From dd9e2fc98892e90d8ef97ec5e1f502f885c560ec Mon Sep 17 00:00:00 2001 From: Georgi Gerganov Date: Thu, 17 Aug 2023 19:32:14 +0300 Subject: [PATCH] ci : update ".bin" to ".gguf" extension ggml-ci --- README.md | 22 ++++++------ ci/run.sh | 44 +++++++++++------------ docs/token_generation_performance_tips.md | 6 ++-- examples/common.h | 2 +- examples/metal/metal.cpp | 2 +- examples/server/README.md | 6 ++-- llama.cpp | 2 +- 7 files changed, 42 insertions(+), 42 deletions(-) diff --git a/README.md b/README.md index 6900b1152..a24e91653 100644 --- a/README.md +++ b/README.md @@ -284,7 +284,7 @@ When built with Metal support, you can enable GPU inference with the `--gpu-laye Any value larger than 0 will offload the computation to the GPU. For example: ```bash -./main -m ./models/7B/ggml-model-q4_0.bin -n 128 -ngl 1 +./main -m ./models/7B/ggml-model-q4_0.gguf -n 128 -ngl 1 ``` ### MPI Build @@ -323,7 +323,7 @@ The above will distribute the computation across 2 processes on the first host a Finally, you're ready to run a computation using `mpirun`: ```bash -mpirun -hostfile hostfile -n 3 ./main -m ./models/7B/ggml-model-q4_0.bin -n 128 +mpirun -hostfile hostfile -n 3 ./main -m ./models/7B/ggml-model-q4_0.gguf -n 128 ``` ### BLAS Build @@ -506,10 +506,10 @@ python3 convert.py models/7B/ python convert.py models/7B/ --vocabtype bpe # quantize the model to 4-bits (using q4_0 method) -./quantize ./models/7B/ggml-model-f16.bin ./models/7B/ggml-model-q4_0.bin q4_0 +./quantize ./models/7B/ggml-model-f16.gguf ./models/7B/ggml-model-q4_0.gguf q4_0 # run the inference -./main -m ./models/7B/ggml-model-q4_0.bin -n 128 +./main -m ./models/7B/ggml-model-q4_0.gguf -n 128 ``` When running the larger models, make sure you have enough disk space to store all the intermediate files. @@ -565,7 +565,7 @@ Here is an example of a few-shot interaction, invoked with the command ./examples/chat-13B.sh # custom arguments using a 13B model -./main -m ./models/13B/ggml-model-q4_0.bin -n 256 --repeat_penalty 1.0 --color -i -r "User:" -f prompts/chat-with-bob.txt +./main -m ./models/13B/ggml-model-q4_0.gguf -n 256 --repeat_penalty 1.0 --color -i -r "User:" -f prompts/chat-with-bob.txt ``` Note the use of `--color` to distinguish between user input and generated text. Other parameters are explained in more detail in the [README](examples/main/README.md) for the `main` example program. @@ -628,6 +628,8 @@ OpenLLaMA is an openly licensed reproduction of Meta's original LLaMA model. It ### Using [GPT4All](https://github.com/nomic-ai/gpt4all) +*Note: these instructions are likely obsoleted by the GGUF update* + - Obtain the `tokenizer.model` file from LLaMA model and put it to `models` - Obtain the `added_tokens.json` file from Alpaca model and put it to `models` - Obtain the `gpt4all-lora-quantized.bin` file from GPT4All model and put it to `models/gpt4all-7B` @@ -703,7 +705,7 @@ If your issue is with model generation quality, then please at least scan the fo #### How to run 1. Download/extract: https://s3.amazonaws.com/research.metamind.io/wikitext/wikitext-2-raw-v1.zip?ref=salesforce-research -2. Run `./perplexity -m models/7B/ggml-model-q4_0.bin -f wiki.test.raw` +2. Run `./perplexity -m models/7B/ggml-model-q4_0.gguf -f wiki.test.raw` 3. Output: ``` perplexity : calculating perplexity over 655 chunks @@ -802,13 +804,13 @@ docker run -v /path/to/models:/models ghcr.io/ggerganov/llama.cpp:full --all-in- 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.bin -p "Building a website can be done in 10 simple steps:" -n 512 +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.bin -p "Building a website can be done in 10 simple steps:" -n 512 +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 ``` ### Docker With CUDA @@ -839,8 +841,8 @@ The resulting images, are essentially the same as the non-CUDA images: 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.bin -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.bin -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: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 ``` ### Contributing diff --git a/ci/run.sh b/ci/run.sh index 8dc394964..54ba6d710 100644 --- a/ci/run.sh +++ b/ci/run.sh @@ -159,17 +159,17 @@ function gg_run_open_llama_3b_v2 { python3 ../convert.py ${path_models} - model_f16="${path_models}/ggml-model-f16.bin" - model_q8_0="${path_models}/ggml-model-q8_0.bin" - model_q4_0="${path_models}/ggml-model-q4_0.bin" - model_q4_1="${path_models}/ggml-model-q4_1.bin" - model_q5_0="${path_models}/ggml-model-q5_0.bin" - model_q5_1="${path_models}/ggml-model-q5_1.bin" - model_q2_k="${path_models}/ggml-model-q2_k.bin" - model_q3_k="${path_models}/ggml-model-q3_k.bin" - model_q4_k="${path_models}/ggml-model-q4_k.bin" - model_q5_k="${path_models}/ggml-model-q5_k.bin" - model_q6_k="${path_models}/ggml-model-q6_k.bin" + model_f16="${path_models}/ggml-model-f16.gguf" + model_q8_0="${path_models}/ggml-model-q8_0.gguf" + model_q4_0="${path_models}/ggml-model-q4_0.gguf" + model_q4_1="${path_models}/ggml-model-q4_1.gguf" + model_q5_0="${path_models}/ggml-model-q5_0.gguf" + model_q5_1="${path_models}/ggml-model-q5_1.gguf" + model_q2_k="${path_models}/ggml-model-q2_k.gguf" + model_q3_k="${path_models}/ggml-model-q3_k.gguf" + model_q4_k="${path_models}/ggml-model-q4_k.gguf" + model_q5_k="${path_models}/ggml-model-q5_k.gguf" + model_q6_k="${path_models}/ggml-model-q6_k.gguf" wiki_test_60="${path_wiki}/wiki.test-60.raw" @@ -285,17 +285,17 @@ function gg_run_open_llama_7b_v2 { python3 ../convert.py ${path_models} - model_f16="${path_models}/ggml-model-f16.bin" - model_q8_0="${path_models}/ggml-model-q8_0.bin" - model_q4_0="${path_models}/ggml-model-q4_0.bin" - model_q4_1="${path_models}/ggml-model-q4_1.bin" - model_q5_0="${path_models}/ggml-model-q5_0.bin" - model_q5_1="${path_models}/ggml-model-q5_1.bin" - model_q2_k="${path_models}/ggml-model-q2_k.bin" - model_q3_k="${path_models}/ggml-model-q3_k.bin" - model_q4_k="${path_models}/ggml-model-q4_k.bin" - model_q5_k="${path_models}/ggml-model-q5_k.bin" - model_q6_k="${path_models}/ggml-model-q6_k.bin" + model_f16="${path_models}/ggml-model-f16.gguf" + model_q8_0="${path_models}/ggml-model-q8_0.gguf" + model_q4_0="${path_models}/ggml-model-q4_0.gguf" + model_q4_1="${path_models}/ggml-model-q4_1.gguf" + model_q5_0="${path_models}/ggml-model-q5_0.gguf" + model_q5_1="${path_models}/ggml-model-q5_1.gguf" + model_q2_k="${path_models}/ggml-model-q2_k.gguf" + model_q3_k="${path_models}/ggml-model-q3_k.gguf" + model_q4_k="${path_models}/ggml-model-q4_k.gguf" + model_q5_k="${path_models}/ggml-model-q5_k.gguf" + model_q6_k="${path_models}/ggml-model-q6_k.gguf" wiki_test="${path_wiki}/wiki.test.raw" diff --git a/docs/token_generation_performance_tips.md b/docs/token_generation_performance_tips.md index 69ba6173c..c9acff7d4 100644 --- a/docs/token_generation_performance_tips.md +++ b/docs/token_generation_performance_tips.md @@ -3,7 +3,7 @@ ## Verifying that the model is running on the GPU with cuBLAS Make sure you compiled llama with the correct env variables according to [this guide](../README.md#cublas), so that llama accepts the `-ngl N` (or `--n-gpu-layers N`) flag. When running llama, you may configure `N` to be very large, and llama will offload the maximum possible number of layers to the GPU, even if it's less than the number you configured. For example: ```shell -./main -m "path/to/model.bin" -ngl 200000 -p "Please sir, may I have some " +./main -m "path/to/model.gguf" -ngl 200000 -p "Please sir, may I have some " ``` When running llama, before it starts the inference work, it will output diagnostic information that shows whether cuBLAS is offloading work to the GPU. Look for these lines: @@ -25,9 +25,9 @@ GPU: A6000 (48GB VRAM) CPU: 7 physical cores RAM: 32GB -Model: `TheBloke_Wizard-Vicuna-30B-Uncensored-GGML/Wizard-Vicuna-30B-Uncensored.ggmlv3.q4_0.bin` (30B parameters, 4bit quantization, GGML) +Model: `TheBloke_Wizard-Vicuna-30B-Uncensored-GGML/Wizard-Vicuna-30B-Uncensored.q4_0.gguf` (30B parameters, 4bit quantization, GGML) -Run command: `./main -m "path/to/model.bin" -p "-p "An extremely detailed description of the 10 best ethnic dishes will follow, with recipes: " -n 1000 [additional benchmark flags]` +Run command: `./main -m "path/to/model.gguf" -p "An extremely detailed description of the 10 best ethnic dishes will follow, with recipes: " -n 1000 [additional benchmark flags]` Result: diff --git a/examples/common.h b/examples/common.h index b8510dfe4..62772eff7 100644 --- a/examples/common.h +++ b/examples/common.h @@ -52,7 +52,7 @@ struct gpt_params { std::string cfg_negative_prompt; // string to help guidance float cfg_scale = 1.f; // How strong is guidance - std::string model = "models/7B/ggml-model.bin"; // model path + std::string model = "models/7B/ggml-model-f16.bin"; // model path std::string model_alias = "unknown"; // model alias std::string prompt = ""; std::string path_prompt_cache = ""; // path to file for saving/loading prompt eval state diff --git a/examples/metal/metal.cpp b/examples/metal/metal.cpp index 7438defde..c05a4fa93 100644 --- a/examples/metal/metal.cpp +++ b/examples/metal/metal.cpp @@ -2,7 +2,7 @@ // // - First, export a LLaMA graph: // -// $ ./bin/main -m ../models/7B/ggml-model-q4_0.bin --export +// $ ./bin/main -m ../models/7B/ggml-model-q4_0.gguf --export // // - Run this tool to evaluate the exported graph: // diff --git a/examples/server/README.md b/examples/server/README.md index 1559dd3f2..ce2d2c9c6 100644 --- a/examples/server/README.md +++ b/examples/server/README.md @@ -5,7 +5,7 @@ This example demonstrates a simple HTTP API server and a simple web front end to Command line options: - `--threads N`, `-t N`: Set the number of threads to use during computation. -- `-m FNAME`, `--model FNAME`: Specify the path to the LLaMA model file (e.g., `models/7B/ggml-model.bin`). +- `-m FNAME`, `--model FNAME`: Specify the path to the LLaMA model file (e.g., `models/7B/ggml-model.gguf`). - `-m ALIAS`, `--alias ALIAS`: Set an alias for the model. The alias will be returned in API responses. - `-c N`, `--ctx-size N`: Set the size of the prompt context. The default is 512, but LLaMA models were built with a context of 2048, which will provide better results for longer input/inference. The size may differ in other models, for example, baichuan models were build with a context of 4096. - `-ngl N`, `--n-gpu-layers N`: When compiled with appropriate support (currently CLBlast or cuBLAS), this option allows offloading some layers to the GPU for computation. Generally results in increased performance. @@ -48,14 +48,12 @@ To get started right away, run the following command, making sure to use the cor ### Unix-based systems (Linux, macOS, etc.): ```bash -./server -m models/7B/ggml-model.bin -c 2048 +./server -m models/7B/ggml-model.gguf -c 2048 ``` ### Windows: ```powershell -server.exe -m models\7B\ggml-model.bin -c 2048 -``` The above command will start a server that by default listens on `127.0.0.1:8080`. You can consume the endpoints with Postman or NodeJS with axios library. You can visit the web front end at the same url. diff --git a/llama.cpp b/llama.cpp index 1d1642bd8..018f52e71 100644 --- a/llama.cpp +++ b/llama.cpp @@ -3575,7 +3575,7 @@ static void llama_model_quantize_internal(const std::string & fname_inp, const s } else { size_t counter = 0; new_size = 0; - auto compute = [&mutex, &counter, &hist_cur, &new_size, new_type, f32_data, new_data, nelements] () { + auto compute = [&mutex, &counter, &hist_cur, &new_size, new_type, f32_data, new_data, nelements, chunk_size]() { // NOLINT std::vector local_hist; size_t local_size = 0; while (true) {