base on The official PyTorch implementation of Google's Gemma models # Gemma in PyTorch **Gemma** is a family of lightweight, state-of-the art open models built from research and technology used to create Google Gemini models. They are text-to-text, decoder-only large language models, available in English, with open weights, pre-trained variants, and instruction-tuned variants. For more details, please check out the following links: * [Gemma on Google AI](https://ai.google.dev/gemma) * [Gemma on Kaggle](https://www.kaggle.com/models/google/gemma) * [Gemma on Vertex AI Model Garden](https://console.cloud.google.com/vertex-ai/publishers/google/model-garden/335) This is the official PyTorch implementation of Gemma models. We provide model and inference implementations using both PyTorch and PyTorch/XLA, and support running inference on CPU, GPU and TPU. ## Updates * [June 26th 🔥] Support Gemma v2. You can find the checkpoints [on Kaggle](https://www.kaggle.com/models/google/gemma-2/pytorch) and Hugging Face * [April 9th] Support CodeGemma. You can find the checkpoints [on Kaggle](https://www.kaggle.com/models/google/codegemma/pytorch) and [Hugging Face](https://huggingface.co/collections/google/codegemma-release-66152ac7b683e2667abdee11) * [April 5] Support Gemma v1.1. You can find the v1.1 checkpoints [on Kaggle](https://www.kaggle.com/models/google/gemma/frameworks/pyTorch) and [Hugging Face](https://huggingface.co/collections/google/gemma-release-65d5efbccdbb8c4202ec078b). ## Download Gemma model checkpoint You can find the model checkpoints on Kaggle [here](https://www.kaggle.com/models/google/gemma/frameworks/pyTorch). Alternatively, you can find the model checkpoints on the Hugging Face Hub [here](https://huggingface.co/models?other=gemma_torch). To download the models, go the the model repository of the model of interest and click the `Files and versions` tab, and download the model and tokenizer files. For programmatic downloading, if you have `huggingface_hub` installed, you can also run: ``` huggingface-cli download google/gemma-7b-it-pytorch ``` Note that you can choose between the 2B, 2B V2, 7B, 7B int8 quantized, 9B, and 27B variants. ``` VARIANT=<2b or 7b or 9b or 27b> CKPT_PATH=<Insert ckpt path here> ``` ## Try it free on Colab Follow the steps at [https://ai.google.dev/gemma/docs/pytorch_gemma](https://ai.google.dev/gemma/docs/pytorch_gemma). ## Try it out with PyTorch Prerequisite: make sure you have setup docker permission properly as a non-root user. ```bash sudo usermod -aG docker $USER newgrp docker ``` ### Build the docker image. ```bash DOCKER_URI=gemma:${USER} docker build -f docker/Dockerfile ./ -t ${DOCKER_URI} ``` ### Run Gemma inference on CPU. ```bash PROMPT="The meaning of life is" docker run -t --rm \ -v ${CKPT_PATH}:/tmp/ckpt \ ${DOCKER_URI} \ python scripts/run.py \ --ckpt=/tmp/ckpt \ --variant="${VARIANT}" \ --prompt="${PROMPT}" # add `--quant` for the int8 quantized model. ``` ### Run Gemma inference on GPU. ```bash PROMPT="The meaning of life is" docker run -t --rm \ --gpus all \ -v ${CKPT_PATH}:/tmp/ckpt \ ${DOCKER_URI} \ python scripts/run.py \ --device=cuda \ --ckpt=/tmp/ckpt \ --variant="${VARIANT}" \ --prompt="${PROMPT}" # add `--quant` for the int8 quantized model. ``` ## Try It out with PyTorch/XLA ### Build the docker image (CPU, TPU). ```bash DOCKER_URI=gemma_xla:${USER} docker build -f docker/xla.Dockerfile ./ -t ${DOCKER_URI} ``` ### Build the docker image (GPU). ```bash DOCKER_URI=gemma_xla_gpu:${USER} docker build -f docker/xla_gpu.Dockerfile ./ -t ${DOCKER_URI} ``` ### Run Gemma inference on CPU. ```bash docker run -t --rm \ --shm-size 4gb \ -e PJRT_DEVICE=CPU \ -v ${CKPT_PATH}:/tmp/ckpt \ ${DOCKER_URI} \ python scripts/run_xla.py \ --ckpt=/tmp/ckpt \ --variant="${VARIANT}" \ # add `--quant` for the int8 quantized model. ``` ### Run Gemma inference on TPU. Note: be sure to use the docker container built from `xla.Dockerfile`. ```bash docker run -t --rm \ --shm-size 4gb \ -e PJRT_DEVICE=TPU \ -v ${CKPT_PATH}:/tmp/ckpt \ ${DOCKER_URI} \ python scripts/run_xla.py \ --ckpt=/tmp/ckpt \ --variant="${VARIANT}" \ # add `--quant` for the int8 quantized model. ``` ### Run Gemma inference on GPU. Note: be sure to use the docker container built from `xla_gpu.Dockerfile`. ```bash docker run -t --rm --privileged \ --shm-size=16g --net=host --gpus all \ -e USE_CUDA=1 \ -e PJRT_DEVICE=CUDA \ -v ${CKPT_PATH}:/tmp/ckpt \ ${DOCKER_URI} \ python scripts/run_xla.py \ --ckpt=/tmp/ckpt \ --variant="${VARIANT}" \ # add `--quant` for the int8 quantized model. ``` ### Tokenizer Notes 99 unused tokens are reserved in the pretrained tokenizer model to assist with more efficient training/fine-tuning. Unused tokens are in the string format of `<unused[0-98]>` with token id range of `[7-105]`. ``` "<unused0>": 7, "<unused1>": 8, "<unused2>": 9, ... "<unused98>": 105, ``` ## Disclaimer This is not an officially supported Google product. ", Assign "at most 3 tags" to the expected json: {"id":"8031","tags":[]} "only from the tags list I provide: [{"id":77,"name":"3d"},{"id":89,"name":"agent"},{"id":17,"name":"ai"},{"id":54,"name":"algorithm"},{"id":24,"name":"api"},{"id":44,"name":"authentication"},{"id":3,"name":"aws"},{"id":27,"name":"backend"},{"id":60,"name":"benchmark"},{"id":72,"name":"best-practices"},{"id":39,"name":"bitcoin"},{"id":37,"name":"blockchain"},{"id":1,"name":"blog"},{"id":45,"name":"bundler"},{"id":58,"name":"cache"},{"id":21,"name":"chat"},{"id":49,"name":"cicd"},{"id":4,"name":"cli"},{"id":64,"name":"cloud-native"},{"id":48,"name":"cms"},{"id":61,"name":"compiler"},{"id":68,"name":"containerization"},{"id":92,"name":"crm"},{"id":34,"name":"data"},{"id":47,"name":"database"},{"id":8,"name":"declarative-gui "},{"id":9,"name":"deploy-tool"},{"id":53,"name":"desktop-app"},{"id":6,"name":"dev-exp-lib"},{"id":59,"name":"dev-tool"},{"id":13,"name":"ecommerce"},{"id":26,"name":"editor"},{"id":66,"name":"emulator"},{"id":62,"name":"filesystem"},{"id":80,"name":"finance"},{"id":15,"name":"firmware"},{"id":73,"name":"for-fun"},{"id":2,"name":"framework"},{"id":11,"name":"frontend"},{"id":22,"name":"game"},{"id":81,"name":"game-engine "},{"id":23,"name":"graphql"},{"id":84,"name":"gui"},{"id":91,"name":"http"},{"id":5,"name":"http-client"},{"id":51,"name":"iac"},{"id":30,"name":"ide"},{"id":78,"name":"iot"},{"id":40,"name":"json"},{"id":83,"name":"julian"},{"id":38,"name":"k8s"},{"id":31,"name":"language"},{"id":10,"name":"learning-resource"},{"id":33,"name":"lib"},{"id":41,"name":"linter"},{"id":28,"name":"lms"},{"id":16,"name":"logging"},{"id":76,"name":"low-code"},{"id":90,"name":"message-queue"},{"id":42,"name":"mobile-app"},{"id":18,"name":"monitoring"},{"id":36,"name":"networking"},{"id":7,"name":"node-version"},{"id":55,"name":"nosql"},{"id":57,"name":"observability"},{"id":46,"name":"orm"},{"id":52,"name":"os"},{"id":14,"name":"parser"},{"id":74,"name":"react"},{"id":82,"name":"real-time"},{"id":56,"name":"robot"},{"id":65,"name":"runtime"},{"id":32,"name":"sdk"},{"id":71,"name":"search"},{"id":63,"name":"secrets"},{"id":25,"name":"security"},{"id":85,"name":"server"},{"id":86,"name":"serverless"},{"id":70,"name":"storage"},{"id":75,"name":"system-design"},{"id":79,"name":"terminal"},{"id":29,"name":"testing"},{"id":12,"name":"ui"},{"id":50,"name":"ux"},{"id":88,"name":"video"},{"id":20,"name":"web-app"},{"id":35,"name":"web-server"},{"id":43,"name":"webassembly"},{"id":69,"name":"workflow"},{"id":87,"name":"yaml"}]" returns me the "expected json"