SmileyGPT 1.2.0 EXPERIMENTAL. The fuck is this doing here
Added 2025-02-27 00:53:40 +0000 UTCSo decided totake a break from SchizoSeek (help my brain hurts) to play with this a little. Improved some things
- Added prefill prompt to "Assistant Prefill` field with slight modifications to improve Claude performance.
- Set default OpenAI Model to `gpt-4o-2024-11-20`.
- Set default Claude Model to `claude-3-7-sonnet-20250219`.
- Added `π | Anti-Claudism` mutator.
- Added `π | Message Simulator` mutator.
- Slightly tweaked `π | Status Box` mutator.
- Edited `π | Clau's role (char)` to match SmileySeek prompt.
- Edited `π | Clau's role (user)` to match SmileySeek prompt.
- Edited `π | Clau's role (world)` to match SmileySeek prompt.
It's been a while since I changed this one, huh? There was not much change for GPT as its performance should be more or less the same with the default model (a minute of silence for `chatgpt-4o-latest`).
Most of the changes are things I made while using JB and didn't like or just thought they were better than the old stuff. I have tested it with Sorbet 3.7 and I can say that it FUCKING COOKS, not perfect but it works. However, I recommend doing the following to remove as many filters as possible for that model:
- Disable `π | Policies Thingy` prompt.
- Disable `π | Assistant Instructions` prompt.
Enjoy your meal you degenerate people.
