llama_cpp

llama_cpp

CONTAINERS IMAGES RUN BUILD

• llama.cpp from https://github.com/ggerganov/llama.cpp with CUDA enabled (found under /opt/llama.cpp)
• Python bindings from https://github.com/abetlen/llama-cpp-python (found under /opt/llama-cpp-python)

Warning

Starting with version 0.1.79, the model format has changed from GGML to GGUF. Existing GGML models can be converted using the convert-llama-ggmlv3-to-gguf.py script in llama.cpp (or you can often find the GGUF conversions on HuggingFace Hub)

There are two branches of this container for backwards compatability:

• llama_cpp:gguf (the default, which tracks upstream master)
• llama_cpp:ggml (which still supports GGML model format)

There are a couple patches applied to the legacy GGML fork:

• fixed __fp16 typedef in llama.h on ARM64 (use half with NVCC)
• parsing of BOS/EOS tokens (see ggerganov/llama.cpp#1931)

Inference Benchmark

You can use llama.cpp's built-in main tool to run GGUF models (from HuggingFace Hub or elsewhere)

./run.sh --workdir=/opt/llama.cpp/bin $(./autotag llama_cpp) /bin/bash -c \
 './main --model $(huggingface-downloader TheBloke/Llama-2-7B-GGUF/llama-2-7b.Q4_K_S.gguf) \
         --prompt "Once upon a time," \
         --n-predict 128 --ctx-size 192 --batch-size 192 \
         --n-gpu-layers 999 --threads $(nproc)'

> the --model argument expects a .gguf filename (typically the Q4_K_S quantization is used)
> if you're trying to load Llama-2-70B, add the --gqa 8 flag

To use the Python API and benchmark.py instead:

./run.sh --workdir=/opt/llama.cpp/bin $(./autotag llama_cpp) /bin/bash -c \
 'python3 benchmark.py --model $(huggingface-downloader TheBloke/Llama-2-7B-GGUF/llama-2-7b.Q4_K_S.gguf) \
            --prompt "Once upon a time," \
            --n-predict 128 --ctx-size 192 --batch-size 192 \
            --n-gpu-layers 999 --threads $(nproc)'

Memory Usage

Model	Quantization	Memory (MB)
TheBloke/Llama-2-7B-GGUF	llama-2-7b.Q4_K_S.gguf	5,268
TheBloke/Llama-2-13B-GGUF	llama-2-13b.Q4_K_S.gguf	8,609
TheBloke/LLaMA-30b-GGUF	llama-30b.Q4_K_S.gguf	19,045
TheBloke/Llama-2-70B-GGUF	llama-2-70b.Q4_K_S.gguf	37,655

CONTAINERS

llama_cpp:0.2.57
Aliases	llama_cpp
Requires	L4T ['>=34.1.0']
Dependencies	build-essential cuda cudnn python cmake numpy huggingface_hub
Dependants	langchain langchain:samples text-generation-webui:1.7 text-generation-webui:6a7cd01 text-generation-webui:main
Dockerfile	Dockerfile

CONTAINER IMAGES

Repository/Tag	Date	Arch	Size
dustynv/llama_cpp:ggml-r35.2.1	2023-12-05	arm64	5.2GB
dustynv/llama_cpp:ggml-r35.3.1	2023-12-06	arm64	5.2GB
dustynv/llama_cpp:ggml-r35.4.1	2023-12-19	arm64	5.2GB
dustynv/llama_cpp:ggml-r36.2.0	2023-12-19	arm64	5.1GB
dustynv/llama_cpp:gguf-r35.2.1	2023-12-15	arm64	5.1GB
dustynv/llama_cpp:gguf-r35.3.1	2023-12-19	arm64	5.2GB
dustynv/llama_cpp:gguf-r35.4.1	2023-12-15	arm64	5.1GB
dustynv/llama_cpp:gguf-r36.2.0	2023-12-19	arm64	5.1GB
dustynv/llama_cpp:r35.2.1	2023-08-29	arm64	5.2GB
dustynv/llama_cpp:r35.3.1	2023-08-15	arm64	5.2GB
dustynv/llama_cpp:r35.4.1	2023-08-13	arm64	5.1GB
dustynv/llama_cpp:r36.2.0	2024-02-22	arm64	5.3GB

_{Container images are compatible with other minor versions of JetPack/L4T:
    • L4T R32.7 containers can run on other versions of L4T R32.7 (JetPack 4.6+)
    • L4T R35.x containers can run on other versions of L4T R35.x (JetPack 5.1+)}

RUN CONTAINER

To start the container, you can use jetson-containers run and autotag, or manually put together a docker run command:

# automatically pull or build a compatible container image
jetson-containers run $(autotag llama_cpp)

# or explicitly specify one of the container images above
jetson-containers run dustynv/llama_cpp:r36.2.0

# or if using 'docker run' (specify image and mounts/ect)
sudo docker run --runtime nvidia -it --rm --network=host dustynv/llama_cpp:r36.2.0

^{jetson-containers run forwards arguments to docker run with some defaults added (like --runtime nvidia, mounts a /data cache, and detects devices)
autotag finds a container image that's compatible with your version of JetPack/L4T - either locally, pulled from a registry, or by building it.}

To mount your own directories into the container, use the -v or --volume flags:

jetson-containers run -v /path/on/host:/path/in/container $(autotag llama_cpp)

To launch the container running a command, as opposed to an interactive shell:

jetson-containers run $(autotag llama_cpp) my_app --abc xyz

You can pass any options to it that you would to docker run, and it'll print out the full command that it constructs before executing it.

BUILD CONTAINER

If you use autotag as shown above, it'll ask to build the container for you if needed. To manually build it, first do the system setup, then run:

jetson-containers build llama_cpp

The dependencies from above will be built into the container, and it'll be tested during. Run it with --help for build options.