local-llm

Run LLMs locally on Cloud Workstations. Uses:

• Quantized models from 🤗
• llama-cpp-python's webserver

In this guide:

• Running as a Cloud Workstation
• local-llm commands
• Running locally
• Debugging
• LLM disclaimer

Running as a Cloud Workstation

This repository includes a Dockerfile that can be used to create a custom base image for a Cloud Workstation environment that includes the llm tool.

To get started, you'll need to have a GCP Project and have the gcloud CLI installed.

1. Set environment variables

   1. Set the PROJECT_ID and PROJECT_NUM environment variables from your GCP project. You must modify the values.

      export PROJECT_ID=<project-id>
      export PROJECT_NUM=<project-num>

   2. Set other needed environment variables. You can modify the values.

      export REGION=us-central1
      export LOCALLLM_REGISTRY=localllm-registry
      export LOCALLLM_IMAGE_NAME=localllm
      export LOCALLLM_CLUSTER=localllm-cluster
      export LOCALLLM_WORKSTATION=localllm-workstation
      export LOCALLLM_PORT=8000

2. Set the default project.

   gcloud config set project $PROJECT_ID

3. Enable needed services.

   gcloud services enable \
     cloudbuild.googleapis.com \
     workstations.googleapis.com \
     container.googleapis.com \
     containeranalysis.googleapis.com \
     containerscanning.googleapis.com \
     artifactregistry.googleapis.com

4. Create an Artifact Registry repository for docker images.

   gcloud artifacts repositories create $LOCALLLM_REGISTRY \
     --location=$REGION \
     --repository-format=docker

5. Build and push the image to Artifact Registry using Cloud Build. Details are in cloudbuild.yaml.

   gcloud builds submit . \
       --substitutions=_IMAGE_REGISTRY=$LOCALLLM_REGISTRY,_IMAGE_NAME=$LOCALLLM_IMAGE_NAME

6. Configure a Cloud Workstation cluster.

   Wait for this to complete before moving forward which can take up to 20 minutes.

   gcloud workstations clusters create $LOCALLLM_CLUSTER \
     --region=$REGION

7. Create a Cloud Workstation configuration. We suggest using a machine type of e2-standard-32 which has 32 vCPU, 16 core and 128 GB memory.

   gcloud beta workstations configs create $LOCALLLM_WORKSTATION \
   --region=$REGION \
   --cluster=$LOCALLLM_CLUSTER \
   --machine-type=e2-standard-32 \
   --container-custom-image=us-central1-docker.pkg.dev/${PROJECT_ID}/${LOCALLLM_REGISTRY}/${LOCALLLM_IMAGE_NAME}:latest

8. Create a Cloud Workstation.

   gcloud workstations create $LOCALLLM_WORKSTATION \
   --cluster=$LOCALLLM_CLUSTER \
   --config=$LOCALLLM_WORKSTATION \
   --region=$REGION

9. Grant access to the default Cloud Workstation service account.

   gcloud artifacts repositories add-iam-policy-binding $LOCALLLM_REGISTRY \
     --location=$REGION \
     --member=serviceAccount:service-$PROJECT_NUM@gcp-sa-workstationsvm.iam.gserviceaccount.com \
     --role=roles/artifactregistry.reader

10. Start the workstation.

    gcloud workstations start $LOCALLLM_WORKSTATION \
      --cluster=$LOCALLLM_CLUSTER \
      --config=$LOCALLLM_WORKSTATION \
      --region=$REGION

11. Connect to the workstation using ssh. Alternatively, you can connect to the workstation interactively in the browser.

    gcloud workstations ssh $LOCALLLM_WORKSTATION \
      --cluster=$LOCALLLM_CLUSTER \
      --config=$LOCALLLM_WORKSTATION \
      --region=$REGION

12. Start serving the default model from the repo.

    local-llm run TheBloke/Llama-2-13B-Ensemble-v5-GGUF $LOCALLLM_PORT

13. Get the hostname of the workstation using:

    gcloud workstations describe $LOCALLLM_WORKSTATION \
      --cluster=$LOCALLLM_CLUSTER \
      --config=$LOCALLLM_WORKSTATION \
      --region=$REGION

14. Interact with the model by visiting the live OpenAPI documentation page: https://$LOCALLLM_PORT-$LLM_HOSTNAME/docs.

local-llm commands

[!NOTE] The command is now local-llm, however the original command (llm) is supported inside of the cloud workstations image.

Assumes that models are downloaded to ~/.cache/huggingface/hub/. This is the default cache path used by Hugging Face Hub library and only supports .gguf files.

If you're using models from TheBloke and you don't specify a filename, we'll attempt to use the model with 4 bit medium quantization, or you can specify a filename explicitly.

1. List downloaded models.

   local-llm list

2. List running models.

   local-llm ps

3. Start serving models.

   1. Start serving the default model from the repo. Download if not present.

      local-llm run TheBloke/Llama-2-13B-Ensemble-v5-GGUF 8000

   2. Start serving a specific model. Download if not present.

      local-llm run TheBloke/Llama-2-13B-Ensemble-v5-GGUF --filename llama-2-13b-ensemble-v5.Q4_K_S.gguf 8000

4. Stop serving models.

   1. Stop serving all models from the repo.

      local-llm kill TheBloke/Llama-2-13B-Ensemble-v5-GGUF

   2. Stop serving a specific model.

      local-llm kill TheBloke/Llama-2-13B-Ensemble-v5-GGUF --filename llama-2-13b-ensemble-v5.Q4_K_S.gguf

5. Download models.

   1. Download the default model from the repo.

      local-llm pull TheBloke/Llama-2-13B-Ensemble-v5-GGUF

   2. Download a specific model from the repo.

      local-llm pull TheBloke/Llama-2-13B-Ensemble-v5-GGUF --filename llama-2-13b-ensemble-v5.Q4_K_S.gguf

6. Remove models.

   1. Remove all models downloaded from the repo.

      local-llm rm TheBloke/Llama-2-13B-Ensemble-v5-GGUF

   2. Remove a specific model from the repo.

      local-llm rm TheBloke/Llama-2-13B-Ensemble-v5-GGUF --filename llama-2-13b-ensemble-v5.Q4_K_S.gguf

Running locally

1. Install the tools.

   # Install the tools
   pip3 install openai
   pip3 install ./local-llm/.

2. Download and run a model.

   local-llm run TheBloke/Llama-2-13B-Ensemble-v5-GGUF 8000

3. Try out a query. The default query is for a haiku about cats.

   python3 querylocal.py

4. Interact with the Open API interface via the /docs extension. For the above, visit http://localhost:8000/docs.

Debugging

To assist with debugging, you can configure model startup to write logs to a log file by providing a yaml python logging configuration file:

local-llm run TheBloke/Llama-2-13B-Ensemble-v5-GGUF 8000 --log-config <some config file>

To run locally using the bundled log config (log_config.yaml):

sudo touch /var/log/local-llm.log
sudo chown user:user /var/log/local-llm.log # use your user and group

# provide the log config manually
local-llm run TheBloke/Llama-2-13B-Ensemble-v5-GGUF 8000 --log-config local-llm/log_config.yaml

# or use an environment variable so you don't have to pass the argument
export LOG_CONFIG=$(pip show local-llm | grep Location | awk '{print $2}')/log_config.yaml
local-llm run TheBloke/Llama-2-13B-Ensemble-v5-GGUF 8000

You can follow the logs with:

tail -f /var/log/local-llm.log

If you are running multiple models, the logs from each will be written to the same file and interleaved.

If running from Cloud Workstations, logs from running models will be written to /var/log/local-llm.log (log_config.yaml is provided by default via the environment variable LOG_CONFIG within the image).

LLM Disclaimer

This project imports freely available LLMs and makes them available from Cloud Workstations. We recommend independently verifying any content generated by the models. We do not assume any responsibility or liability for the use or interpretation of generated content.