peonist-ai/halogen-qwen3.8-flash-next · Discussions

Resources

View closed (0)

Variable MTP and ngram

#9 opened 6 days ago by

Can the model size be further reduced to lower initial loading overhead

#8 opened 8 days ago by

What an absolute godlike tool. The speed is insane!

#7 opened 11 days ago by

Q5 or Q6 quants ?

#6 opened 13 days ago by

Donato's video on halogen & qwen3.8 flash

#5 opened 15 days ago by

Ablated model?

#4 opened 17 days ago by

How about a 3-bit quant?

#3 opened 18 days ago by

Engram SSD Offloading

#2 opened 21 days ago by

Very fast.

#1 opened 22 days ago by