-
Notifications
You must be signed in to change notification settings - Fork 8
Models Configuration
Skills is another name of AI models running using the Python backend Elemental Golem. These skills are mostly local models but can also be wrappers for accessing APIs in OpenAI. The online skills list displays all the models that are running across all the backend servers. It allows any administrator account to stop a skill or change its configuration settings.
- #1 Configure Skill Change configuration data like API keys and other default settings
- #2 Stop Skill Stop a running skill
- #3 Skills List List of all the running skills/models, shows estimated ram, server, status. etc
- #4 Refresh Refresh the list of running skills
Each server will display back information like RAM, GPU RAM, HD and other stats that are useful for deciding where to download and run a model. The information about the server can be refreshed with the button in the upper right-hand corner.
Under each server there is a list of Installed Skills. You can run as many copies of the model as your system ram will hold and they will all be used for tasks.
- #1 Run Skill Opens the dialog box to select what device to run the model on
- #2 Device Options Select what device to run the skill on
- #3 Run model Run the model/skill
- #1 Download Skill Downloads the model from Hugging Face, Civit.ai or GIT repository
- #2 Custom Skill Opens dialog box to define a custom skill
Click on the Add Skill button. For the most part, you can copy and paste an existing definition and change a few key fields to point to your model on HuggingFace.
Label Must be unique.
Routing Key Must be unique.
Moe Domain Definitions for model routing.
Moe Function Definitions for model routing.
Model > Name The last part of the Hugging Face URL for the model repository. For example TheBloke/Mixtral-8x7B-Instruct-v0.1-GGUF for HF URL https://huggingface.co/TheBloke/Mixtral-8x7B-Instruct-v0.1-GGUF