News for hackers. Stuff that matters.

A comment on Qwen3-235B-A22B-Thinking-2507. Parent comment

Re:Qwen3-235B-A22B-Thinking-2507 by danielhanchen on Friday July 25, 2025 @07:41AM (#44682047)
Oh I would always use Q2_K_XL :) It uses our dynamic methodology to quantize certain layers in different bits ie 2, 3, 4, 5, 6, 8 bits - the more important the layer is, the higher the bitrate
[ Parent | on HN ]
Re:Qwen3-235B-A22B-Thinking-2507 by Squeeze2664 on Friday July 25, 2025 @09:32AM (#44682941)
How do you determine the importance of a layer in this case?
[ Parent | on HN ]
Re:Qwen3-235B-A22B-Thinking-2507 by kkzz99 on Friday July 25, 2025 @12:04PM (#44684746)
Afaik they have a test bench that they use and take the activation data from that.
[ Parent | on HN ]
Re:Qwen3-235B-A22B-Thinking-2507 by danielhanchen on Friday July 25, 2025 @04:18PM (#44687932)
Yes we have around 1 to 3 million tokens of high quality self verified data that we use to calibrate models!
[ Parent | on HN ]
Re:Qwen3-235B-A22B-Thinking-2507 by smallerize on Friday July 25, 2025 @01:14PM (#44685574)
https://unsloth.ai/blog/dynamic-v2
[ Parent | on HN ]