feat: add IBM Granite code models to the catalog

Fixes containers#1157

Signed-off-by: Jeff MAURY <[email protected]>

packages/backend/src/assets/ai.json

@@ -49,6 +49,8 @@
       "basedir": "recipes/natural_language_processing/codegen",
       "readme": "# Code Generation Application\n\n  This recipe helps developers start building their own custom LLM enabled code generation applications. It consists of two main components: the Model Service and the AI Application.\n\n  There are a few options today for local Model Serving, but this recipe will use [`llama-cpp-python`](https://github.com/abetlen/llama-cpp-python) and their OpenAI compatible Model Service. There is a Containerfile provided that can be used to build this Model Service within the repo, [`model_servers/llamacpp_python/base/Containerfile`](/model_servers/llamacpp_python/base/Containerfile).\n\n  The AI Application will connect to the Model Service via its OpenAI compatible API. The recipe relies on [Langchain's](https://python.langchain.com/docs/get_started/introduction) python package to simplify communication with the Model Service and uses [Streamlit](https://streamlit.io/) for the UI layer. You can find an example of the code generation application below.\n\n![](/assets/codegen_ui.png) \n\n\n## Try the Code Generation Application\n\nThe [Podman Desktop](https://podman-desktop.io) [AI Lab Extension](https://github.com/containers/podman-desktop-extension-ai-lab) includes this recipe among others. To try it out, open `Recipes Catalog` -> `Code Generation` and follow the instructions to start the application.\n\n# Build the Application\n\nThe rest of this document will explain how to build and run the application from the terminal, and will\ngo into greater detail on how each container in the Pod above is built, run, and \nwhat purpose it serves in the overall application. All the recipes use a central [Makefile](../../common/Makefile.common) that includes variables populated with default values to simplify getting started. Please review the [Makefile docs](../../common/README.md), to learn about further customizing your application.\n\n\nThis application requires a model, a model service and an AI inferencing application.\n\n* [Quickstart](#quickstart)\n* [Download a model](#download-a-model)\n* [Build the Model Service](#build-the-model-service)\n* [Deploy the Model Service](#deploy-the-model-service)\n* [Build the AI Application](#build-the-ai-application)\n* [Deploy the AI Application](#deploy-the-ai-application)\n* [Interact with the AI Application](#interact-with-the-ai-application)\n* [Embed the AI Application in a Bootable Container Image](#embed-the-ai-application-in-a-bootable-container-image)\n\n\n## Quickstart\nTo run the application with pre-built images from `quay.io/ai-lab`, use `make quadlet`. This command\nbuilds the application's metadata and generates Kubernetes YAML at `./build/codegen.yaml` to spin up a Pod that can then be launched locally.\nTry it with:\n\n```\nmake quadlet\npodman kube play build/codegen.yaml\n```\n\nThis will take a few minutes if the model and model-server container images need to be downloaded. \nThe Pod is named `codegen`, so you may use [Podman](https://podman.io) to manage the Pod and its containers:\n\n```\npodman pod list\npodman ps\n```\n\nOnce the Pod and its containers are running, the application can be accessed at `http://localhost:8501`. \nPlease refer to the section below for more details about [interacting with the codegen application](#interact-with-the-ai-application).\n\nTo stop and remove the Pod, run:\n\n```\npodman pod stop codegen\npodman pod rm codgen\n```\n\n## Download a model\n\nIf you are just getting started, we recommend using [Mistral-7B-code-16k-qlora](https://huggingface.co/Nondzu/Mistral-7B-code-16k-qlora). This is a well\nperformant mid-sized model with an apache-2.0 license fine tuned for code generation. In order to use it with our Model Service we need it converted\nand quantized into the [GGUF format](https://github.com/ggerganov/ggml/blob/master/docs/gguf.md). There are a number of\nways to get a GGUF version of Mistral-7B-code-16k-qlora, but the simplest is to download a pre-converted one from\n[huggingface.co](https://huggingface.co) here: https://huggingface.co/TheBloke/Mistral-7B-Code-16K-qlora-GGUF.\n\nThere are a number of options for quantization level, but we recommend `Q4_K_M`. \n\nThe recommended model can be downloaded using the code snippet below:\n\n```bash\ncd ../../../models\ncurl -sLO https://huggingface.co/TheBloke/Mistral-7B-Code-16K-qlora-GGUF/resolve/main/mistral-7b-code-16k-qlora.Q4_K_M.gguf\ncd ../recipes/natural_language_processing/codgen\n```\n\n_A full list of supported open models is forthcoming._  \n\n\n## Build the Model Service\n\nThe complete instructions for building and deploying the Model Service can be found in the\n[llamacpp_python model-service document](../../../model_servers/llamacpp_python/README.md).\n\nThe Model Service can be built from make commands from the [llamacpp_python directory](../../../model_servers/llamacpp_python/).\n\n```bash\n# from path model_servers/llamacpp_python from repo containers/ai-lab-recipes\nmake build\n```\nCheckout the [Makefile](../../../model_servers/llamacpp_python/Makefile) to get more details on different options for how to build.\n\n## Deploy the Model Service\n\nThe local Model Service relies on a volume mount to the localhost to access the model files. It also employs environment variables to dictate the model used and where its served. You can start your local Model Service using the following `make` command from `model_servers/llamacpp_python` set with reasonable defaults:\n\n```bash\n# from path model_servers/llamacpp_python from repo containers/ai-lab-recipes\nmake run\n```\n\n## Build the AI Application\n\nThe AI Application can be built from the make command:\n\n```bash\n# Run this from the current directory (path recipes/natural_language_processing/codegen from repo containers/ai-lab-recipes)\nmake build\n```\n\n## Deploy the AI Application\n\nMake sure the Model Service is up and running before starting this container image. When starting the AI Application container image we need to direct it to the correct `MODEL_ENDPOINT`. This could be any appropriately hosted Model Service (running locally or in the cloud) using an OpenAI compatible API. In our case the Model Service is running inside the Podman machine so we need to provide it with the appropriate address `10.88.0.1`. To deploy the AI application use the following:\n\n```bash\n# Run this from the current directory (path recipes/natural_language_processing/codegen from repo containers/ai-lab-recipes)\nmake run \n```\n\n## Interact with the AI Application\n\nEverything should now be up an running with the code generation application available at [`http://localhost:8501`](http://localhost:8501). By using this recipe and getting this starting point established, users should now have an easier time customizing and building their own LLM enabled code generation applications.   \n\n## Embed the AI Application in a Bootable Container Image\n\nTo build a bootable container image that includes this sample code generation workload as a service that starts when a system is booted, run: `make -f Makefile bootc`. You can optionally override the default image / tag you want to give the make command by specifying it as follows: `make -f Makefile BOOTC_IMAGE=<your_bootc_image> bootc`.\n\nSubstituting the bootc/Containerfile FROM command is simple using the Makefile FROM option.\n\n```bash\nmake FROM=registry.redhat.io/rhel9/rhel-bootc:9.4 bootc\n```\n\nSelecting the ARCH for the bootc/Containerfile is simple using the Makefile ARCH= variable.\n\n```\nmake ARCH=x86_64 bootc\n```\n\nThe magic happens when you have a bootc enabled system running. If you do, and you'd like to update the operating system to the OS you just built\nwith the code generation application, it's as simple as ssh-ing into the bootc system and running:\n\n```bash\nbootc switch quay.io/ai-lab/codegen-bootc:latest\n```\n\nUpon a reboot, you'll see that the codegen service is running on the system. Check on the service with:\n\n```bash\nssh user@bootc-system-ip\nsudo systemctl status codegen\n```\n\n### What are bootable containers?\n\nWhat's a [bootable OCI container](https://containers.github.io/bootc/) and what's it got to do with AI?\n\nThat's a good question! We think it's a good idea to embed AI workloads (or any workload!) into bootable images at _build time_ rather than\nat _runtime_. This extends the benefits, such as portability and predictability, that containerizing applications provides to the operating system.\nBootable OCI images bake exactly what you need to run your workloads into the operating system at build time by using your favorite containerization\ntools. Might I suggest [podman](https://podman.io/)?\n\nOnce installed, a bootc enabled system can be updated by providing an updated bootable OCI image from any OCI\nimage registry with a single `bootc` command. This works especially well for fleets of devices that have fixed workloads - think\nfactories or appliances. Who doesn't want to add a little AI to their appliance, am I right?\n\nBootable images lend toward immutable operating systems, and the more immutable an operating system is, the less that can go wrong at runtime!\n\n#### Creating bootable disk images\n\nYou can convert a bootc image to a bootable disk image using the\n[quay.io/centos-bootc/bootc-image-builder](https://github.com/osbuild/bootc-image-builder) container image.\n\nThis container image allows you to build and deploy [multiple disk image types](../../common/README_bootc_image_builder.md) from bootc container images.\n\nDefault image types can be set via the DISK_TYPE Makefile variable.\n\n`make bootc-image-builder DISK_TYPE=ami`\n",
       "recommended": [
+        "hf.ibm-granite.granite-3b-code-instruct",
+        "hf.ibm-granite.granite-8b-code-instruct",
         "hf.TheBloke.mistral-7b-code-16k-qlora.Q4_K_M",
         "hf.TheBloke.mistral-7b-codealpaca-lora.Q4_K_M"
       ],
@@ -177,6 +179,30 @@
       "sha256": "94f3a16321c9604ca22e970f3b89931ae5b4bbfd4c5d996e2bb606c506590666",
       "backend": "llama-cpp"
     },
+    {
+      "id": "hf.ibm-granite.granite-3b-code-instruct",
+      "name": "ibm-granite/granite-3b-code-instruct-GGUF",
+      "description": "![image/png](https://cdn-uploads.huggingface.co/production/uploads/62cd5057674cdb524450093d/1hzxoPwqkBJXshKVVe6_9.png)\n\n# ibm-granite/granite-3b-code-instruct-GGUF\nThis is the Q4_K_M converted version of the original [`ibm-granite/granite-3b-code-instruct`](https://huggingface.co/ibm-granite/granite-3b-code-instruct).\nRefer to the [original model card](https://huggingface.co/ibm-granite/granite-3b-code-instruct) for more details.\n\n## Use with llama.cpp\n```shell\ngit clone https://github.com/ggerganov/llama.cpp\ncd llama.cpp\n\n# install\nmake\n\n# run generation\n./main -m granite-3b-code-instruct-GGUF/granite-3b-code-instruct.Q4_K_M.gguf -n 128 -p \"def generate_random(x: int):\" --color\n```",
+      "hw": "CPU",
+      "registry": "Hugging Face",
+      "license": "Apache-2.0",
+      "url": "https://huggingface.co/ibm-granite/granite-3b-code-instruct-GGUF/resolve/main/granite-3b-code-instruct.Q4_K_M.gguf",
+      "memory": 2126008812,
+      "sha256": "5bd783ab3925f425f17764fd34c1f7119fb64a023ccf9dd48654c3c3f252a8ff",
+      "backend": "llama-cpp"
+    },
+    {
+      "id": "hf.ibm-granite.granite-8b-code-instruct",
+      "name": "ibm-granite/granite-8b-code-instruct-GGUF",
+      "description": "![image/png](https://cdn-uploads.huggingface.co/production/uploads/62cd5057674cdb524450093d/1hzxoPwqkBJXshKVVe6_9.png)\n\n# ibm-granite/granite-8b-code-instruct-GGUF\nThis is the Q4_K_M converted version of the original [`ibm-granite/granite-8b-code-instruct`](https://huggingface.co/ibm-granite/granite-8b-code-instruct).\nRefer to the [original model card](https://huggingface.co/ibm-granite/granite-8b-code-instruct) for more details.\n\n## Use with llama.cpp\n```shell\ngit clone https://github.com/ggerganov/llama.cpp\ncd llama.cpp\n\n# install\nmake\n\n# run generation\n./main -m granite-8b-code-instruct-GGUF/granite-8b-code-instruct.Q4_K_M.gguf -n 128 -p \"def generate_random(x: int):\" --color\n```",
+      "hw": "CPU",
+      "registry": "Hugging Face",
+      "license": "Apache-2.0",
+      "url": "https://huggingface.co/ibm-granite/granite-8b-code-instruct-GGUF/resolve/main/granite-8b-code-instruct.Q4_K_M.gguf",
+      "memory": 5347234284,
+      "sha256": "bc8804cb43c4e1e82e2188658569b147587f83a89640600a64d5f7d7de2565b4",
+      "backend": "llama-cpp"
+    },
     {
       "id": "hf.TheBloke.mistral-7b-codealpaca-lora.Q4_K_M",
       "name": "TheBloke/Mistral-7B-codealpaca-lora-GGUF",

packages/backend/src/workers/provider/LlamaCppPython.ts

@@ -25,7 +25,7 @@ import type { TaskRegistry } from '../../registries/TaskRegistry';
 import { InferenceType } from '@shared/src/models/IInference';
 
 export const LLAMA_CPP_INFERENCE_IMAGE =
-  'ghcr.io/containers/podman-desktop-extension-ai-lab-playground-images/ai-lab-playground-chat:0.3.2';
+  'ghcr.io/containers/podman-desktop-extension-ai-lab-playground-images/ai-lab-playground-chat:0.4';
 
 export const SECOND: number = 1_000_000_000;