lunarg.com/lunarg-signing-key-pub.asc | apt-key add - - wget -qO /etc/apt/sources.list.d/lunarg-vulkan-jammy.list https://packages.lunarg.com/vulkan/lunarg-vulkan-jammy.list - apt update -y - apt-get install -y vulkan-sdk - # To verify the installation, use the command below: - vulkaninfo - ``` - - Alternatively your package manager might be able to provide the appropriate libraries. - For example for Ubuntu 22.04 you can install `libvulkan-dev` instead. - For Fedora 40, you can install `vulkan-devel`, `glslc` and `glslang` packages. - - Then, build llama.cpp using the cmake command below: - - ```bash - cmake -B build -DGGML_VULKAN=1 - cmake --build build --config Release - # Test the output binary (with "-ngl 33" to offload all layers to GPU) - ./bin/llama-cli -m "PATH_TO_MODEL" -p "Hi you how are you" -n 50 -e -ngl 33 -t 4 - - # You should see in the output, ggml_vulkan detected your GPU. For example: - # ggml_vulkan: Using Intel(R) Graphics (ADL GT2) | uma: 1 | fp16: 1 | warp size: 32 - ``` - -### Prepare and Quantize - -> [!NOTE] -> You can use the [GGUF-my-repo](https://huggingface.co/spaces/ggml-org/gguf-my-repo) space on Hugging Face to quantise your model weights without any setup too. It is synced from `llama.cpp` main every 6 hours. - -To obtain the official LLaMA 2 weights please see the Obtaining and using the Facebook LLaMA 2 model section. There is also a large selection of pre-quantized `gguf` models available on Hugging Face. - -Note: `convert.py` has been moved to `examples/convert_legacy_llama.py` and shouldn't be used for anything other than `Llama/Llama2/Mistral` models and their derivatives. -It does not support LLaMA 3, you can use `convert_hf_to_gguf.py` with LLaMA 3 downloaded from Hugging Face. +If you want a more ChatGPT-like experience, you can run in conversation mode by passing `-cnv` as a parameter: ```bash -# obtain the official LLaMA model weights and place them in ./models -ls ./models -llama-2-7b tokenizer_checklist.chk tokenizer.model -# [Optional] for models using BPE tokenizers -ls ./models - vocab.json -# [Optional] for PyTorch .bin models like Mistral-7B -ls ./models - +llama-cli -m your_model.gguf -p "You are a helpful assistant" -cnv -# install Python dependencies -python3 -m pip install -r requirements.txt - -# convert the model to ggml FP16 format -python3 convert_hf_to_gguf.py models/mymodel/ - -# quantize the model to 4-bits (using Q4_K_M method) -./llama-quantize ./models/mymodel/ggml-model-f16.gguf ./models/mymodel/ggml-model-Q4_K_M.gguf Q4_K_M - -# update the gguf filetype to current version if older version is now unsupported -./llama-quantize ./models/mymodel/ggml-model-Q4_K_M.gguf ./models/mymodel/ggml-model-Q4_K_M-v2.gguf COPY +# Output: +# > hi, who are you? +# Hi there! I'm your helpful assistant! I'm an AI-powered chatbot designed to assist and provide information to users like you. I'm here to help answer your questions, provide guidance, and offer support on a wide range of topics. I'm a friendly and knowledgeable AI, and I'm always happy to help with anything you need. What's on your mind, and how can I assist you today? +# +# > what is 1+1? +# Easy peasy! The answer to 1+1 is... 2! ``` -### Run the quantized model +By default, the chat template will be taken from the input model. If you want to use another chat template, pass `--chat-template NAME` as a parameter. See the list of [supported templates](https://github.com/ggerganov/llama.cpp/wiki/Templates-supported-by-llama_chat_apply_template) ```bash -# start inference on a gguf model -./llama-cli -m ./models/mymodel/ggml-model-Q4_K_M.gguf -n 128 +./llama-cli -m your_model.gguf -p "You are a helpful assistant" -cnv --chat-template chatml ``` -When running the larger models, make sure you have enough disk space to store all the intermediate files. +You can also use your own template via in-prefix, in-suffix and reverse-prompt parameters: -### Running on Windows with prebuilt binaries - -You will find prebuilt Windows binaries on the release page. - -Simply download and extract the latest zip package of choice: (e.g. `llama-b1380-bin-win-avx2-x64.zip`) - -From the unzipped folder, open a terminal/cmd window here and place a pre-converted `.gguf` model file. Test out the main example like so: - -``` -.\main -m llama-2-7b.Q4_0.gguf -n 128 +```bash +./llama-cli -m your_model.gguf -p "You are a helpful assistant" -cnv --in-prefix 'User: ' --reverse-prompt 'User:' ``` -### Memory/Disk Requirements +### Web server -As the models are currently fully loaded into memory, you will need adequate disk space to save them and sufficient RAM to load them. At the moment, memory and disk requirements are the same. +[llama.cpp web server](./examples/server/README.md) is a lightweight [OpenAI API](https://github.com/openai/openai-openapi) compatible HTTP server that can be used to serve local models and easily connect them to existing clients. -| Model | Original size | Quantized size (Q4_0) | -|------:|--------------:|----------------------:| -| 7B | 13 GB | 3.9 GB | -| 13B | 24 GB | 7.8 GB | -| 30B | 60 GB | 19.5 GB | -| 65B | 120 GB | 38.5 GB | +Example usage: -### Quantization +```bash +./llama-server -m your_model.gguf --port 8080 -Several quantization methods are supported. They differ in the resulting model disk size and inference speed. - -*(outdated)* - -| Model | Measure | F16 | Q4_0 | Q4_1 | Q5_0 | Q5_1 | Q8_0 | -|------:|--------------|-------:|-------:|-------:|-------:|-------:|-------:| -| 7B | perplexity | 5.9066 | 6.1565 | 6.0912 | 5.9862 | 5.9481 | 5.9070 | -| 7B | file size | 13.0G | 3.5G | 3.9G | 4.3G | 4.7G | 6.7G | -| 7B | ms/tok @ 4th | 127 | 55 | 54 | 76 | 83 | 72 | -| 7B | ms/tok @ 8th | 122 | 43 | 45 | 52 | 56 | 67 | -| 7B | bits/weight | 16.0 | 4.5 | 5.0 | 5.5 | 6.0 | 8.5 | -| 13B | perplexity | 5.2543 | 5.3860 | 5.3608 | 5.2856 | 5.2706 | 5.2548 | -| 13B | file size | 25.0G | 6.8G | 7.6G | 8.3G | 9.1G | 13G | -| 13B | ms/tok @ 4th | - | 103 | 105 | 148 | 160 | 131 | -| 13B | ms/tok @ 8th | - | 73 | 82 | 98 | 105 | 128 | -| 13B | bits/weight | 16.0 | 4.5 | 5.0 | 5.5 | 6.0 | 8.5 | - -- [k-quants](https://github.com/ggerganov/llama.cpp/pull/1684) -- recent k-quants improvements and new i-quants - - [#2707](https://github.com/ggerganov/llama.cpp/pull/2707) - - [#2807](https://github.com/ggerganov/llama.cpp/pull/2807) - - [#4773 - 2-bit i-quants (inference)](https://github.com/ggerganov/llama.cpp/pull/4773) - - [#4856 - 2-bit i-quants (inference)](https://github.com/ggerganov/llama.cpp/pull/4856) - - [#4861 - importance matrix](https://github.com/ggerganov/llama.cpp/pull/4861) - - [#4872 - MoE models](https://github.com/ggerganov/llama.cpp/pull/4872) - - [#4897 - 2-bit quantization](https://github.com/ggerganov/llama.cpp/pull/4897) - - [#4930 - imatrix for all k-quants](https://github.com/ggerganov/llama.cpp/pull/4930) - - [#4951 - imatrix on the GPU](https://github.com/ggerganov/llama.cpp/pull/4957) - - [#4969 - imatrix for legacy quants](https://github.com/ggerganov/llama.cpp/pull/4969) - - [#4996 - k-qunats tuning](https://github.com/ggerganov/llama.cpp/pull/4996) - - [#5060 - Q3_K_XS](https://github.com/ggerganov/llama.cpp/pull/5060) - - [#5196 - 3-bit i-quants](https://github.com/ggerganov/llama.cpp/pull/5196) - - [quantization tuning](https://github.com/ggerganov/llama.cpp/pull/5320), [another one](https://github.com/ggerganov/llama.cpp/pull/5334), and [another one](https://github.com/ggerganov/llama.cpp/pull/5361) - -### Perplexity (measuring model quality) - -You can use the `perplexity` example to measure perplexity over a given prompt (lower perplexity is better). -For more information, see [https://huggingface.co/docs/transformers/perplexity](https://huggingface.co/docs/transformers/perplexity). - -The perplexity measurements in table above are done against the `wikitext2` test dataset (https://paperswithcode.com/dataset/wikitext-2), with context length of 512. -The time per token is measured on a MacBook M1 Pro 32GB RAM using 4 and 8 threads. - -#### How to run - -1. Download/extract: https://huggingface.co/datasets/ggml-org/ci/resolve/main/wikitext-2-raw-v1.zip -2. Run `./llama-perplexity -m models/7B/ggml-model-q4_0.gguf -f wiki.test.raw` -3. Output: +# Basic web UI can be accessed via browser: http://localhost:8080 +# Chat completion endpoint: http://localhost:8080/v1/chat/completions ``` -perplexity : calculating perplexity over 655 chunks -24.43 seconds per pass - ETA 4.45 hours -[1]4.5970,[2]5.1807,[3]6.0382,... -``` -And after 4.45 hours, you will have the final perplexity. ### Interactive mode -If you want a more ChatGPT-like experience, you can run in interactive mode by passing `-i` as a parameter. +> [!NOTE] +> If you prefer basic usage, please consider using conversation mode instead of interactive mode + In this mode, you can always interrupt generation by pressing Ctrl+C and entering one or more lines of text, which will be converted into tokens and appended to the current context. You can also specify a *reverse prompt* with the parameter `-r "reverse prompt string"`. This will result in user input being prompted whenever the exact tokens of the reverse prompt string are encountered in the generation. A typical use is to use a prompt that makes LLaMA emulate a chat between multiple users, say Alice and Bob, and pass `-r "Alice:"`. Here is an example of a few-shot interaction, invoked with the command @@ -827,150 +414,28 @@ If your issue is with model generation quality, then please at least scan the fo - [Aligning language models to follow instructions](https://openai.com/research/instruction-following) - [Training language models to follow instructions with human feedback](https://arxiv.org/abs/2203.02155) -### Android +## Tools -#### Build on Android using Termux -[Termux](https://github.com/termux/termux-app#installation) is a method to execute `llama.cpp` on an Android device (no root required). -``` -apt update && apt upgrade -y -apt install git make cmake -``` +### Prepare and Quantize -It's recommended to move your model inside the `~/` directory for best performance: -``` -cd storage/downloads -mv model.gguf ~/ -``` +> [!NOTE] +> You can use the [GGUF-my-repo](https://huggingface.co/spaces/ggml-org/gguf-my-repo) space on Hugging Face to quantise your model weights without any setup too. It is synced from `llama.cpp` main every 6 hours. -[Get the code](https://github.com/ggerganov/llama.cpp#get-the-code) & [follow the Linux build instructions](https://github.com/ggerganov/llama.cpp#build) to build `llama.cpp`. +To obtain the official LLaMA 2 weights please see the Obtaining and using the Facebook LLaMA 2 model section. There is also a large selection of pre-quantized `gguf` models available on Hugging Face. -#### Building the Project using Android NDK -Obtain the [Android NDK](https://developer.android.com/ndk) and then build with CMake. +Note: `convert.py` has been moved to `examples/convert_legacy_llama.py` and shouldn't be used for anything other than `Llama/Llama2/Mistral` models and their derivatives. +It does not support LLaMA 3, you can use `convert_hf_to_gguf.py` with LLaMA 3 downloaded from Hugging Face. -Execute the following commands on your computer to avoid downloading the NDK to your mobile. Alternatively, you can also do this in Termux: -``` -$ mkdir build-android -$ cd build-android -$ export NDK= -$ cmake -DCMAKE_TOOLCHAIN_FILE=$NDK/build/cmake/android.toolchain.cmake -DANDROID_ABI=arm64-v8a -DANDROID_PLATFORM=android-23 -DCMAKE_C_FLAGS=-march=armv8.4a+dotprod .. -$ make -``` +To learn more about quantizing model, [read this documentation](./examples/quantize/README.md) -Install [termux](https://github.com/termux/termux-app#installation) on your device and run `termux-setup-storage` to get access to your SD card (if Android 11+ then run the command twice). +### Perplexity (measuring model quality) -Finally, copy these built `llama` binaries and the model file to your device storage. Because the file permissions in the Android sdcard cannot be changed, you can copy the executable files to the `/data/data/com.termux/files/home/bin` path, and then execute the following commands in Termux to add executable permission: +You can use the `perplexity` example to measure perplexity over a given prompt (lower perplexity is better). +For more information, see [https://huggingface.co/docs/transformers/perplexity](https://huggingface.co/docs/transformers/perplexity). -(Assumed that you have pushed the built executable files to the /sdcard/llama.cpp/bin path using `adb push`) -``` -$cp -r /sdcard/llama.cpp/bin /data/data/com.termux/files/home/ -$cd /data/data/com.termux/files/home/bin -$chmod +x ./* -``` +To learn more how to measure perplexity using llama.cpp, [read this documentation](./examples/perplexity/README.md) -Download model [llama-2-7b-chat.Q4_K_M.gguf](https://huggingface.co/TheBloke/Llama-2-7B-Chat-GGUF/blob/main/llama-2-7b-chat.Q4_K_M.gguf), and push it to `/sdcard/llama.cpp/`, then move it to `/data/data/com.termux/files/home/model/` - -``` -$mv /sdcard/llama.cpp/llama-2-7b-chat.Q4_K_M.gguf /data/data/com.termux/files/home/model/ -``` - -Now, you can start chatting: -``` -$cd /data/data/com.termux/files/home/bin -$./llama-cli -m ../model/llama-2-7b-chat.Q4_K_M.gguf -n 128 -cml -``` - -Here's a demo of an interactive session running on Pixel 5 phone: - -https://user-images.githubusercontent.com/271616/225014776-1d567049-ad71-4ef2-b050-55b0b3b9274c.mp4 - -### Docker - -#### Prerequisites -* Docker must be installed and running on your system. -* Create a folder to store big models & intermediate files (ex. /llama/models) - -#### Images -We have three Docker images available for this project: - -1. `ghcr.io/ggerganov/llama.cpp:full`: This image includes both the main executable file and the tools to convert LLaMA models into ggml and convert into 4-bit quantization. (platforms: `linux/amd64`, `linux/arm64`) -2. `ghcr.io/ggerganov/llama.cpp:light`: This image only includes the main executable file. (platforms: `linux/amd64`, `linux/arm64`) -3. `ghcr.io/ggerganov/llama.cpp:server`: This image only includes the server executable file. (platforms: `linux/amd64`, `linux/arm64`) - -Additionally, there the following images, similar to the above: - -- `ghcr.io/ggerganov/llama.cpp:full-cuda`: Same as `full` but compiled with CUDA support. (platforms: `linux/amd64`) -- `ghcr.io/ggerganov/llama.cpp:light-cuda`: Same as `light` but compiled with CUDA support. (platforms: `linux/amd64`) -- `ghcr.io/ggerganov/llama.cpp:server-cuda`: Same as `server` but compiled with CUDA support. (platforms: `linux/amd64`) -- `ghcr.io/ggerganov/llama.cpp:full-rocm`: Same as `full` but compiled with ROCm support. (platforms: `linux/amd64`, `linux/arm64`) -- `ghcr.io/ggerganov/llama.cpp:light-rocm`: Same as `light` but compiled with ROCm support. (platforms: `linux/amd64`, `linux/arm64`) -- `ghcr.io/ggerganov/llama.cpp:server-rocm`: Same as `server` but compiled with ROCm support. (platforms: `linux/amd64`, `linux/arm64`) - -The GPU enabled images are not currently tested by CI beyond being built. They are not built with any variation from the ones in the Dockerfiles defined in [.devops/](.devops/) and the GitHub Action defined in [.github/workflows/docker.yml](.github/workflows/docker.yml). If you need different settings (for example, a different CUDA or ROCm library, you'll need to build the images locally for now). - -#### Usage - -The easiest way to download the models, convert them to ggml and optimize them is with the --all-in-one command which includes the full docker image. - -Replace `/path/to/models` below with the actual path where you downloaded the models. - -```bash -docker run -v /path/to/models:/models ghcr.io/ggerganov/llama.cpp:full --all-in-one "/models/" 7B -``` - -On completion, you are ready to play! - -```bash -docker run -v /path/to/models:/models ghcr.io/ggerganov/llama.cpp:full --run -m /models/7B/ggml-model-q4_0.gguf -p "Building a website can be done in 10 simple steps:" -n 512 -``` - -or with a light image: - -```bash -docker run -v /path/to/models:/models ghcr.io/ggerganov/llama.cpp:light -m /models/7B/ggml-model-q4_0.gguf -p "Building a website can be done in 10 simple steps:" -n 512 -``` - -or with a server image: - -```bash -docker run -v /path/to/models:/models -p 8000:8000 ghcr.io/ggerganov/llama.cpp:server -m /models/7B/ggml-model-q4_0.gguf --port 8000 --host 0.0.0.0 -n 512 -``` - -### Docker With CUDA - -Assuming one has the [nvidia-container-toolkit](https://github.com/NVIDIA/nvidia-container-toolkit) properly installed on Linux, or is using a GPU enabled cloud, `cuBLAS` should be accessible inside the container. - -#### Building Locally - -```bash -docker build -t local/llama.cpp:full-cuda -f .devops/full-cuda.Dockerfile . -docker build -t local/llama.cpp:light-cuda -f .devops/llama-cli-cuda.Dockerfile . -docker build -t local/llama.cpp:server-cuda -f .devops/llama-server-cuda.Dockerfile . -``` - -You may want to pass in some different `ARGS`, depending on the CUDA environment supported by your container host, as well as the GPU architecture. - -The defaults are: - -- `CUDA_VERSION` set to `11.7.1` -- `CUDA_DOCKER_ARCH` set to `all` - -The resulting images, are essentially the same as the non-CUDA images: - -1. `local/llama.cpp:full-cuda`: This image includes both the main executable file and the tools to convert LLaMA models into ggml and convert into 4-bit quantization. -2. `local/llama.cpp:light-cuda`: This image only includes the main executable file. -3. `local/llama.cpp:server-cuda`: This image only includes the server executable file. - -#### Usage - -After building locally, Usage is similar to the non-CUDA examples, but you'll need to add the `--gpus` flag. You will also want to use the `--n-gpu-layers` flag. - -```bash -docker run --gpus all -v /path/to/models:/models local/llama.cpp:full-cuda --run -m /models/7B/ggml-model-q4_0.gguf -p "Building a website can be done in 10 simple steps:" -n 512 --n-gpu-layers 1 -docker run --gpus all -v /path/to/models:/models local/llama.cpp:light-cuda -m /models/7B/ggml-model-q4_0.gguf -p "Building a website can be done in 10 simple steps:" -n 512 --n-gpu-layers 1 -docker run --gpus all -v /path/to/models:/models local/llama.cpp:server-cuda -m /models/7B/ggml-model-q4_0.gguf --port 8000 --host 0.0.0.0 -n 512 --n-gpu-layers 1 -``` - -### Contributing +## Contributing - Contributors can open PRs - Collaborators can push to branches in the `llama.cpp` repo and merge PRs into the `master` branch @@ -981,12 +446,17 @@ docker run --gpus all -v /path/to/models:/models local/llama.cpp:server-cuda -m - Make sure to read this: [Inference at the edge](https://github.com/ggerganov/llama.cpp/discussions/205) - A bit of backstory for those who are interested: [Changelog podcast](https://changelog.com/podcast/532) -### Docs +## Other documentations - [main (cli)](./examples/main/README.md) - [server](./examples/server/README.md) - [jeopardy](./examples/jeopardy/README.md) -- [BLIS](./docs/BLIS.md) +- [GBNF grammars](./grammars/README.md) + +**Development documentations** + +- [How to build](./docs/build.md) +- [Running on Docker](./docs/docker.md) +- [Build on Android](./docs/android.md) - [Performance troubleshooting](./docs/token_generation_performance_tips.md) - [GGML tips & tricks](https://github.com/ggerganov/llama.cpp/wiki/GGML-Tips-&-Tricks) -- [GBNF grammars](./grammars/README.md) diff --git a/docs/android.md b/docs/android.md new file mode 100644 index 000000000..cec4358d9 --- /dev/null +++ b/docs/android.md @@ -0,0 +1,56 @@ + +# Android + +## Build on Android using Termux +[Termux](https://github.com/termux/termux-app#installation) is a method to execute `llama.cpp` on an Android device (no root required). +``` +apt update && apt upgrade -y +apt install git make cmake +``` + +It's recommended to move your model inside the `~/` directory for best performance: +``` +cd storage/downloads +mv model.gguf ~/ +``` + +[Get the code](https://github.com/ggerganov/llama.cpp#get-the-code) & [follow the Linux build instructions](https://github.com/ggerganov/llama.cpp#build) to build `llama.cpp`. + +## Building the Project using Android NDK +Obtain the [Android NDK](https://developer.android.com/ndk) and then build with CMake. + +Execute the following commands on your computer to avoid downloading the NDK to your mobile. Alternatively, you can also do this in Termux: +``` +$ mkdir build-android +$ cd build-android +$ export NDK= +$ cmake -DCMAKE_TOOLCHAIN_FILE=$NDK/build/cmake/android.toolchain.cmake -DANDROID_ABI=arm64-v8a -DANDROID_PLATFORM=android-23 -DCMAKE_C_FLAGS=-march=armv8.4a+dotprod .. +$ make +``` + +Install [termux](https://github.com/termux/termux-app#installation) on your device and run `termux-setup-storage` to get access to your SD card (if Android 11+ then run the command twice). + +Finally, copy these built `llama` binaries and the model file to your device storage. Because the file permissions in the Android sdcard cannot be changed, you can copy the executable files to the `/data/data/com.termux/files/home/bin` path, and then execute the following commands in Termux to add executable permission: + +(Assumed that you have pushed the built executable files to the /sdcard/llama.cpp/bin path using `adb push`) +``` +$cp -r /sdcard/llama.cpp/bin /data/data/com.termux/files/home/ +$cd /data/data/com.termux/files/home/bin +$chmod +x ./* +``` + +Download model [llama-2-7b-chat.Q4_K_M.gguf](https://huggingface.co/TheBloke/Llama-2-7B-Chat-GGUF/blob/main/llama-2-7b-chat.Q4_K_M.gguf), and push it to `/sdcard/llama.cpp/`, then move it to `/data/data/com.termux/files/home/model/` + +``` +$mv /sdcard/llama.cpp/llama-2-7b-chat.Q4_K_M.gguf /data/data/com.termux/files/home/model/ +``` + +Now, you can start chatting: +``` +$cd /data/data/com.termux/files/home/bin +$./llama-cli -m ../model/llama-2-7b-chat.Q4_K_M.gguf -n 128 -cml +``` + +Here's a demo of an interactive session running on Pixel 5 phone: + +https://user-images.githubusercontent.com/271616/225014776-1d567049-ad71-4ef2-b050-55b0b3b9274c.mp4 diff --git a/docs/BLIS.md b/docs/backend/BLIS.md similarity index 100% rename from docs/BLIS.md rename to docs/backend/BLIS.md diff --git a/README-sycl.md b/docs/backend/SYCL.md similarity index 100% rename from README-sycl.md rename to docs/backend/SYCL.md diff --git a/docs/build.md b/docs/build.md new file mode 100644 index 000000000..cf42a4eb9 --- /dev/null +++ b/docs/build.md @@ -0,0 +1,288 @@ +# Build llama.cpp locally + +**To get the Code:** + +```bash +git clone https://github.com/ggerganov/llama.cpp +cd llama.cpp +``` + +In order to build llama.cpp you have four different options. + +- Using `make`: + - On Linux or MacOS: + + ```bash + make + ``` + + - On Windows: + + 1. Download the latest fortran version of [w64devkit](https://github.com/skeeto/w64devkit/releases). + 2. Extract `w64devkit` on your pc. + 3. Run `w64devkit.exe`. + 4. Use the `cd` command to reach the `llama.cpp` folder. + 5. From here you can run: + ```bash + make + ``` + + - Notes: + - For faster compilation, add the `-j` argument to run multiple jobs in parallel. For example, `make -j 8` will run 8 jobs in parallel. + - For faster repeated compilation, install [ccache](https://ccache.dev/). + - For debug builds, run `make LLAMA_DEBUG=1` + +- Using `CMake`: + + ```bash + cmake -B build + cmake --build build --config Release + ``` + + **Notes**: + + - For faster compilation, add the `-j` argument to run multiple jobs in parallel. For example, `cmake --build build --config Release -j 8` will run 8 jobs in parallel. + - For faster repeated compilation, install [ccache](https://ccache.dev/). + - For debug builds, there are two cases: + + 1. Single-config generators (e.g. default = `Unix Makefiles`; note that they just ignore the `--config` flag): + + ```bash + cmake -B build -DCMAKE_BUILD_TYPE=Debug + cmake --build build + ``` + + 2. Multi-config generators (`-G` param set to Visual Studio, XCode...): + + ```bash + cmake -B build -G "Xcode" + cmake --build build --config Debug + ``` + +- Using `gmake` (FreeBSD): + + 1. Install and activate [DRM in FreeBSD](https://wiki.freebsd.org/Graphics) + 2. Add your user to **video** group + 3. Install compilation dependencies. + + ```bash + sudo pkg install gmake automake autoconf pkgconf llvm15 openblas + + gmake CC=/usr/local/bin/clang15 CXX=/usr/local/bin/clang++15 -j4 + ``` + +## Metal Build + +On MacOS, Metal is enabled by default. Using Metal makes the computation run on the GPU. +To disable the Metal build at compile time use the `GGML_NO_METAL=1` flag or the `GGML_METAL=OFF` cmake option. + +When built with Metal support, you can explicitly disable GPU inference with the `--n-gpu-layers|-ngl 0` command-line +argument. + +## BLAS Build + +Building the program with BLAS support may lead to some performance improvements in prompt processing using batch sizes higher than 32 (the default is 512). Support with CPU-only BLAS implementations doesn't affect the normal generation performance. We may see generation performance improvements with GPU-involved BLAS implementations, e.g. cuBLAS, hipBLAS. There are currently several different BLAS implementations available for build and use: + +### Accelerate Framework: + + This is only available on Mac PCs and it's enabled by default. You can just build using the normal instructions. + +### OpenBLAS: + +This provides BLAS acceleration using only the CPU. Make sure to have OpenBLAS installed on your machine. + +- Using `make`: + - On Linux: + ```bash + make GGML_OPENBLAS=1 + ``` + + - On Windows: + + 1. Download the latest fortran version of [w64devkit](https://github.com/skeeto/w64devkit/releases). + 2. Download the latest version of [OpenBLAS for Windows](https://github.com/xianyi/OpenBLAS/releases). + 3. Extract `w64devkit` on your pc. + 4. From the OpenBLAS zip that you just downloaded copy `libopenblas.a`, located inside the `lib` folder, inside `w64devkit\x86_64-w64-mingw32\lib`. + 5. From the same OpenBLAS zip copy the content of the `include` folder inside `w64devkit\x86_64-w64-mingw32\include`. + 6. Run `w64devkit.exe`. + 7. Use the `cd` command to reach the `llama.cpp` folder. + 8. From here you can run: + + ```bash + make GGML_OPENBLAS=1 + ``` + +- Using `CMake` on Linux: + + ```bash + cmake -B build -DGGML_BLAS=ON -DGGML_BLAS_VENDOR=OpenBLAS + cmake --build build --config Release + ``` + +### BLIS + +Check [BLIS.md](./backend/BLIS.md) for more information. + +### SYCL + +SYCL is a higher-level programming model to improve programming productivity on various hardware accelerators. + +llama.cpp based on SYCL is used to **support Intel GPU** (Data Center Max series, Flex series, Arc series, Built-in GPU and iGPU). + +For detailed info, please refer to [llama.cpp for SYCL](./backend/SYCL.md). + +### Intel oneMKL + +Building through oneAPI compilers will make avx_vnni instruction set available for intel processors that do not support avx512 and avx512_vnni. Please note that this build config **does not support Intel GPU**. For Intel GPU support, please refer to [llama.cpp for SYCL](./backend/SYCL.md). + +- Using manual oneAPI installation: + By default, `GGML_BLAS_VENDOR` is set to `Generic`, so if you already sourced intel environment script and assign `-DGGML_BLAS=ON` in cmake, the mkl version of Blas will automatically been selected. Otherwise please install oneAPI and follow the below steps: + ```bash + source /opt/intel/oneapi/setvars.sh # You can skip this step if in oneapi-basekit docker image, only required for manual installation + cmake -B build -DGGML_BLAS=ON -DGGML_BLAS_VENDOR=Intel10_64lp -DCMAKE_C_COMPILER=icx -DCMAKE_CXX_COMPILER=icpx -DGGML_NATIVE=ON + cmake --build build --config Release + ``` + +- Using oneAPI docker image: + If you do not want to source the environment vars and install oneAPI manually, you can also build the code using intel docker container: [oneAPI-basekit](https://hub.docker.com/r/intel/oneapi-basekit). Then, you can use the commands given above. + +Check [Optimizing and Running LLaMA2 on Intel® CPU](https://www.intel.com/content/www/us/en/content-details/791610/optimizing-and-running-llama2-on-intel-cpu.html) for more information. + +### CUDA + +This provides GPU acceleration using the CUDA cores of your Nvidia GPU. Make sure to have the CUDA toolkit installed. You can download it from your Linux distro's package manager (e.g. `apt install nvidia-cuda-toolkit`) or from here: [CUDA Toolkit](https://developer.nvidia.com/cuda-downloads). + +For Jetson user, if you have Jetson Orin, you can try this: [Offical Support](https://www.jetson-ai-lab.com/tutorial_text-generation.html). If you are using an old model(nano/TX2), need some additional operations before compiling. + +- Using `make`: + ```bash + make GGML_CUDA=1 + ``` +- Using `CMake`: + + ```bash + cmake -B build -DGGML_CUDA=ON + cmake --build build --config Release + ``` + +The environment variable [`CUDA_VISIBLE_DEVICES`](https://docs.nvidia.com/cuda/cuda-c-programming-guide/index.html#env-vars) can be used to specify which GPU(s) will be used. The following compilation options are also available to tweak performance: + +| Option | Legal values | Default | Description | +|-------------------------------|------------------------|---------|-----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------| +| GGML_CUDA_FORCE_DMMV | Boolean | false | Force the use of dequantization + matrix vector multiplication kernels instead of using kernels that do matrix vector multiplication on quantized data. By default the decision is made based on compute capability (MMVQ for 6.1/Pascal/GTX 1000 or higher). Does not affect k-quants. | +| GGML_CUDA_DMMV_X | Positive integer >= 32 | 32 | Number of values in x direction processed by the CUDA dequantization + matrix vector multiplication kernel per iteration. Increasing this value can improve performance on fast GPUs. Power of 2 heavily recommended. Does not affect k-quants. | +| GGML_CUDA_MMV_Y | Positive integer | 1 | Block size in y direction for the CUDA mul mat vec kernels. Increasing this value can improve performance on fast GPUs. Power of 2 recommended. | +| GGML_CUDA_FORCE_MMQ | Boolean | false | Force the use of custom matrix multiplication kernels for quantized models instead of FP16 cuBLAS even if there is no int8 tensor core implementation available (affects V100, RDNA3). MMQ kernels are enabled by default on GPUs with int8 tensor core support. With MMQ force enabled, speed for large batch sizes will be worse but VRAM consumption will be lower. | +| GGML_CUDA_FORCE_CUBLAS | Boolean | false | Force the use of FP16 cuBLAS instead of custom matrix multiplication kernels for quantized models | +| GGML_CUDA_F16 | Boolean | false | If enabled, use half-precision floating point arithmetic for the CUDA dequantization + mul mat vec kernels and for the q4_1 and q5_1 matrix matrix multiplication kernels. Can improve performance on relatively recent GPUs. | +| GGML_CUDA_KQUANTS_ITER | 1 or 2 | 2 | Number of values processed per iteration and per CUDA thread for Q2_K and Q6_K quantization formats. Setting this value to 1 can improve performance for slow GPUs. | +| GGML_CUDA_PEER_MAX_BATCH_SIZE | Positive integer | 128 | Maximum batch size for which to enable peer access between multiple GPUs. Peer access requires either Linux or NVLink. When using NVLink enabling peer access for larger batch sizes is potentially beneficial. | +| GGML_CUDA_FA_ALL_QUANTS | Boolean | false | Compile support for all KV cache quantization type (combinations) for the FlashAttention CUDA kernels. More fine-grained control over KV cache size but compilation takes much longer. | + +### hipBLAS + +This provides BLAS acceleration on HIP-supported AMD GPUs. +Make sure to have ROCm installed. +You can download it from your Linux distro's package manager or from here: [ROCm Quick Start (Linux)](https://rocm.docs.amd.com/projects/install-on-linux/en/latest/tutorial/quick-start.html#rocm-install-quick). + +- Using `make`: + ```bash + make GGML_HIPBLAS=1 + ``` +- Using `CMake` for Linux (assuming a gfx1030-compatible AMD GPU): + ```bash + HIPCXX="$(hipconfig -l)/clang" HIP_PATH="$(hipconfig -R)" \ + cmake -S . -B build -DGGML_HIPBLAS=ON -DAMDGPU_TARGETS=gfx1030 -DCMAKE_BUILD_TYPE=Release \ + && cmake --build build --config Release -- -j 16 + ``` + On Linux it is also possible to use unified memory architecture (UMA) to share main memory between the CPU and integrated GPU by setting `-DGGML_HIP_UMA=ON`. + However, this hurts performance for non-integrated GPUs (but enables working with integrated GPUs). + + Note that if you get the following error: + ``` + clang: error: cannot find ROCm device library; provide its path via '--rocm-path' or '--rocm-device-lib-path', or pass '-nogpulib' to build without ROCm device library + ``` + Try searching for a directory under `HIP_PATH` that contains the file + `oclc_abi_version_400.bc`. Then, add the following to the start of the + command: `HIP_DEVICE_LIB_PATH=`, so something + like: + ```bash + HIPCXX="$(hipconfig -l)/clang" HIP_PATH="$(hipconfig -p)" \ + HIP_DEVICE_LIB_PATH= \ + cmake -S . -B build -DGGML_HIPBLAS=ON -DAMDGPU_TARGETS=gfx1030 -DCMAKE_BUILD_TYPE=Release \ + && cmake --build build -- -j 16 + ``` + +- Using `make` (example for target gfx1030, build with 16 CPU threads): + ```bash + make -j16 GGML_HIPBLAS=1 GGML_HIP_UMA=1 AMDGPU_TARGETS=gfx1030 + ``` + +- Using `CMake` for Windows (using x64 Native Tools Command Prompt for VS, and assuming a gfx1100-compatible AMD GPU): + ```bash + set PATH=%HIP_PATH%\bin;%PATH% + cmake -S . -B build -G Ninja -DAMDGPU_TARGETS=gfx1100 -DGGML_HIPBLAS=ON -DCMAKE_C_COMPILER=clang -DCMAKE_CXX_COMPILER=clang++ -DCMAKE_BUILD_TYPE=Release + cmake --build build + ``` + Make sure that `AMDGPU_TARGETS` is set to the GPU arch you want to compile for. The above example uses `gfx1100` that corresponds to Radeon RX 7900XTX/XT/GRE. You can find a list of targets [here](https://llvm.org/docs/AMDGPUUsage.html#processors) + Find your gpu version string by matching the most significant version information from `rocminfo | grep gfx | head -1 | awk '{print $2}'` with the list of processors, e.g. `gfx1035` maps to `gfx1030`. + + +The environment variable [`HIP_VISIBLE_DEVICES`](https://rocm.docs.amd.com/en/latest/understand/gpu_isolation.html#hip-visible-devices) can be used to specify which GPU(s) will be used. +If your GPU is not officially supported you can use the environment variable [`HSA_OVERRIDE_GFX_VERSION`] set to a similar GPU, for example 10.3.0 on RDNA2 (e.g. gfx1030, gfx1031, or gfx1035) or 11.0.0 on RDNA3. +The following compilation options are also available to tweak performance (yes, they refer to CUDA, not HIP, because it uses the same code as the cuBLAS version above): + +| Option | Legal values | Default | Description | +|------------------------|------------------------|---------|------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------| +| GGML_CUDA_DMMV_X | Positive integer >= 32 | 32 | Number of values in x direction processed by the HIP dequantization + matrix vector multiplication kernel per iteration. Increasing this value can improve performance on fast GPUs. Power of 2 heavily recommended. Does not affect k-quants. | +| GGML_CUDA_MMV_Y | Positive integer | 1 | Block size in y direction for the HIP mul mat vec kernels. Increasing this value can improve performance on fast GPUs. Power of 2 recommended. Does not affect k-quants. | +| GGML_CUDA_KQUANTS_ITER | 1 or 2 | 2 | Number of values processed per iteration and per HIP thread for Q2_K and Q6_K quantization formats. Setting this value to 1 can improve performance for slow GPUs. | + +### Vulkan + +**With docker**: + +You don't need to install Vulkan SDK. It will be installed inside the container. + +```sh +# Build the image +docker build -t llama-cpp-vulkan -f .devops/llama-cli-vulkan.Dockerfile . + +# Then, use it: +docker run -it --rm -v "$(pwd):/app:Z" --device /dev/dri/renderD128:/dev/dri/renderD128 --device /dev/dri/card1:/dev/dri/card1 llama-cpp-vulkan -m "/app/models/YOUR_MODEL_FILE" -p "Building a website can be done in 10 simple steps:" -n 400 -e -ngl 33 +``` + +**Without docker**: + +Firstly, you need to make sure you have installed [Vulkan SDK](https://vulkan.lunarg.com/doc/view/latest/linux/getting_started_ubuntu.html) + +For example, on Ubuntu 22.04 (jammy), use the command below: + +```bash +wget -qO - https://packages.lunarg.com/lunarg-signing-key-pub.asc | apt-key add - +wget -qO /etc/apt/sources.list.d/lunarg-vulkan-jammy.list https://packages.lunarg.com/vulkan/lunarg-vulkan-jammy.list +apt update -y +apt-get install -y vulkan-sdk +# To verify the installation, use the command below: +vulkaninfo +``` + +Alternatively your package manager might be able to provide the appropriate libraries. +For example for Ubuntu 22.04 you can install `libvulkan-dev` instead. +For Fedora 40, you can install `vulkan-devel`, `glslc` and `glslang` packages. + +Then, build llama.cpp using the cmake command below: + +```bash +cmake -B build -DGGML_VULKAN=1 +cmake --build build --config Release +# Test the output binary (with "-ngl 33" to offload all layers to GPU) +./bin/llama-cli -m "PATH_TO_MODEL" -p "Hi you how are you" -n 50 -e -ngl 33 -t 4 + +# You should see in the output, ggml_vulkan detected your GPU. For example: +# ggml_vulkan: Using Intel(R) Graphics (ADL GT2) | uma: 1 | fp16: 1 | warp size: 32 +``` + +### Android + +To read documentation for how to build on Android, [click here](./android.md) diff --git a/docs/HOWTO-add-model.md b/docs/development/HOWTO-add-model.md similarity index 99% rename from docs/HOWTO-add-model.md rename to docs/development/HOWTO-add-model.md index 87093cedd..2712b66c1 100644 --- a/docs/HOWTO-add-model.md +++ b/docs/development/HOWTO-add-model.md @@ -1,4 +1,4 @@ -## Add a new model architecture to `llama.cpp` +# Add a new model architecture to `llama.cpp` Adding a model requires few steps: diff --git a/docs/debugging-tests.md b/docs/development/debugging-tests.md similarity index 100% rename from docs/debugging-tests.md rename to docs/development/debugging-tests.md diff --git a/docs/llama-star/idea-arch.key b/docs/development/llama-star/idea-arch.key similarity index 100% rename from docs/llama-star/idea-arch.key rename to docs/development/llama-star/idea-arch.key diff --git a/docs/llama-star/idea-arch.pdf b/docs/development/llama-star/idea-arch.pdf similarity index 100% rename from docs/llama-star/idea-arch.pdf rename to docs/development/llama-star/idea-arch.pdf diff --git a/docs/token_generation_performance_tips.md b/docs/development/token_generation_performance_tips.md similarity index 100% rename from docs/token_generation_performance_tips.md rename to docs/development/token_generation_performance_tips.md diff --git a/docs/docker.md b/docs/docker.md new file mode 100644 index 000000000..d8922d77d --- /dev/null +++ b/docs/docker.md @@ -0,0 +1,86 @@ +# Docker + +## Prerequisites +* Docker must be installed and running on your system. +* Create a folder to store big models & intermediate files (ex. /llama/models) + +## Images +We have three Docker images available for this project: + +1. `ghcr.io/ggerganov/llama.cpp:full`: This image includes both the main executable file and the tools to convert LLaMA models into ggml and convert into 4-bit quantization. (platforms: `linux/amd64`, `linux/arm64`) +2. `ghcr.io/ggerganov/llama.cpp:light`: This image only includes the main executable file. (platforms: `linux/amd64`, `linux/arm64`) +3. `ghcr.io/ggerganov/llama.cpp:server`: This image only includes the server executable file. (platforms: `linux/amd64`, `linux/arm64`) + +Additionally, there the following images, similar to the above: + +- `ghcr.io/ggerganov/llama.cpp:full-cuda`: Same as `full` but compiled with CUDA support. (platforms: `linux/amd64`) +- `ghcr.io/ggerganov/llama.cpp:light-cuda`: Same as `light` but compiled with CUDA support. (platforms: `linux/amd64`) +- `ghcr.io/ggerganov/llama.cpp:server-cuda`: Same as `server` but compiled with CUDA support. (platforms: `linux/amd64`) +- `ghcr.io/ggerganov/llama.cpp:full-rocm`: Same as `full` but compiled with ROCm support. (platforms: `linux/amd64`, `linux/arm64`) +- `ghcr.io/ggerganov/llama.cpp:light-rocm`: Same as `light` but compiled with ROCm support. (platforms: `linux/amd64`, `linux/arm64`) +- `ghcr.io/ggerganov/llama.cpp:server-rocm`: Same as `server` but compiled with ROCm support. (platforms: `linux/amd64`, `linux/arm64`) + +The GPU enabled images are not currently tested by CI beyond being built. They are not built with any variation from the ones in the Dockerfiles defined in [.devops/](.devops/) and the GitHub Action defined in [.github/workflows/docker.yml](.github/workflows/docker.yml). If you need different settings (for example, a different CUDA or ROCm library, you'll need to build the images locally for now). + +## Usage + +The easiest way to download the models, convert them to ggml and optimize them is with the --all-in-one command which includes the full docker image. + +Replace `/path/to/models` below with the actual path where you downloaded the models. + +```bash +docker run -v /path/to/models:/models ghcr.io/ggerganov/llama.cpp:full --all-in-one "/models/" 7B +``` + +On completion, you are ready to play! + +```bash +docker run -v /path/to/models:/models ghcr.io/ggerganov/llama.cpp:full --run -m /models/7B/ggml-model-q4_0.gguf -p "Building a website can be done in 10 simple steps:" -n 512 +``` + +or with a light image: + +```bash +docker run -v /path/to/models:/models ghcr.io/ggerganov/llama.cpp:light -m /models/7B/ggml-model-q4_0.gguf -p "Building a website can be done in 10 simple steps:" -n 512 +``` + +or with a server image: + +```bash +docker run -v /path/to/models:/models -p 8000:8000 ghcr.io/ggerganov/llama.cpp:server -m /models/7B/ggml-model-q4_0.gguf --port 8000 --host 0.0.0.0 -n 512 +``` + +## Docker With CUDA + +Assuming one has the [nvidia-container-toolkit](https://github.com/NVIDIA/nvidia-container-toolkit) properly installed on Linux, or is using a GPU enabled cloud, `cuBLAS` should be accessible inside the container. + +## Building Docker locally + +```bash +docker build -t local/llama.cpp:full-cuda -f .devops/full-cuda.Dockerfile . +docker build -t local/llama.cpp:light-cuda -f .devops/llama-cli-cuda.Dockerfile . +docker build -t local/llama.cpp:server-cuda -f .devops/llama-server-cuda.Dockerfile . +``` + +You may want to pass in some different `ARGS`, depending on the CUDA environment supported by your container host, as well as the GPU architecture. + +The defaults are: + +- `CUDA_VERSION` set to `11.7.1` +- `CUDA_DOCKER_ARCH` set to `all` + +The resulting images, are essentially the same as the non-CUDA images: + +1. `local/llama.cpp:full-cuda`: This image includes both the main executable file and the tools to convert LLaMA models into ggml and convert into 4-bit quantization. +2. `local/llama.cpp:light-cuda`: This image only includes the main executable file. +3. `local/llama.cpp:server-cuda`: This image only includes the server executable file. + +## Usage + +After building locally, Usage is similar to the non-CUDA examples, but you'll need to add the `--gpus` flag. You will also want to use the `--n-gpu-layers` flag. + +```bash +docker run --gpus all -v /path/to/models:/models local/llama.cpp:full-cuda --run -m /models/7B/ggml-model-q4_0.gguf -p "Building a website can be done in 10 simple steps:" -n 512 --n-gpu-layers 1 +docker run --gpus all -v /path/to/models:/models local/llama.cpp:light-cuda -m /models/7B/ggml-model-q4_0.gguf -p "Building a website can be done in 10 simple steps:" -n 512 --n-gpu-layers 1 +docker run --gpus all -v /path/to/models:/models local/llama.cpp:server-cuda -m /models/7B/ggml-model-q4_0.gguf --port 8000 --host 0.0.0.0 -n 512 --n-gpu-layers 1 +``` diff --git a/docs/install.md b/docs/install.md new file mode 100644 index 000000000..10a568506 --- /dev/null +++ b/docs/install.md @@ -0,0 +1,39 @@ +# Install pre-built version of llama.cpp + +## Homebrew + +On Mac and Linux, the homebrew package manager can be used via + +```sh +brew install llama.cpp +``` +The formula is automatically updated with new `llama.cpp` releases. More info: https://github.com/ggerganov/llama.cpp/discussions/7668 + +## Nix + +On Mac and Linux, the Nix package manager can be used via + +```sh +nix profile install nixpkgs#llama-cpp +``` +For flake enabled installs. + +Or + +```sh +nix-env --file '' --install --attr llama-cpp +``` + +For non-flake enabled installs. + +This expression is automatically updated within the [nixpkgs repo](https://github.com/NixOS/nixpkgs/blob/nixos-24.05/pkgs/by-name/ll/llama-cpp/package.nix#L164). + +## Flox + +On Mac and Linux, Flox can be used to install llama.cpp within a Flox environment via + +```sh +flox install llama-cpp +``` + +Flox follows the nixpkgs build of llama.cpp. diff --git a/examples/quantize/README.md b/examples/quantize/README.md index b78ece4e7..553c2701b 100644 --- a/examples/quantize/README.md +++ b/examples/quantize/README.md @@ -4,7 +4,89 @@ You can also use the [GGUF-my-repo](https://huggingface.co/spaces/ggml-org/gguf- Note: It is synced from llama.cpp `main` every 6 hours. -## Llama 2 7B +Example usage: + +```bash +# obtain the official LLaMA model weights and place them in ./models +ls ./models +llama-2-7b tokenizer_checklist.chk tokenizer.model +# [Optional] for models using BPE tokenizers +ls ./models + vocab.json +# [Optional] for PyTorch .bin models like Mistral-7B +ls ./models + + +# install Python dependencies +python3 -m pip install -r requirements.txt + +# convert the model to ggml FP16 format +python3 convert_hf_to_gguf.py models/mymodel/ + +# quantize the model to 4-bits (using Q4_K_M method) +./llama-quantize ./models/mymodel/ggml-model-f16.gguf ./models/mymodel/ggml-model-Q4_K_M.gguf Q4_K_M + +# update the gguf filetype to current version if older version is now unsupported +./llama-quantize ./models/mymodel/ggml-model-Q4_K_M.gguf ./models/mymodel/ggml-model-Q4_K_M-v2.gguf COPY +``` + +Run the quantized model: + +```bash +# start inference on a gguf model +./llama-cli -m ./models/mymodel/ggml-model-Q4_K_M.gguf -n 128 +``` + +When running the larger models, make sure you have enough disk space to store all the intermediate files. + +## Memory/Disk Requirements + +As the models are currently fully loaded into memory, you will need adequate disk space to save them and sufficient RAM to load them. At the moment, memory and disk requirements are the same. + +| Model | Original size | Quantized size (Q4_0) | +|------:|--------------:|----------------------:| +| 7B | 13 GB | 3.9 GB | +| 13B | 24 GB | 7.8 GB | +| 30B | 60 GB | 19.5 GB | +| 65B | 120 GB | 38.5 GB | + +## Quantization + +Several quantization methods are supported. They differ in the resulting model disk size and inference speed. + +*(outdated)* + +| Model | Measure | F16 | Q4_0 | Q4_1 | Q5_0 | Q5_1 | Q8_0 | +|------:|--------------|-------:|-------:|-------:|-------:|-------:|-------:| +| 7B | perplexity | 5.9066 | 6.1565 | 6.0912 | 5.9862 | 5.9481 | 5.9070 | +| 7B | file size | 13.0G | 3.5G | 3.9G | 4.3G | 4.7G | 6.7G | +| 7B | ms/tok @ 4th | 127 | 55 | 54 | 76 | 83 | 72 | +| 7B | ms/tok @ 8th | 122 | 43 | 45 | 52 | 56 | 67 | +| 7B | bits/weight | 16.0 | 4.5 | 5.0 | 5.5 | 6.0 | 8.5 | +| 13B | perplexity | 5.2543 | 5.3860 | 5.3608 | 5.2856 | 5.2706 | 5.2548 | +| 13B | file size | 25.0G | 6.8G | 7.6G | 8.3G | 9.1G | 13G | +| 13B | ms/tok @ 4th | - | 103 | 105 | 148 | 160 | 131 | +| 13B | ms/tok @ 8th | - | 73 | 82 | 98 | 105 | 128 | +| 13B | bits/weight | 16.0 | 4.5 | 5.0 | 5.5 | 6.0 | 8.5 | + +- [k-quants](https://github.com/ggerganov/llama.cpp/pull/1684) +- recent k-quants improvements and new i-quants + - [#2707](https://github.com/ggerganov/llama.cpp/pull/2707) + - [#2807](https://github.com/ggerganov/llama.cpp/pull/2807) + - [#4773 - 2-bit i-quants (inference)](https://github.com/ggerganov/llama.cpp/pull/4773) + - [#4856 - 2-bit i-quants (inference)](https://github.com/ggerganov/llama.cpp/pull/4856) + - [#4861 - importance matrix](https://github.com/ggerganov/llama.cpp/pull/4861) + - [#4872 - MoE models](https://github.com/ggerganov/llama.cpp/pull/4872) + - [#4897 - 2-bit quantization](https://github.com/ggerganov/llama.cpp/pull/4897) + - [#4930 - imatrix for all k-quants](https://github.com/ggerganov/llama.cpp/pull/4930) + - [#4951 - imatrix on the GPU](https://github.com/ggerganov/llama.cpp/pull/4957) + - [#4969 - imatrix for legacy quants](https://github.com/ggerganov/llama.cpp/pull/4969) + - [#4996 - k-qunats tuning](https://github.com/ggerganov/llama.cpp/pull/4996) + - [#5060 - Q3_K_XS](https://github.com/ggerganov/llama.cpp/pull/5060) + - [#5196 - 3-bit i-quants](https://github.com/ggerganov/llama.cpp/pull/5196) + - [quantization tuning](https://github.com/ggerganov/llama.cpp/pull/5320), [another one](https://github.com/ggerganov/llama.cpp/pull/5334), and [another one](https://github.com/ggerganov/llama.cpp/pull/5361) + +**Llama 2 7B** | Quantization | Bits per Weight (BPW) | |--------------|-----------------------| @@ -18,7 +100,8 @@ Note: It is synced from llama.cpp `main` every 6 hours. | Q5_K_M | 5.68 | | Q6_K | 6.56 | -## Llama 2 13B +**Llama 2 13B** + Quantization | Bits per Weight (BPW) -- | -- Q2_K | 3.34 @@ -31,7 +114,7 @@ Q5_K_S | 5.51 Q5_K_M | 5.67 Q6_K | 6.56 -# Llama 2 70B +**Llama 2 70B** Quantization | Bits per Weight (BPW) -- | -- From 213701b51a17175d0d326b566efc03f30ec7fbe6 Mon Sep 17 00:00:00 2001 From: jaime-m-p <167997752+jaime-m-p@users.noreply.github.com> Date: Fri, 5 Jul 2024 19:01:35 +0200 Subject: [PATCH 2/3] Detokenizer fixes (#8039) * Add llama_detokenize(): - Update header files location - UNKNOWN and CONTROL are 'special pieces' - Remove space after UNKNOWN and CONTROL - Refactor llama_token_to_piece() - Add flag: clean_up_tokenization_spaces - Symmetric params for llama_tokenize() and llama_detokenize() * Update and fix tokenizer tests: - Using llama_detokenize() - Unexpected vocab type as test fail instead of error - Useful when automating tests: - If you don't know in advance the vocab type - Differenciate other loading errors - Skip unicode surrogaes and undefined - Gracefully exit threads - Using exit() is throwing random exceptions - Clean old known problematic codepoints - Minor: confusing hexadecimal codepoint * Update bruteforce random tests - Add detokenizer checks - New generator: ascii_lr_strip - New generator: apostrophe - Add more vocabs files - Detokenize special tokens. - Replace errors with '\uFFFD' when detokenizing to 'utf-8' - More edge cases - Better detokenization results check * Fix add_space_prefix, set false by default * Better leading space removal * Do not remove space when decoding special tokens * Bugfix: custom regexs splits undefined unicode codepoints * 'viking' detokenizer clean spaces --- common/common.cpp | 58 ++-- common/common.h | 16 +- examples/batched.swift/Sources/main.swift | 3 +- .../llama.cpp.swift/LibLlama.swift | 4 +- include/llama.h | 19 +- src/llama.cpp | 254 +++++++++----- src/unicode.cpp | 18 +- tests/test-tokenizer-0.cpp | 12 +- tests/test-tokenizer-1-bpe.cpp | 35 +- tests/test-tokenizer-1-spm.cpp | 33 +- tests/test-tokenizer-random.py | 312 +++++++++++++----- 11 files changed, 499 insertions(+), 265 deletions(-) diff --git a/common/common.cpp b/common/common.cpp index 42ae85f5f..c548bcb28 100644 --- a/common/common.cpp +++ b/common/common.cpp @@ -2592,51 +2592,35 @@ std::vector llama_tokenize( } std::string llama_token_to_piece(const struct llama_context * ctx, llama_token token, bool special) { - std::vector result(8, 0); - const int n_tokens = llama_token_to_piece(llama_get_model(ctx), token, result.data(), result.size(), special); - if (n_tokens < 0) { - result.resize(-n_tokens); - int check = llama_token_to_piece(llama_get_model(ctx), token, result.data(), result.size(), special); - GGML_ASSERT(check == -n_tokens); - } else { - result.resize(n_tokens); + std::string piece; + piece.resize(piece.capacity()); // using string internal cache, 15 bytes + '\n' + const int n_chars = llama_token_to_piece(llama_get_model(ctx), token, &piece[0], piece.size(), 0, special); + if (n_chars < 0) { + piece.resize(-n_chars); + int check = llama_token_to_piece(llama_get_model(ctx), token, &piece[0], piece.size(), 0, special); + GGML_ASSERT(check == -n_chars); + } + else { + piece.resize(n_chars); } - return std::string(result.data(), result.size()); + return piece; } -std::string llama_detokenize_spm(llama_context * ctx, const std::vector & tokens) { - const llama_token bos_id = llama_token_bos(llama_get_model(ctx)); - - std::string piece; - std::string result; - - for (size_t i = 0; i < tokens.size(); ++i) { - piece = llama_token_to_piece(ctx, tokens[i]); - - // remove the leading space of the first non-BOS token - if (((tokens[0] == bos_id && i == 1) || (tokens[0] != bos_id && i == 0)) && piece[0] == ' ') { - piece = piece.substr(1); - } - - result += piece; +std::string llama_detokenize(llama_context * ctx, const std::vector & tokens, bool special) { + std::string text; + text.resize(std::max(text.capacity(), tokens.size())); + int32_t n_chars = llama_detokenize(llama_get_model(ctx), tokens.data(), (int32_t)tokens.size(), &text[0], (int32_t)text.size(), false, special); + if (n_chars < 0) { + text.resize(-n_chars); + n_chars = llama_detokenize(llama_get_model(ctx), tokens.data(), (int32_t)tokens.size(), &text[0], (int32_t)text.size(), false, special); + GGML_ASSERT(n_chars <= (int32_t)text.size()); // whitespace trimming is performed after per-token detokenization } - return result; -} - -std::string llama_detokenize_bpe(llama_context * ctx, const std::vector & tokens) { - std::string piece; - std::string result; - - for (size_t i = 0; i < tokens.size(); ++i) { - piece = llama_token_to_piece(ctx, tokens[i]); - - result += piece; - } + text.resize(n_chars); // NOTE: the original tokenizer decodes bytes after collecting the pieces. - return result; + return text; } bool llama_should_add_bos_token(const llama_model * model) { diff --git a/common/common.h b/common/common.h index ac86483ff..dabf02b59 100644 --- a/common/common.h +++ b/common/common.h @@ -350,21 +350,13 @@ std::string llama_token_to_piece( llama_token token, bool special = true); -// TODO: these should be moved in llama.h C-style API under single `llama_detokenize` function -// that takes into account the tokenizer type and decides how to handle the leading space -// // detokenizes a vector of tokens into a string // should work similar to Python's `tokenizer.decode` -// removes the leading space from the first non-BOS token -std::string llama_detokenize_spm( +// optionally renders special/control tokens +std::string llama_detokenize( llama_context * ctx, - const std::vector & tokens); - -// detokenizes a vector of tokens into a string -// should work similar to Python's `tokenizer.decode` -std::string llama_detokenize_bpe( - llama_context * ctx, - const std::vector & tokens); + const std::vector & tokens, + bool special = true); // Uses the value from the model metadata if possible, otherwise // defaults to true when model type is SPM, otherwise false. diff --git a/examples/batched.swift/Sources/main.swift b/examples/batched.swift/Sources/main.swift index dbbd06da5..616494d2d 100644 --- a/examples/batched.swift/Sources/main.swift +++ b/examples/batched.swift/Sources/main.swift @@ -229,7 +229,7 @@ private func tokenize(text: String, add_bos: Bool) -> [llama_token] { private func token_to_piece(token: llama_token, buffer: inout [CChar]) -> String? { var result = [CChar](repeating: 0, count: 8) - let nTokens = llama_token_to_piece(model, token, &result, Int32(result.count), false) + let nTokens = llama_token_to_piece(model, token, &result, Int32(result.count), 0, false) if nTokens < 0 { let actualTokensCount = -Int(nTokens) result = .init(repeating: 0, count: actualTokensCount) @@ -238,6 +238,7 @@ private func token_to_piece(token: llama_token, buffer: inout [CChar]) -> String token, &result, Int32(result.count), + 0, false ) assert(check == actualTokensCount) diff --git a/examples/llama.swiftui/llama.cpp.swift/LibLlama.swift b/examples/llama.swiftui/llama.cpp.swift/LibLlama.swift index 737f882fb..2a3f9f758 100644 --- a/examples/llama.swiftui/llama.cpp.swift/LibLlama.swift +++ b/examples/llama.swiftui/llama.cpp.swift/LibLlama.swift @@ -322,7 +322,7 @@ actor LlamaContext { defer { result.deallocate() } - let nTokens = llama_token_to_piece(model, token, result, 8, false) + let nTokens = llama_token_to_piece(model, token, result, 8, 0, false) if nTokens < 0 { let newResult = UnsafeMutablePointer.allocate(capacity: Int(-nTokens)) @@ -330,7 +330,7 @@ actor LlamaContext { defer { newResult.deallocate() } - let nNewTokens = llama_token_to_piece(model, token, newResult, -nTokens, false) + let nNewTokens = llama_token_to_piece(model, token, newResult, -nTokens, 0, false) let bufferPointer = UnsafeBufferPointer(start: newResult, count: Int(nNewTokens)) return Array(bufferPointer) } else { diff --git a/include/llama.h b/include/llama.h index 7a9a25609..865ace994 100644 --- a/include/llama.h +++ b/include/llama.h @@ -904,6 +904,7 @@ extern "C" { /// @param tokens The tokens pointer must be large enough to hold the resulting tokens. /// @return Returns the number of tokens on success, no more than n_tokens_max /// @return Returns a negative number on failure - the number of tokens that would have been returned + /// @param add_special Allow to add BOS and EOS tokens if model is configured to do so. /// @param parse_special Allow tokenizing special and/or control tokens which otherwise are not exposed and treated /// as plaintext. Does not insert a leading space. LLAMA_API int32_t llama_tokenize( @@ -918,15 +919,31 @@ extern "C" { // Token Id -> Piece. // Uses the vocabulary in the provided context. // Does not write null terminator to the buffer. - // User code is responsible to remove the leading whitespace of the first non-BOS token when decoding multiple tokens. + // User can skip up to 'lstrip' leading spaces before copying (useful when encoding/decoding multiple tokens with 'add_space_prefix') // @param special If true, special tokens are rendered in the output. LLAMA_API int32_t llama_token_to_piece( const struct llama_model * model, llama_token token, char * buf, int32_t length, + int32_t lstrip, bool special); + /// @details Convert the provided tokens into text (inverse of llama_tokenize()). + /// @param text The char pointer must be large enough to hold the resulting text. + /// @return Returns the number of chars/bytes on success, no more than text_len_max. + /// @return Returns a negative number on failure - the number of chars/bytes that would have been returned. + /// @param remove_special Allow to remove BOS and EOS tokens if model is configured to do so. + /// @param unparse_special If true, special tokens are rendered in the output. + LLAMA_API int32_t llama_detokenize( + const struct llama_model * model, + const llama_token * tokens, + int32_t n_tokens, + char * text, + int32_t text_len_max, + bool remove_special, + bool unparse_special); + /// Apply chat template. Inspired by hf apply_chat_template() on python. /// Both "model" and "custom_template" are optional, but at least one is required. "custom_template" has higher precedence than "model" /// NOTE: This function does not use a jinja parser. It only support a pre-defined list of template. See more: https://github.com/ggerganov/llama.cpp/wiki/Templates-supported-by-llama_chat_apply_template diff --git a/src/llama.cpp b/src/llama.cpp index 18956d441..b770ca5bc 100644 --- a/src/llama.cpp +++ b/src/llama.cpp @@ -1995,18 +1995,19 @@ using llama_mlocks = std::vector>; // NOTE: avoid ever using this except for building the token_to_piece caches static std::string llama_token_to_piece(const struct llama_model * model, llama_token token, bool special) { - std::vector result(8, 0); - const int n_tokens = llama_token_to_piece(model, token, result.data(), result.size(), special); - if (n_tokens < 0) { - result.resize(-n_tokens); - int check = llama_token_to_piece(model, token, result.data(), result.size(), special); - GGML_ASSERT(check == -n_tokens); + std::string piece; + piece.resize(piece.capacity()); // using string internal cache + const int n_chars = llama_token_to_piece(model, token, &piece[0], piece.size(), 0, special); + if (n_chars < 0) { + piece.resize(-n_chars); + int check = llama_token_to_piece(model, token, &piece[0], piece.size(), 0, special); + GGML_ASSERT(check == -n_chars); } else { - result.resize(n_tokens); + piece.resize(n_chars); } - return std::string(result.data(), result.size()); + return piece; } static ggml_backend_buffer_type_t llama_default_buffer_type_cpu(bool host_buffer) { @@ -2586,10 +2587,11 @@ struct llama_vocab { id special_eot_id = -1; // TODO: move above after "eos_id", and here add "file separator" token // tokenizer flags - bool tokenizer_add_space_prefix = true; + bool tokenizer_add_space_prefix = false; bool tokenizer_add_bos = false; bool tokenizer_add_eos = false; bool tokenizer_ignore_merges = false; + bool tokenizer_clean_spaces = false; // clean_up_tokenization_spaces bool tokenizer_remove_extra_whitespaces = false; bool tokenizer_escape_whitespaces = true; bool tokenizer_treat_whitespace_as_suffix = false; @@ -5230,11 +5232,6 @@ static void llm_load_vocab( vocab.special_pad_id = -1; vocab.special_cls_id = -1; vocab.special_mask_id = -1; - - const int add_space_prefix_keyidx = gguf_find_key(ctx, kv(LLM_KV_TOKENIZER_ADD_PREFIX).c_str()); - if (add_space_prefix_keyidx != -1) { - vocab.tokenizer_add_space_prefix = gguf_get_val_bool(ctx, add_space_prefix_keyidx); - } // The default value of add_space_prefix is true. } else if (tokenizer_model == "bert") { vocab.type = LLAMA_VOCAB_TYPE_WPM; @@ -5246,15 +5243,9 @@ static void llm_load_vocab( vocab.special_pad_id = 0; vocab.special_cls_id = 101; vocab.special_mask_id = 103; - vocab.tokenizer_add_space_prefix = false; } else if (tokenizer_model == "gpt2") { vocab.type = LLAMA_VOCAB_TYPE_BPE; - const int add_space_prefix_keyidx = gguf_find_key(ctx, kv(LLM_KV_TOKENIZER_ADD_PREFIX).c_str()); - if (add_space_prefix_keyidx != -1) { - vocab.tokenizer_add_space_prefix = gguf_get_val_bool(ctx, add_space_prefix_keyidx); - } - // read bpe merges and populate bpe ranks const int merges_keyidx = gguf_find_key(ctx, kv(LLM_KV_TOKENIZER_MERGES).c_str()); if (merges_keyidx == -1) { @@ -5333,6 +5324,8 @@ static void llm_load_vocab( // for now, only BPE models have pre-tokenizers if (vocab.type == LLAMA_VOCAB_TYPE_BPE) { + vocab.tokenizer_add_space_prefix = false; + vocab.tokenizer_clean_spaces = true; if (tokenizer_pre.empty()) { LLAMA_LOG_WARN("%s: missing pre-tokenizer type, using: 'default'\n", __func__); LLAMA_LOG_WARN("%s: \n", __func__); @@ -5354,9 +5347,11 @@ static void llm_load_vocab( } else if ( tokenizer_pre == "deepseek-llm") { vocab.type_pre = LLAMA_VOCAB_PRE_TYPE_DEEPSEEK_LLM; + vocab.tokenizer_clean_spaces = false; } else if ( tokenizer_pre == "deepseek-coder") { vocab.type_pre = LLAMA_VOCAB_PRE_TYPE_DEEPSEEK_CODER; + vocab.tokenizer_clean_spaces = false; } else if ( tokenizer_pre == "falcon") { vocab.type_pre = LLAMA_VOCAB_PRE_TYPE_FALCON; @@ -5368,6 +5363,7 @@ static void llm_load_vocab( vocab.type_pre = LLAMA_VOCAB_PRE_TYPE_STARCODER; } else if ( tokenizer_pre == "gpt-2" || + tokenizer_pre == "phi-2" || tokenizer_pre == "jina-es" || tokenizer_pre == "jina-de" || tokenizer_pre == "jina-v2-es" || @@ -5383,6 +5379,7 @@ static void llm_load_vocab( } else if ( tokenizer_pre == "qwen2") { vocab.type_pre = LLAMA_VOCAB_PRE_TYPE_QWEN2; + vocab.tokenizer_clean_spaces = false; } else if ( tokenizer_pre == "stablelm2") { vocab.type_pre = LLAMA_VOCAB_PRE_TYPE_STABLELM2; @@ -5398,9 +5395,11 @@ static void llm_load_vocab( } else if ( tokenizer_pre == "poro-chat") { vocab.type_pre = LLAMA_VOCAB_PRE_TYPE_PORO; + vocab.tokenizer_clean_spaces = false; } else if ( tokenizer_pre == "viking") { vocab.type_pre = LLAMA_VOCAB_PRE_TYPE_VIKING; + vocab.tokenizer_clean_spaces = false; } else if ( tokenizer_pre == "jais") { vocab.type_pre = LLAMA_VOCAB_PRE_TYPE_JAIS; @@ -5409,10 +5408,14 @@ static void llm_load_vocab( } } else if (vocab.type == LLAMA_VOCAB_TYPE_SPM) { vocab.type_pre = LLAMA_VOCAB_PRE_TYPE_DEFAULT; + vocab.tokenizer_add_space_prefix = true; + vocab.tokenizer_clean_spaces = false; vocab.tokenizer_add_bos = true; vocab.tokenizer_add_eos = false; } else if (vocab.type == LLAMA_VOCAB_TYPE_WPM) { vocab.type_pre = LLAMA_VOCAB_PRE_TYPE_DEFAULT; + vocab.tokenizer_add_space_prefix = false; + vocab.tokenizer_clean_spaces = true; vocab.tokenizer_add_bos = true; vocab.tokenizer_add_eos = false; } else if (vocab.type == LLAMA_VOCAB_TYPE_UGM) { @@ -5422,6 +5425,11 @@ static void llm_load_vocab( } else { vocab.type_pre = LLAMA_VOCAB_PRE_TYPE_DEFAULT; } + + const int add_space_prefix_keyidx = gguf_find_key(ctx, kv(LLM_KV_TOKENIZER_ADD_PREFIX).c_str()); + if (add_space_prefix_keyidx != -1) { + vocab.tokenizer_add_space_prefix = gguf_get_val_bool(ctx, add_space_prefix_keyidx); + } } const int token_idx = gguf_find_key(ctx, kv(LLM_KV_TOKENIZER_LIST).c_str()); @@ -5603,7 +5611,7 @@ static void llm_load_vocab( } } - std::sort( vocab.cache_special_tokens.begin(), vocab.cache_special_tokens.end(), + std::sort(vocab.cache_special_tokens.begin(), vocab.cache_special_tokens.end(), [&] (const llama_vocab::id a, const llama_vocab::id b) { return vocab.id_to_token[a].text.size() > vocab.id_to_token[b].text.size(); } @@ -16098,7 +16106,7 @@ static std::vector llama_tokenize_internal(const llama_vocab & // tokenizer.encode('', add_special_tokens=True) returns [1] // tokenizer.encode('', add_special_tokens=False) returns [] - bool is_prev_special = false; + bool is_prev_special = true; // prefix with space if first token if (add_special && vocab.tokenizer_add_bos) { GGML_ASSERT(vocab.special_bos_id != -1); @@ -16110,10 +16118,9 @@ static std::vector llama_tokenize_internal(const llama_vocab & if (fragment.type == FRAGMENT_BUFFER_VARIANT_TYPE_RAW_TEXT) { auto raw_text = fragment.raw_text.substr(fragment.offset, fragment.length); - if (vocab.tokenizer_add_space_prefix) { - if (!output.size() || is_prev_special) { // prefix with space if first token - raw_text = " " + raw_text; - } + // prefix with space if previous is special + if (vocab.tokenizer_add_space_prefix && is_prev_special) { + raw_text = " " + raw_text; } #ifdef PRETOKENIZERDEBUG @@ -16122,6 +16129,7 @@ static std::vector llama_tokenize_internal(const llama_vocab & llm_tokenizer_spm tokenizer(vocab); llama_escape_whitespace(raw_text); tokenizer.tokenize(raw_text, output); + is_prev_special = false; } else { // if (fragment.type == FRAGMENT_BUFFER_VARIANT_TYPE_TOKEN) output.push_back(fragment.token); is_prev_special = true; @@ -20904,85 +20912,66 @@ static std::string llama_decode_text(const std::string & text) { } // does not write null-terminator to buf -int32_t llama_token_to_piece(const struct llama_model * model, llama_token token, char * buf, int32_t length, bool special) { +int32_t llama_token_to_piece(const struct llama_model * model, llama_token token, char * buf, int32_t length, int32_t lstrip, bool special) { // ref: https://github.com/ggerganov/llama.cpp/pull/7587#discussion_r1620983843 - if (!special && llama_is_control_token(model->vocab, token)) { + static const int attr_special = LLAMA_TOKEN_ATTR_UNKNOWN | LLAMA_TOKEN_ATTR_CONTROL; + const llama_token_attr attr = llama_token_get_attr(model, token); + if (!special && (attr & attr_special)) { return 0; } + // copy piece chars to output text buffer + // skip up to 'lstrip' leading spaces before copying + auto _try_copy = [=] (const char * token, size_t size) -> int32_t { + for (int32_t i = 0; i < lstrip && size && *token == ' '; ++i) { + token++; + size--; + } + if (length < (int32_t)size) { + return (int32_t) -size; + } + memcpy(buf, token, size); + return (int32_t) size; + }; + // if we have a cache - use it { const auto & cache = model->vocab.cache_token_to_piece; if (!cache.empty()) { - const auto & res = cache.at(token); - if (length < (int) res.size()) { - return -(int) res.size(); - } - memcpy(buf, res.c_str(), res.size()); - return res.size(); + const auto & result = cache.at(token); + return _try_copy(result.data(), result.size()); } } if (0 <= token && token < llama_n_vocab(model)) { + const std::string & token_text = model->vocab.id_to_token[token].text; switch (llama_vocab_get_type(model->vocab)) { case LLAMA_VOCAB_TYPE_WPM: case LLAMA_VOCAB_TYPE_SPM: case LLAMA_VOCAB_TYPE_UGM: { // NOTE: we accept all unsupported token types, // suppressing them like CONTROL tokens. - if (llama_is_normal_token(model->vocab, token)) { - std::string result = model->vocab.id_to_token[token].text; + if (attr & (attr_special | LLAMA_TOKEN_ATTR_USER_DEFINED)) { + return _try_copy(token_text.data(), token_text.size()); + } else if (attr & LLAMA_TOKEN_ATTR_NORMAL) { + std::string result = token_text; llama_unescape_whitespace(result); - if (length < (int) result.length()) { - return -(int) result.length(); - } - memcpy(buf, result.c_str(), result.length()); - return result.length(); - } else if ( - (llama_is_user_defined_token(model->vocab, token)) || - (llama_is_control_token (model->vocab, token) && special)) { - std::string result = model->vocab.id_to_token[token].text; - if (length < (int) result.length()) { - return -(int) result.length(); - } - memcpy(buf, result.c_str(), result.length()); - return result.length(); - } else if (llama_is_unknown_token(model->vocab, token)) { // NOLINT - if (length < 3) { - return -3; - } - memcpy(buf, "\xe2\x96\x85", 3); - return 3; - } else if (llama_is_byte_token(model->vocab, token)) { - if (length < 1) { - return -1; - } - buf[0] = llama_token_to_byte(model->vocab, token); - return 1; + return _try_copy(result.data(), result.size()); + } else if (attr & LLAMA_TOKEN_ATTR_BYTE) { + char byte = (char) llama_token_to_byte(model->vocab, token); + return _try_copy((char*) &byte, 1); } break; } case LLAMA_VOCAB_TYPE_BPE: { // NOTE: we accept all unsupported token types, // suppressing them like CONTROL tokens. - if (llama_is_normal_token(model->vocab, token)) { - std::string result = model->vocab.id_to_token[token].text; - result = llama_decode_text(result); - if (length < (int) result.length()) { - return -(int) result.length(); - } - memcpy(buf, result.c_str(), result.length()); - return result.length(); - } else if ( - (llama_is_user_defined_token(model->vocab, token)) || - (llama_is_control_token (model->vocab, token) && special)) { - std::string result = model->vocab.id_to_token[token].text; - if (length < (int) result.length()) { - return -(int) result.length(); - } - memcpy(buf, result.c_str(), result.length()); - return result.length(); + if (attr & (attr_special | LLAMA_TOKEN_ATTR_USER_DEFINED)) { + return _try_copy(token_text.data(), token_text.size()); + } else if (attr & LLAMA_TOKEN_ATTR_NORMAL) { + std::string result = llama_decode_text(token_text); + return _try_copy(result.data(), result.size()); } break; } @@ -20993,6 +20982,113 @@ int32_t llama_token_to_piece(const struct llama_model * model, llama_token token return 0; } +int32_t llama_detokenize( + const struct llama_model * model, + const llama_token * tokens, + int32_t n_tokens, + char * text, + int32_t text_len_max, + bool remove_special, + bool unparse_special) { + int32_t avail = text_len_max; + int32_t total = 0; + + // remove the leading space + bool remove_space = model->vocab.tokenizer_add_space_prefix; + + if (remove_special && model->vocab.tokenizer_add_bos) { + if (n_tokens > 0 && tokens[0] == model->vocab.special_bos_id) { + remove_space = false; + n_tokens--; + tokens++; + } + } + + if (remove_special && model->vocab.tokenizer_add_eos) { + if (n_tokens > 0 && tokens[n_tokens-1] == model->vocab.special_eos_id) { + n_tokens--; + } + } + + for (int32_t i = 0; i < n_tokens; ++i) { + GGML_ASSERT(avail >= 0); + int32_t n_chars = llama_token_to_piece(model, tokens[i], text, avail, remove_space, unparse_special); + remove_space = false; + if (n_chars < 0) { + avail = 0; + total -= n_chars; + } else if (n_chars > 0) { + avail -= n_chars; + text += n_chars; + total += n_chars; + } + } + + if (total > text_len_max) { + return -total; + } + + if (model->vocab.tokenizer_clean_spaces) { + text -= total; // restart text + + // first pass: characters ?!., //TODO: where do these characters come from? + const int32_t total1 = total; + total = total ? 1 : 0; + for (int32_t i = 1; i < total1; ++i) { + const char x = text[i]; + if (text[i - 1] == ' ') { + if (x == '?' || x == '!' || x == '.' || x == ',') { // " ?", " !", " .", " ," + total--; // remove space + } + } + text[total++] = x; + } + + // second pass: strip single apostrophe between spaces + const int32_t total2 = total; + total = total ? 1 : 0; + for (int32_t i = 1; i < total2; ++i) { + const char x = text[i]; + if (x == '\'' && i + 1 < total2 && text[i - 1] == ' ' && text[i + 1] == ' ') { // " ' " + total--; // remove prev space + text[++i] = '\0'; // remove next space + } + text[total++] = x; + } + + // third pass: apostrophe contractions //NOTE: this makes sense? + const int32_t total3 = total; + total = total ? 1 : 0; + for (int32_t i = 1; i < total3; ++i) { + const char x = text[i]; + if (text[i - 1] == ' ') { + if (x == '\'' && i + 1 < total3) { + const char x1 = text[i + 1]; + if (x1 == 't' || x1 == 'd') { // " 't", " 'd" + //total--; // remove space + } else if (x1 == 's' || x1 == 'm') { // " 's", " 'm" + total--; // remove space + } else if (i + 2 < total3) { + const char x2 = text[i + 2]; + if ((x1 == 'l' && x2 == 'l')) { // " 'll" + //total--; // remove space + } else if ((x1 == 'r' && x2 == 'e') || (x1 == 'v' && x2 == 'e')) { // " 're", " 've" + total--; // remove space + } else { + //total--; // remove space + } + } else { + //total--; // remove space + } + } + } + text[total++] = x; + } + } + + return total <= text_len_max ? total : -total; +} + // trim whitespace from the beginning and end of a string static std::string trim(const std::string & str) { size_t start = 0; diff --git a/src/unicode.cpp b/src/unicode.cpp index 8692924b9..51daa15af 100644 --- a/src/unicode.cpp +++ b/src/unicode.cpp @@ -232,8 +232,7 @@ static std::vector unicode_regex_split_custom_gpt2(const std::string & t }; auto _get_flags = [&] (const size_t pos) -> codepoint_flags { - static const codepoint_flags undef(codepoint_flags::UNDEFINED); - return (offset_ini <= pos && pos < offset_end) ? unicode_cpt_flags(cpts[pos]) : undef; + return (offset_ini <= pos && pos < offset_end) ? unicode_cpt_flags(cpts[pos]) : codepoint_flags{}; }; size_t _prev_end = offset_ini; @@ -295,9 +294,9 @@ static std::vector unicode_regex_split_custom_gpt2(const std::string & t continue; } // regex: ?[^\s\p{L}\p{N}]+ - if (!(flags2.is_whitespace || flags2.is_letter || flags2.is_number || flags2.is_undefined)) { + if (!(flags2.is_whitespace | flags2.is_letter | flags2.is_number) && flags2.as_uint()) { pos += (cpt == ' '); - while (!(flags2.is_whitespace || flags2.is_letter || flags2.is_number || flags2.is_undefined)) { + while (!(flags2.is_whitespace | flags2.is_letter | flags2.is_number) && flags2.as_uint()) { flags2 = _get_flags(++pos); } _add_token(pos); @@ -351,8 +350,7 @@ static std::vector unicode_regex_split_custom_llama3(const std::string & }; auto _get_flags = [&] (const size_t pos) -> codepoint_flags { - static const codepoint_flags undef(codepoint_flags::UNDEFINED); - return (offset_ini <= pos && pos < offset_end) ? unicode_cpt_flags(cpts[pos]) : undef; + return (offset_ini <= pos && pos < offset_end) ? unicode_cpt_flags(cpts[pos]) : codepoint_flags{}; }; size_t _prev_end = offset_ini; @@ -394,8 +392,8 @@ static std::vector unicode_regex_split_custom_llama3(const std::string & } } - // regex: [^\r\n\p{L}\p{N}]?\p{L}+ //####FIXME: the first \p{L} is correct? - if (!(cpt == '\r' || cpt == '\n' || /*flags.is_letter |*/ flags.is_number)) { + // regex: [^\r\n\p{L}\p{N}]?\p{L}+ + if (!(cpt == '\r' || cpt == '\n' || flags.is_number)) { if (flags.is_letter || _get_flags(pos+1).is_letter) { // one or more letters pos++; while (_get_flags(pos).is_letter) { @@ -421,9 +419,9 @@ static std::vector unicode_regex_split_custom_llama3(const std::string & // regex: ?[^\s\p{L}\p{N}]+[\r\n]* auto flags2 = (cpt == ' ' ? _get_flags(pos+1) : flags); - if (!(flags2.is_whitespace || flags2.is_letter || flags2.is_number || flags2.is_undefined)) { + if (!(flags2.is_whitespace | flags2.is_letter | flags2.is_number) && flags.as_uint()) { pos += (cpt == ' '); - while (!(flags2.is_whitespace || flags2.is_letter || flags2.is_number || flags2.is_undefined)) { + while (!(flags2.is_whitespace | flags2.is_letter | flags2.is_number) && flags2.as_uint()) { flags2 = _get_flags(++pos); } uint32_t cpt2 = _get_cpt(pos); diff --git a/tests/test-tokenizer-0.cpp b/tests/test-tokenizer-0.cpp index d478f1041..1f04b6f34 100644 --- a/tests/test-tokenizer-0.cpp +++ b/tests/test-tokenizer-0.cpp @@ -195,11 +195,11 @@ int main(int argc, char **argv) { const bool add_special = false; for (const auto & test_kv : k_tests) { - const std::vector res = llama_tokenize(ctx, test_kv.first, add_special); + const std::vector res = llama_tokenize(ctx, test_kv.first, add_special, true); printf("\n"); printf("src: '%s'\n", test_kv.first.c_str()); - printf("res: '%s'\n", llama_detokenize_bpe(ctx, res).c_str()); + printf("res: '%s'\n", llama_detokenize(ctx, res).c_str()); printf("tok: "); for (const auto & tok : res) { printf("%d ", tok); @@ -216,8 +216,8 @@ int main(int argc, char **argv) { if (!correct) { fprintf(stderr, "%s : failed test: '%s'\n", __func__, test_kv.first.c_str()); fprintf(stderr, "%s : detokenized to: '%s' instead of '%s'\n", __func__, - llama_detokenize_bpe(ctx, res).c_str(), - llama_detokenize_bpe(ctx, test_kv.second).c_str()); + llama_detokenize(ctx, res).c_str(), + llama_detokenize(ctx, test_kv.second).c_str()); fprintf(stderr, "%s : expected tokens: ", __func__); for (const auto & t : test_kv.second) { fprintf(stderr, "%6d '%s', ", t, llama_token_to_piece(ctx, t).c_str()); @@ -253,7 +253,7 @@ int main(int argc, char **argv) { { const auto t_start = ggml_time_us(); - res = llama_tokenize(ctx, text, add_special); + res = llama_tokenize(ctx, text, add_special, true); const auto t_end = ggml_time_us(); @@ -272,7 +272,7 @@ int main(int argc, char **argv) { } for (const auto & tok : res) { - //ofs << tok << " '" << string_strip(llama_detokenize_bpe(ctx, std::vector{tok})) << "'" << std::endl; + //ofs << tok << " '" << string_strip(llama_detokenize(ctx, std::vector{tok})) << "'" << std::endl; ofs << tok << "\n"; } } diff --git a/tests/test-tokenizer-1-bpe.cpp b/tests/test-tokenizer-1-bpe.cpp index 209a04ad6..9498387e0 100644 --- a/tests/test-tokenizer-1-bpe.cpp +++ b/tests/test-tokenizer-1-bpe.cpp @@ -11,6 +11,7 @@ #include #include #include +#include int main(int argc, char **argv) { if (argc < 2 || argc > 3) { @@ -63,7 +64,10 @@ int main(int argc, char **argv) { } } - GGML_ASSERT(llama_vocab_type(model) == LLAMA_VOCAB_TYPE_BPE); + //GGML_ASSERT(llama_vocab_type(model) == LLAMA_VOCAB_TYPE_BPE); + if (llama_vocab_type(model) != LLAMA_VOCAB_TYPE_BPE) { + return 99; + } #ifdef _WIN32 // We need this for unicode console support @@ -74,7 +78,7 @@ int main(int argc, char **argv) { const int n_vocab = llama_n_vocab(model); for (int i = 0; i < n_vocab; ++i) { - std::string str = llama_detokenize_bpe(ctx, std::vector(1, i)); + std::string str = llama_detokenize(ctx, std::vector(1, i)); try { auto cps = unicode_cpts_from_utf8(str); std::vector tokens = llama_tokenize(ctx, str, false, true); @@ -90,7 +94,7 @@ int main(int argc, char **argv) { fprintf(stderr, "]\n"); return 2; } - std::string check = llama_detokenize_bpe(ctx, tokens); + std::string check = llama_detokenize(ctx, tokens); if (check != str) { fprintf(stderr, "%s : error: token %d detokenizes to '%s'(%zu) but tokenization of this detokenizes to '%s'(%zu)\n", __func__, i, str.c_str(), str.length(), check.c_str(), check.length()); @@ -108,26 +112,23 @@ int main(int argc, char **argv) { std::vector threads(nthread); + std::atomic_int errcode = {}; + for (int i = 0; i < nthread; ++i) { - threads[i] = std::thread([i, nthread, ctx]() { - for (uint32_t cp = i; cp < 0x0010ffff; cp += nthread) { - if (!( // NOLINT - (cp < 0x03 || cp > 0x05) && cp != 0x0b && cp != 0x11 && - (cp < 0x13 || cp > 0x17) && cp != 0x19 && - (cp < 0x1c || cp > 0x1e) && - (cp < 0xd800 || cp > 0xdfff) && - (cp < 0x00040000 || cp >= 0x000e0000) - )) { + threads[i] = std::thread([i, nthread, ctx, &errcode]() { + for (uint32_t cp = i; !errcode && cp < 0x00110000; cp += nthread) { + if ((0x0000D800 <= cp && cp <= 0x0000DFFF) || // surrogates \p{Cs} + (0x00040000 <= cp && cp <= 0x000E0000)) { // undefined \p{Cn} continue; } std::string str = unicode_cpt_to_utf8(cp); std::vector tokens = llama_tokenize(ctx, str, false); - std::string check = llama_detokenize_bpe(ctx, tokens); + std::string check = llama_detokenize(ctx, tokens); if (cp != 9601 && str != check) { - fprintf(stderr, "error: codepoint %x detokenizes to '%s'(%zu) instead of '%s'(%zu)\n", + fprintf(stderr, "error: codepoint 0x%x detokenizes to '%s'(%zu) instead of '%s'(%zu)\n", cp, check.c_str(), check.length(), str.c_str(), str.length()); - std::exit(3); + errcode = 3; } } }); @@ -136,6 +137,10 @@ int main(int argc, char **argv) { for (auto & t : threads) { t.join(); } + + if (errcode) { + return errcode; + } } llama_free_model(model); diff --git a/tests/test-tokenizer-1-spm.cpp b/tests/test-tokenizer-1-spm.cpp index ac2333dda..7ca9e2ca6 100644 --- a/tests/test-tokenizer-1-spm.cpp +++ b/tests/test-tokenizer-1-spm.cpp @@ -11,6 +11,7 @@ #include #include #include +#include int main(int argc, char ** argv) { if (argc < 2) { @@ -51,7 +52,10 @@ int main(int argc, char ** argv) { } } - GGML_ASSERT(llama_vocab_type(model) == LLAMA_VOCAB_TYPE_SPM); + //GGML_ASSERT(llama_vocab_type(model) == LLAMA_VOCAB_TYPE_SPM); + if (llama_vocab_type(model) != LLAMA_VOCAB_TYPE_SPM) { + return 99; + } #ifdef _WIN32 // We need this for unicode console support @@ -62,9 +66,9 @@ int main(int argc, char ** argv) { const int n_vocab = llama_n_vocab(model); for (int i = 0; i < n_vocab; ++i) { - std::string str = llama_detokenize_spm(ctx, std::vector(1, i)); - std::vector tokens = llama_tokenize(ctx, str, false); - std::string check = llama_detokenize_spm(ctx, tokens); + std::string str = llama_detokenize(ctx, std::vector(1, i), true); + std::vector tokens = llama_tokenize(ctx, str, false, true); + std::string check = llama_detokenize(ctx, tokens); if (check != str) { fprintf(stderr, "%s : error: token %d detokenizes to '%s'(%zu) but tokenization of this detokenizes to '%s'(%zu)\n", __func__, i, str.c_str(), str.length(), check.c_str(), check.length()); @@ -78,20 +82,23 @@ int main(int argc, char ** argv) { std::vector threads(nthread); + std::atomic_int errcode = {}; + for (int i = 0; i < nthread; ++i) { - threads[i] = std::thread([i, nthread, ctx]() { - for (uint32_t cp = i; cp < 0x0010ffff; cp += nthread) { - if (cp >= 0xd800 && cp <= 0xdfff) { + threads[i] = std::thread([i, nthread, ctx, &errcode]() { + for (uint32_t cp = i; !errcode && cp < 0x00110000; cp += nthread) { + if ((0x0000D800 <= cp && cp <= 0x0000DFFF) || // surrogates \p{Cs} + (0x00040000 <= cp && cp <= 0x000E0000)) { // undefined \p{Cn} continue; } std::string str = unicode_cpt_to_utf8(cp); - std::vector tokens = llama_tokenize(ctx, str, false); - std::string check = llama_detokenize_spm(ctx, tokens); + std::vector tokens = llama_tokenize(ctx, str, false, true); + std::string check = llama_detokenize(ctx, tokens); if (cp != 9601 && str != check) { - fprintf(stderr, "error: codepoint %x detokenizes to '%s'(%zu) instead of '%s'(%zu)\n", + fprintf(stderr, "error: codepoint 0x%x detokenizes to '%s'(%zu) instead of '%s'(%zu)\n", cp, check.c_str(), check.length(), str.c_str(), str.length()); - std::exit(3); + errcode = 3; } } }); @@ -100,6 +107,10 @@ int main(int argc, char ** argv) { for (auto & t : threads) { t.join(); } + + if(errcode) { + return errcode; + } } llama_free_model(model); diff --git a/tests/test-tokenizer-random.py b/tests/test-tokenizer-random.py index a07c52fb3..48cab8a1e 100644 --- a/tests/test-tokenizer-random.py +++ b/tests/test-tokenizer-random.py @@ -13,7 +13,7 @@ import subprocess import random import unicodedata -from typing import Callable, Iterator +from typing import Iterator import cffi from transformers import AutoTokenizer @@ -24,17 +24,20 @@ logger = logging.getLogger("test-tokenizer-random") class LibLlama: - DEFAULT_PATH_LLAMA_H = "./llama.h" - DEFAULT_PATH_LIBLLAMA = "./build/libllama.so" # CMakeLists.txt: BUILD_SHARED_LIBS ON + DEFAULT_PATH_LLAMA_H = "./include/llama.h" + DEFAULT_PATH_INCLUDES = ["./ggml/include/", "./include/"] + DEFAULT_PATH_LIBLLAMA = "./build/src/libllama.so" # CMakeLists.txt: BUILD_SHARED_LIBS ON - def __init__(self, path_llama_h: str = None, path_libllama: str = None): + def __init__(self, path_llama_h: str = None, path_includes: list[str] = [], path_libllama: str = None): path_llama_h = path_llama_h or self.DEFAULT_PATH_LLAMA_H + path_includes = path_includes or self.DEFAULT_PATH_INCLUDES path_libllama = path_libllama or self.DEFAULT_PATH_LIBLLAMA - (self.ffi, self.lib) = self._load_libllama_cffi(path_llama_h, path_libllama) + (self.ffi, self.lib) = self._load_libllama_cffi(path_llama_h, path_includes, path_libllama) self.lib.llama_backend_init() - def _load_libllama_cffi(self, path_llama_h: str, path_libllama: str): - cmd = ["gcc", "-E", "-P", "-D__restrict=", "-D__attribute__(x)=", "-D__asm__(x)=", path_llama_h] + def _load_libllama_cffi(self, path_llama_h: str, path_includes: list[str], path_libllama: str): + cmd = ["gcc", "-E", "-P", "-D__restrict=", "-D__attribute__(x)=", "-D__asm__(x)="] + cmd += ["-I" + path for path in path_includes] + [path_llama_h] res = subprocess.run(cmd, stdout=subprocess.PIPE) assert (res.returncode == 0) source = res.stdout.decode() @@ -79,6 +82,7 @@ class LibLlamaModel: raise RuntimeError("error: failed to create context for model '%s'" % path_model) n_tokens_max = self.lib.llama_n_ctx(self.ctx) self.token_ids = self.ffi.new("llama_token[]", n_tokens_max) + self.text_buff = self.ffi.new("uint8_t[]", 1024) def free(self): if self.ctx: @@ -89,14 +93,78 @@ class LibLlamaModel: self.model = None self.lib = None - def tokenize(self, text: str, n_tokens_max: int = 0, add_special: bool = False, parse_special: bool = False) -> list[int]: - n_tokens_max = n_tokens_max if n_tokens_max > 0 else len(self.token_ids) + def tokenize(self, text: str, add_special: bool = False, parse_special: bool = False) -> list[int]: text = text.encode("utf-8") - num = self.lib.llama_tokenize(self.model, text, len(text), self.token_ids, n_tokens_max, add_special, parse_special) - if num < 0: - return [] + num = self.lib.llama_tokenize(self.model, text, len(text), self.token_ids, len(self.token_ids), add_special, parse_special) + while num < 0 and len(self.token_ids) < (16 << 20): + self.token_ids = self.ffi.new("llama_token[]", -2 * num) + num = self.lib.llama_tokenize(self.model, text, len(text), self.token_ids, len(self.token_ids), add_special, parse_special) return list(self.token_ids[0:num]) + def detokenize(self, ids: list[int], remove_special: bool = False, unparse_special: bool = False) -> str: + if len(self.token_ids) < len(ids): + self.token_ids = self.ffi.new("llama_token[]", 2 * len(ids)) + for i, id in enumerate(ids): + self.token_ids[i] = id + num = self.lib.llama_detokenize(self.model, self.token_ids, len(ids), self.text_buff, len(self.text_buff), remove_special, unparse_special) + while num < 0 and len(self.text_buff) < (16 << 20): + self.text_buff = self.ffi.new("uint8_t[]", -2 * num) + num = self.lib.llama_detokenize(self.model, self.token_ids, len(ids), self.text_buff, len(self.text_buff), remove_special, unparse_special) + return str(self.ffi.buffer(self.text_buff, num), encoding="utf-8", errors="replace") # replace errors with '\uFFFD' + + +class Tokenizer: + + def encode(self, text: str) -> list[int]: + raise NotImplementedError + + def decode(self, ids: list[int]) -> str: + raise NotImplementedError + + +class TokenizerGroundtruth (Tokenizer): + + def __init__(self, dir_tokenizer: str): + self.model = AutoTokenizer.from_pretrained(dir_tokenizer) + # guess BOS and EOS + ids = self.encode("a") + assert 1 <= len(ids) <= 3 + add_bos_token = len(ids) > 1 and self.model.bos_token_id == ids[0] + add_eos_token = len(ids) > 1 and self.model.eos_token_id == ids[-1] + self.add_bos_token = getattr(self.model, "add_bos_token", add_bos_token) + self.add_eos_token = getattr(self.model, "add_eos_token", add_eos_token) + # build vocab + tokens = list(self.model.get_vocab().values()) + self.vocab = self.model.batch_decode(tokens, skip_special_tokens=True) + self.vocab = list(sorted(self.vocab)) + # tokens and lists + self.special_tokens = list(self.model.all_special_tokens) + self.added_tokens = list(self.model.added_tokens_encoder) + self.bos_token = self.model.bos_token + self.eos_token = self.model.eos_token + + def encode(self, text: str) -> list[int]: + return self.model.encode(text, add_special_tokens=True) + + def decode(self, ids: list[int]) -> str: + return self.model.decode(ids, skip_special_tokens=False) + + +class TokenizerLlamaCpp (Tokenizer): + + libllama: LibLlama = None + + def __init__(self, vocab_file: str): + if not self.libllama: + self.libllama = LibLlama() + self.model = LibLlamaModel(self.libllama, vocab_file, mparams=dict(vocab_only=True), cparams=dict(n_ctx=4096)) + + def encode(self, text: str) -> list[int]: + return self.model.tokenize(text, add_special=True, parse_special=True) + + def decode(self, ids: list[int]) -> str: + return self.model.detokenize(ids, remove_special=False, unparse_special=True) + def generator_custom_text() -> Iterator[str]: """General tests""" @@ -165,19 +233,48 @@ def generator_custom_text_edge_cases() -> Iterator[str]: 'a b', # rstrip phi-3 'a b', # lstrip jina-v2 '\xa0aC', # deepseek + '\u2029 \uA3E4', # deepseek-llm + "a ?", + 'å', # mpt + '\U000ac517', # utf-8 encode error, falcon + '\U000522f4', # utf-8 encode error, starcoder + "abcd", + " abcd", ] -def generator_vocab_words(vocab: list[str]) -> Iterator[str]: +def generator_vocab_words(tokenizer: TokenizerGroundtruth) -> Iterator[str]: """Brute force check all vocab words""" - yield from vocab + yield from tokenizer.vocab -def generator_added_lr_strip(tokenizer) -> Iterator[str]: - WHITESPACES = ["", " ", " ", " "] - special_tokens = list(tokenizer.all_special_tokens) - added_tokens = list(tokenizer.added_tokens_encoder) - all_tokens = list(sorted(set(special_tokens + added_tokens))) +def generator_ascii_lr_strip() -> Iterator[str]: + WHITESPACES = ["", " ", " "] + CHARACTERS = list(chr(i) for i in range(1, 0x80)) + [""] + for char1 in CHARACTERS: + for char2 in CHARACTERS: + for lstrip in WHITESPACES: + for rstrip in WHITESPACES: + yield lstrip + char1 + char2 + rstrip + yield lstrip + char1 + rstrip + char2 + yield char1 + lstrip + char2 + rstrip + + +def generator_apostrophe() -> Iterator[str]: + WHITESPACES = ["", " ", " "] + CHARACTERS = list(chr(i) for i in range(1, 0x80)) + [""] + for char1 in CHARACTERS: + for char2 in CHARACTERS: + for lstrip in WHITESPACES: + for rstrip in WHITESPACES: + yield char1 + lstrip + "'" + rstrip + char2 + yield char1 + char2 + lstrip + "'" + rstrip + "z" + yield "a" + lstrip + "'" + rstrip + char1 + char2 + + +def generator_added_lr_strip(tokenizer: TokenizerGroundtruth) -> Iterator[str]: + WHITESPACES = ["", " ", " ", "\n", "\r\n", "\n\n", "\t", "\t\t"] + all_tokens = list(sorted(set(tokenizer.special_tokens + tokenizer.added_tokens))) for token in all_tokens: for lstrip in WHITESPACES: for rstrip in WHITESPACES: @@ -187,11 +284,9 @@ def generator_added_lr_strip(tokenizer) -> Iterator[str]: yield "a" + lstrip + token + rstrip + "z" -def generator_random_added_tokens(tokenizer, iterations=100) -> Iterator[str]: - special_tokens = list(tokenizer.all_special_tokens) - added_tokens = list(tokenizer.added_tokens_encoder) - separations = [" ", "\n", "\t", "-", "!", "one", "1", "", ""] - all_tokens = list(sorted(set(special_tokens + added_tokens + separations))) +def generator_random_added_tokens(tokenizer: TokenizerGroundtruth, iterations=100) -> Iterator[str]: + separations = [" ", "\n", "\t", "-", "!", "one", "1", "", ""] + all_tokens = list(sorted(set(tokenizer.special_tokens + tokenizer.added_tokens + separations))) rand = random.Random() for m in range(iterations): rand.seed(m) @@ -242,13 +337,13 @@ def generator_unicodes() -> Iterator[str]: def _valid(cpt): if cpt >= 0x30000: # unassigned and supplement­ary return False - if 0x00D800 <= cpt <= 0x00F8FF: # Surrogates - return False - if unicodedata.category(chr(cpt)) == "Cn": + # if cpt == 0x2029: # deepseek-llm + # return False + if unicodedata.category(chr(cpt)) in ("Cn", "Cs", "Co"): # undefined, surrogates, private return False return True - characters = [chr(cpt) for cpt in range(1, MAX_CODEPOINTS) if _valid(cpt)] + characters = [chr(cpt) for cpt in range(0, MAX_CODEPOINTS) if _valid(cpt)] yield from characters @@ -273,11 +368,11 @@ def generator_random_unicodes(iterations=100) -> Iterator[str]: yield "".join(text) -def generator_random_vocab_chars(vocab: list[str], iterations=100) -> Iterator[str]: +def generator_random_vocab_chars(tokenizer: TokenizerGroundtruth, iterations=100) -> Iterator[str]: """Brute force random text with vocab characters""" vocab_chars = set() - for word in vocab: + for word in tokenizer.vocab: vocab_chars.update(word) vocab_chars = list(sorted(vocab_chars)) @@ -288,10 +383,10 @@ def generator_random_vocab_chars(vocab: list[str], iterations=100) -> Iterator[s yield "".join(text) -def generator_random_vocab_words(vocab: list[str], iterations=100) -> Iterator[str]: +def generator_random_vocab_words(tokenizer: TokenizerGroundtruth, iterations=100) -> Iterator[str]: """Brute force random text from vocab words""" - vocab = [w.strip() for w in vocab] + vocab = [w.strip() for w in tokenizer.vocab] yield from vocab rand = random.Random() @@ -307,7 +402,7 @@ def generator_random_vocab_words(vocab: list[str], iterations=100) -> Iterator[s yield "".join(text) -def compare_tokenizers(func_tokenize1: Callable, func_tokenize2: Callable, generator: Iterator[str]): +def compare_tokenizers(tokenizer1: TokenizerGroundtruth, tokenizer2: TokenizerLlamaCpp, generator: Iterator[str]): def find_first_mismatch(ids1: list[int], ids2: list[int]): for i, (a, b) in enumerate(zip(ids1, ids2)): @@ -317,34 +412,67 @@ def compare_tokenizers(func_tokenize1: Callable, func_tokenize2: Callable, gener return -1 return min(len(ids1), len(ids2)) - t_tokenizer1 = 0 - t_tokenizer2 = 0 + def check_detokenizer(text: str, text1: str, text2: str) -> bool: + if text1 == text2: # equal to TokenizerGroundtruth? + return True + # equal to source text? + if tokenizer1.add_bos_token: # remove BOS + if text2.startswith(tokenizer1.bos_token): + text2 = text2[len(tokenizer1.bos_token):] + if tokenizer1.add_eos_token: # remove EOS + if text2.endswith(tokenizer1.eos_token): + text2 = text2[:-len(tokenizer1.eos_token)] + return text == text2 + + t_encode1 = 0 + t_encode2 = 0 + t_decode1 = 0 + t_decode2 = 0 t_start = time.perf_counter() - num_errors = 10 + encode_errors = 0 + decode_errors = 0 + MAX_ERRORS = 10 logger.info("%s: %s" % (generator.__name__, "ini")) for text in generator: + # print(repr(text), text.encode()) # print(repr(text), hex(ord(text[0])), text.encode()) t0 = time.perf_counter() - ids1 = func_tokenize1(text) + ids1 = tokenizer1.encode(text) t1 = time.perf_counter() - ids2 = func_tokenize2(text) + ids2 = tokenizer2.encode(text) t2 = time.perf_counter() - t_tokenizer1 += t1 - t0 - t_tokenizer2 += t2 - t1 - if ids1 != ids2: + text1 = tokenizer1.decode(ids1) + t3 = time.perf_counter() + text2 = tokenizer2.decode(ids1) + t4 = time.perf_counter() + t_encode1 += t1 - t0 + t_encode2 += t2 - t1 + t_decode1 += t3 - t2 + t_decode2 += t4 - t3 + if encode_errors < MAX_ERRORS and ids1 != ids2: i = find_first_mismatch(ids1, ids2) ids1 = list(ids1)[max(0, i - 2) : i + 5 + 1] ids2 = list(ids2)[max(0, i - 2) : i + 5 + 1] - logger.error(" TokenIDs: " + str(ids1)) - logger.error(" Expected: " + str(ids2)) + logger.error(" Expected: " + str(ids1)) + logger.error(" Result: " + str(ids2)) + encode_errors += 1 + logger.error(f" {encode_errors=}") + if decode_errors < MAX_ERRORS and not check_detokenizer(text, text1, text2): + i = find_first_mismatch(text1, text2) + text1 = list(text1[max(0, i - 2) : i + 5 + 1]) + text2 = list(text2[max(0, i - 2) : i + 5 + 1]) + logger.error(" Expected: " + " ".join(hex(ord(x)) for x in text1)) + logger.error(" Result: " + " ".join(hex(ord(x)) for x in text2)) + decode_errors += 1 + logger.error(f" {decode_errors=}") + if encode_errors >= MAX_ERRORS and decode_errors >= MAX_ERRORS: + logger.error(f" EXIT: {encode_errors=} {decode_errors=}") # raise Exception() - num_errors += 1 - if num_errors > 10: - break + break t_total = time.perf_counter() - t_start - logger.info("%s: end, tok1: %.3f tok2: %.3f total: %.3f" % (generator.__name__, t_tokenizer1, t_tokenizer2, t_total)) + logger.info(f"{generator.__name__}: end, {t_encode1=:.3f} {t_encode2=:.3f} {t_decode1=:.3f} {t_decode2=:.3f} {t_total=:.3f}") def main(argv: list[str] = None): @@ -357,74 +485,76 @@ def main(argv: list[str] = None): logging.basicConfig(level = logging.DEBUG if args.verbose else logging.INFO) logger.info(f"VOCABFILE: '{args.vocab_file}'") - model = LibLlamaModel(LibLlama(), args.vocab_file, mparams=dict(vocab_only=True), cparams=dict(n_ctx=4096)) - tokenizer = AutoTokenizer.from_pretrained(args.dir_tokenizer) + tokenizer1 = TokenizerGroundtruth(args.dir_tokenizer) + tokenizer2 = TokenizerLlamaCpp(args.vocab_file) - def func_tokenize1(text: str): - return model.tokenize(text, add_special=True, parse_special=True) + # compare_tokenizers(tokenizer1, tokenizer2, generator_custom_text()) + # compare_tokenizers(tokenizer1, tokenizer2, generator_custom_text_edge_cases()) + compare_tokenizers(tokenizer1, tokenizer2, generator_ascii_lr_strip()) + compare_tokenizers(tokenizer1, tokenizer2, generator_apostrophe()) + compare_tokenizers(tokenizer1, tokenizer2, generator_unicodes()) + compare_tokenizers(tokenizer1, tokenizer2, generator_vocab_words(tokenizer1)) + compare_tokenizers(tokenizer1, tokenizer2, generator_added_lr_strip(tokenizer1)) + # compare_tokenizers(tokenizer1, tokenizer2, generator_random_added_tokens(tokenizer1, 10_000)) + # compare_tokenizers(tokenizer1, tokenizer2, generator_random_chars(10_000)) + # compare_tokenizers(tokenizer1, tokenizer2, generator_random_unicodes(10_000)) + # compare_tokenizers(tokenizer1, tokenizer2, generator_random_vocab_chars(tokenizer1, 10_000)) + # compare_tokenizers(tokenizer1, tokenizer2, generator_random_vocab_words(tokenizer1, 5_000)) - def func_tokenize2(text: str): - return tokenizer.encode(text, add_special_tokens=True) - - ids = func_tokenize2("a") - assert 1 <= len(ids) <= 3 - add_bos_token = len(ids) > 1 and tokenizer.bos_token_id == ids[0] - add_eos_token = len(ids) > 1 and tokenizer.eos_token_id == ids[-1] - tokenizer.add_bos_token = getattr(tokenizer, "add_bos_token", add_bos_token) - tokenizer.add_eos_token = getattr(tokenizer, "add_eos_token", add_eos_token) - - vocab = list(sorted(tokenizer.batch_decode(list(tokenizer.get_vocab().values()), skip_special_tokens=True))) - - compare_tokenizers(func_tokenize1, func_tokenize2, generator_custom_text()) - compare_tokenizers(func_tokenize1, func_tokenize2, generator_custom_text_edge_cases()) - compare_tokenizers(func_tokenize1, func_tokenize2, generator_unicodes()) - compare_tokenizers(func_tokenize1, func_tokenize2, generator_vocab_words(vocab)) - compare_tokenizers(func_tokenize1, func_tokenize2, generator_added_lr_strip(tokenizer)) - compare_tokenizers(func_tokenize1, func_tokenize2, generator_random_added_tokens(tokenizer, 10_000)) - compare_tokenizers(func_tokenize1, func_tokenize2, generator_random_chars(10_000)) - compare_tokenizers(func_tokenize1, func_tokenize2, generator_random_unicodes(10_000)) - compare_tokenizers(func_tokenize1, func_tokenize2, generator_random_vocab_chars(vocab, 10_000)) - compare_tokenizers(func_tokenize1, func_tokenize2, generator_random_vocab_words(vocab, 5_000)) - - model.free() + tokenizer2.model.free() if __name__ == "__main__": # main() + if True: + logging.basicConfig( + level = logging.DEBUG, + format = "%(asctime)s.%(msecs)03d %(name)s %(levelname)s %(message)s", + datefmt = "%Y-%m-%d %H:%M:%S", + filename = logger.name + ".log", + filemode = "a" + ) logging.basicConfig( level = logging.DEBUG, - format = "%(asctime)s.%(msecs)03d %(name)s %(levelname)s %(message)s", - datefmt = "%Y-%m-%d %H:%M:%S", - filename = logger.name + ".log", - filemode = "a" + format = "%(levelname)s %(message)s", ) path_tokenizers = "./models/tokenizers/" path_vocab_format = "./models/ggml-vocab-%s.gguf" - # import os - # tokenizers = os.listdir(path_tokenizers) tokenizers = [ - # "llama-spm", # SPM - # "phi-3", # SPM - # "bert-bge", # WPM - # "jina-v2-en", # WPM - "gpt-2", # BPE + "llama-spm", # SPM + "phi-3", # SPM + "gemma", # SPM + "gemma-2", # SPM + "baichuan", # SPM + "bert-bge", # WPM + "jina-v2-en", # WPM "llama-bpe", # BPE + "phi-2", # BPE + "deepseek-llm", # BPE + "deepseek-coder", # BPE "falcon", # BPE + "mpt", # BPE "starcoder", # BPE + "gpt-2", # BPE + "stablelm2", # BPE + "refact", # BPE + "qwen2", # BPE + "olmo", # BPE "jina-v2-es", # BPE "jina-v2-de", # BPE - "jina-v2-code", # BPE "smaug-bpe", # BPE - "phi-2", # BPE - "deepseek-coder", # BPE - "deepseek-llm", # BPE + "poro-chat", # BPE + "jina-v2-code", # BPE + "viking", # BPE + "jais", # BPE ] + logger.info("=" * 50) for tokenizer in tokenizers: - logger.info("=" * 50) + logger.info("-" * 50) logger.info(f"TOKENIZER: '{tokenizer}'") vocab_file = path_vocab_format % tokenizer dir_tokenizer = path_tokenizers + "/" + tokenizer From 87e25a1d1bd26eb06d1cab9e2ee4e14a7a0be33c Mon Sep 17 00:00:00 2001 From: Daniel Bevenius Date: Sat, 6 Jul 2024 09:22:16 +0200 Subject: [PATCH 3/3] llama : add early return for empty range (#8327) MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit * llama : add early return for empty range This commit adds an early return to the llama_kv_cache_seq_add and llama_kv_cache_seq_div functions. The motivation for adding this is to avoid looping over the cache when the range is empty. I ran into this when using the self-extend feature in main.cpp. Signed-off-by: Daniel Bevenius * llama : add static_cast to fix CI warning/error This commit attempts to fix the following warning/error: ```console src/llama.cpp:7271:31: error: comparison of integer expressions of different signedness: ‘int’ and ‘uint32_t’ {aka ‘unsigned int’} [-Werror=sign-compare] 7271 | if (i < hparams.n_layer_dense_lead) { | ~~^~~~~~~~~~~~~~~~~~~~~~~~~~~~ ``` This can be reproduced locally by setting -Wsign-compare in the Makefile. Signed-off-by: Daniel Bevenius * squash! llama : add early return for empty range Remove the setting of cache.head to 0 when the range is empty. Signed-off-by: Daniel Bevenius * Update src/llama.cpp --------- Signed-off-by: Daniel Bevenius Co-authored-by: Georgi Gerganov --- src/llama.cpp | 4 ++++ 1 file changed, 4 insertions(+) diff --git a/src/llama.cpp b/src/llama.cpp index b770ca5bc..b39906fd5 100644 --- a/src/llama.cpp +++ b/src/llama.cpp @@ -3260,6 +3260,8 @@ static void llama_kv_cache_seq_add( if (p0 < 0) p0 = 0; if (p1 < 0) p1 = std::numeric_limits::max(); + // If there is no range then return early to avoid looping over the cache. + if (p0 == p1) return; if (cache.recurrent) { // for Mamba-like models, only the pos needs to be shifted @@ -3304,6 +3306,8 @@ static void llama_kv_cache_seq_div( int d) { if (p0 < 0) p0 = 0; if (p1 < 0) p1 = std::numeric_limits::max(); + // If there is no range then return early to avoid looping over the cache. + if (p0 == p1) return; if (cache.recurrent) { // for Mamba-like models, only the pos needs to be changed