Install LocalAI Models
Install a Model from the Midori AI Model Repo
Step 1:
- Start the Midori AI Subsystem
Step 2:
- On the Main Menu, Type
5
to Enter the Backend Program Menu
Step 3:
- On the Backend Program Menu, Type
10
to Enter the LocalAI Model Installer
Step 4a:
- If you have LocalAI installed in the subsystem, skip this step.
- If you do not have LocalAI installed in the subsystem, the program will ask you to enter the LocalAI docker’s name. It will look something like
localai-api-1
, but not always. If you need help, reach out on the Midori AI Discord / Email.
Step 4b:
- If you have GPU support installed in that image, type
yes
. - If you do not have GPU support installed in that image, type
no
.
Step 5:
- Type in the size you would like for your LLM and then follow the prompts in the manager!
Step 6:
- Sit Back and Let the Model Download from Midori AI’s Model Repo
- Don’t forget to note the name of the model you just installed so you can request it for OpenAI V1 later.
Need help on how to do that? Stop by - How to send OpenAI request to LocalAI
Install a Hugging Face Model from the Midori AI Model Repo
Step 1:
- Start the Midori AI Subsystem
Step 2:
- On the Main Menu, Type
5
to Enter the Backend Program Menu
Step 3:
- On the Backend Program Menu, Type
10
to Enter the LocalAI Model Installer
Step 4a:
- If you have LocalAI installed in the subsystem, skip this step.
- If you do not have LocalAI installed in the subsystem, the program will ask you to enter the LocalAI docker’s name. It will look something like
localai-api-1
, but not always. If you need help, reach out on the Midori AI Discord / Email.
Step 4b:
- If you have GPU support installed in that image, type
yes
. - If you do not have GPU support installed in that image, type
no
.
Step 5:
- Type
huggingface
when asked what size of model you would like.
Step 6:
- Copy and Paste the Hugging Face Download URL That You Wish to Use
- For example:
https://huggingface.co/mlabonne/gemma-7b-it-GGUF/resolve/main/gemma-7b-it.Q2_K.gguf?download=true
- Or you can use the huggingface naming from their api
- For example:
mlabonne/gemma-7b-it-GGUF/gemma-7b-it.Q2_K.gguf
Step 7:
- Sit Back and Let the Model Download from Midori AI’s Model Repo
- Don’t forget to note the name of the model you just installed so you can request it for OpenAI V1 later.
Need help on how to do that? Stop by - How to send OpenAI request to LocalAI