← Return to tool
Log in with itch.io to leave a comment.
did you make/train your own local model for this or use ollama premade model?
By default, Shatei uses “qwen3.5:9b-q4_k_m”. After testing various models, this is the one that yielded the best results in an 8GB memory environment.
However, you can change the “model”: “qwen3.5:9b-q4_k_m” section in "config.json" to use any model available on your Ollama instance.
If the model listed here is officially supported by Ollama and is not yet present in your environment, it will be automatically downloaded at startup.
Enjoy!
could i set up agentic pipeline utilizing the config?
thank you ,i have to use a smaller pipeline format for low end hardware :_)
if you compile with nuitka i do too! i can possible help with the AV flags. Dillonmontana@yahoo.com
← Return to tool
Comments
Log in with itch.io to leave a comment.
did you make/train your own local model for this or use ollama premade model?
By default, Shatei uses “qwen3.5:9b-q4_k_m”. After testing various models, this is the one that yielded the best results in an 8GB memory environment.
However, you can change the “model”: “qwen3.5:9b-q4_k_m” section in "config.json" to use any model available on your Ollama instance.
If the model listed here is officially supported by Ollama and is not yet present in your environment, it will be automatically downloaded at startup.
Enjoy!
could i set up agentic pipeline utilizing the config?
thank you ,i have to use a smaller pipeline format for low end hardware :_)
if you compile with nuitka i do too! i can possible help with the AV flags. Dillonmontana@yahoo.com