From ea9b7e692862e515e9704d1e6bde3e976db4dc01 Mon Sep 17 00:00:00 2001 From: Matteo Collina Date: Wed, 8 May 2024 13:05:49 +0200 Subject: [PATCH] moar guide Signed-off-by: Matteo Collina --- CONTRIBUTING.md | 22 ++++++++++++++++++++-- 1 file changed, 20 insertions(+), 2 deletions(-) diff --git a/CONTRIBUTING.md b/CONTRIBUTING.md index 0224565..fc28406 100644 --- a/CONTRIBUTING.md +++ b/CONTRIBUTING.md @@ -38,12 +38,30 @@ Steps for downloading and setting up AI Warp for local development. is located at `ai-warp-app/platformatic.json`. **Note: this will be overwrited every time you generate the test app.** - 8. Start the test app. + 8. Start the test app. From the `app-warp-ai` folder, run: ```bash - npm start + node ../dist/cli/start.js ``` +### Testing a local model with llama2 + +To test a local model with with llama2, you can use the following to downloaded a tested model: + +```bash +curl -L -O https://huggingface.co/TheBloke/Mistral-7B-Instruct-v0.2-GGUF/resolve/main/mistral-7b-instruct-v0.2.Q8_0.gguf +``` + +Then, in your `platformatic.json` file, add: + +``` + "aiProvider": { + "llama2": { + "modelPath": "./mistral-7b-instruct-v0.2.Q8_0.gguf" + } + }, +``` + ## Important Notes * AI Warp needs to be rebuilt for any code change to take affect in your test