search

rendezqueue / rendezllama

CLI for llama.cpp with various commands to guide, edit, and regenerate tokens on the fly.
ISC License
10 stars 1 forks source link

feat: Provide a deterministic preset for testing new models #38

Open grencez opened 1 month ago

grencez commented 1 month ago

I'd like to run the example prompts deterministically.

I tried to use a preset with multiple --x_setting options but got really bad results due to how settings override each other. Maybe "preset" is the wrong word here... maybe sampling settings should be removed from all example prompts so that they don't override presets. :shrug: