Meow-GPT

Run all in containers

HTTP_PORT=8888 LLM=deepseek-coder OLLAMA_BASE_URL=http://ollama-service:11434 docker compose --profile container up
# or LLM=deepseek-coder OLLAMA_BASE_URL=http://ollama-service:11434 docker compose --profile container watch

The first time only, you must wait for the complete downloading of the model.

Use the native Ollama install

To do for the first time only:

LLM=deepseek-coder
ollama pull ${LLM}

HTTP_PORT=8888 LLM=deepseek-coder OLLAMA_BASE_URL=http://host.docker.internal:11434 docker compose --profile webapp up
# or LLM=deepseek-coder OLLAMA_BASE_URL=http://host.docker.internal:11434 docker compose --profile webapp watch

Use a specific env file

docker compose --env-file deepseek-coder-instruct.env --profile webapp up

ollama pull gemma:2b-instruct
docker compose --env-file gemma-2b-instruct.env --profile webapp up

Open the Web UI

http://localhost:8888

Development

cd webapp
# yarn install
LLM=deepseek-coder
OLLAMA_BASE_URL=http://localhost:11434 
node index.mjs