melinoe-30b-a3b-thinking-i1/Melinoe-30B-A3B-Thinking.i1-Q6_K.gguf
Hi!, I really like this model, how it is trained. melinoe-30b-a3b-thinking-i1,
This model seemed to be very good, had allot of personality, but it gets "broken" like gets in incoherent loops, talks allot of random stuff, repeats same sentences.
Is this something that will be fixed?🤔
Tried the other one, that was even more "broken".
melinoe-30b-a3b-thinking/Melinoe-30B-A3B-Thinking.Q8_0.gguf
Thank you so much for providing this feedback. It really makes me happy to hear that people are finding the model valuable. What I can do is, I will try diluting the model by merging it with the base model's weights. I believe that should address the most obvious issues. I will also carefully adjust the training procedure of future model fine tunes in order to shore up any identified problematic areas.
Any options to make it speaks other languages qwen3 supports ?