Awesome job.
#1
by
atopwhether
- opened
Kudos on an awesome finetune! This model is a breath of fresh air.
Pros:
- Alpaca, that is all.
- Intelligence: This is the only finetune I've personally tried that successfully maintains the original's intellectual depth.
- Engaging interactions: The model is extremely fun to converse with, having effectively eliminated generic boilerplate responses (e.g., "Certainly! Here's the code"). Instead, it playfully incorporates emojis and creative phrasing, such as "You know your AI will always deliver, here's your refactored code," which is awesome.
- Roleplaying: Although not my primary use case, I experimented with it multiple times. The model proved to be creative, uncensored, and—most importantly—passed the 'vibe check' with flying colors. On a few occasions, I encountered repetitive circular responses, but a simple OCC command resolved the issue: OCC: Conversational stalemate—break the cycle and gradually move forward.
- The list of positives goes on...
Cons:
None