Pinned post
Aesther
Verified account This account is verified. Learn more@AestherML
AI Researcher
635 Following 63K Followers 3.8K Posts
Latest posts 5
I am genuinely impressed with Inkling-2-Mini so far.
Definitely better than Qwen 3.8 27B, and much faster if you get the right inference engine!
This is how American labs should compete with China; through innovation, not regulation.
Definitely better than Qwen 3.8 27B, and much faster if you get the right inference engine!
This is how American labs should compete with China; through innovation, not regulation.
You should actually never set Min P to 0
0 turns the models to an essentially random number generator when things get even remotely confusing
Incredibly enough, 3 out of 5 of my friends that were complaining about Qwen 3.8 27B had this setting wrong 😅
0 turns the models to an essentially random number generator when things get even remotely confusing
Incredibly enough, 3 out of 5 of my friends that were complaining about Qwen 3.8 27B had this setting wrong 😅
GTA 6 will be a total flop
All they had to do was implement AI NPCs and destructible buildings
They completely blew it
All they had to do was implement AI NPCs and destructible buildings
They completely blew it
It took over ten years to get a successor to Vine, and it seemed obvious succession should had been immediate for something so important, but no one wanted to do it before TikTok?