HN Explorer

gslaller

↗ HN profile [ 81.3 ms ] full profile
Karma
0
Created
()
Submissions
0
  1. Show HN: Moe Routing Atlas – which experts fire in Qwen3.5/3.6 35B-A3B (gslaller.github.io)

    Did some work on visualising the firing pattern of MoE (Mixture of Experts) modules for two Qwen models. There are some other interesting statistics, such as router entropy, expert reuse from the previous token…

    1 points 0
  2. MoE Analysis Qwen[3.5|3.6]35B-A3B (gslaller.github.io)
    1 points 1
  3. Edtech, it's failure or yet missing success

    Arguably, YouTube (YT) is the best education technology platform on the internet, yet it is just a non-creative successor of the VHS/TV and not even CD(which had the menu, breaking the linearity of the frames). The bits…

    5 points 4