Exceptional Model
Just wanted to congratulate on this incredible finetune.
I have tested a lot of models that can work on a 16+8 gb vram setup that I have, nothing comes close to the base models EXCEPT this. It surpassed the base model in my opinion. This does what it says caps the thinking and thus produces better results.
Same here, best model i ever used local.
It works so much better for me than stock qwen 27b, mostly in stopping looping. I hardly ever now get thinking loops, which is my main critique of stock qwen3.6. Also, appears that it doesn't compromise on quality, as it looks about on the same level of capability.
It works so much better for me than stock qwen 27b, mostly in stopping looping. I hardly ever now get thinking loops, which is my main critique of stock qwen3.6. Also, appears that it doesn't compromise on quality, as it looks about on the same level of capability.
too bad i found it so late , 3.8 will be out in a week or so.
It works so much better for me than stock qwen 27b, mostly in stopping looping. I hardly ever now get thinking loops, which is my main critique of stock qwen3.6. Also, appears that it doesn't compromise on quality, as it looks about on the same level of capability.
too bad i found it so late , 3.8 will be out in a week or so.
3.8 will need its own "ThinkingCap", yes? 😊
3.8 will need its own "ThinkingCap", yes? 😊
This reminds me of The Wandering Inn. If you don't know what I am talking about, then just forget that I said anything.