Edit model card

Trained on filtered, not fully cleaned Logs & Instruct Data.

experimental

has a bunch of problems I did not sort.

  • </s> at end of responses - didnt update fastchat whoops my bad
  • Occasional Run-off Generations at the end - data filter issue
  • Issues with codeblocks etc at end of generation. - filter issue
  • Sucks at storywriting / narrator - eh not built for that, could work tho. needs prompting.

Recommended params: Temp of 1.32, min-p of 0.1 - temp last.

Format: Vicuna / Alpaca / ChatML / whatever lmao

Stopping Strings: ['\n{{char}}', '\n{{user}}', '```'] Stops most bs like incomplete code blocks randomly inserted etc. [replace macro with your character and user's name if using another frontend]

A bit horny but sticks to char well. meant for rp with character cards

Swipe a few times and you'll get sovl.

Data

Downloads last month
950
Inference API
Model is too large to load in Inference API (serverless). To try the model, launch it on Inference Endpoints (dedicated) instead.