Create freckles_observer_v40_freckles_noise.json
Browse files
freckles_observer_v40_freckles_noise.json
ADDED
|
@@ -0,0 +1,1369 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"activation_stats": {
|
| 3 |
+
"enc_in": {
|
| 4 |
+
"mean": 0.11399824917316437,
|
| 5 |
+
"std": 0.32156893610954285,
|
| 6 |
+
"min": -4.077515602111816,
|
| 7 |
+
"max": 4.074148654937744,
|
| 8 |
+
"abs_mean": 0.25178295373916626,
|
| 9 |
+
"dead_frac": 2.543131586207892e-06,
|
| 10 |
+
"sparsity": 0.0,
|
| 11 |
+
"kurtosis": 4.910905838012695
|
| 12 |
+
},
|
| 13 |
+
"enc_block_0": {
|
| 14 |
+
"mean": -0.02227744832634926,
|
| 15 |
+
"std": 0.2875443994998932,
|
| 16 |
+
"min": -3.1000189781188965,
|
| 17 |
+
"max": 2.062558650970459,
|
| 18 |
+
"abs_mean": 0.17517945170402527,
|
| 19 |
+
"dead_frac": 4.450480446394067e-06,
|
| 20 |
+
"sparsity": 0.0,
|
| 21 |
+
"kurtosis": 19.296550750732422
|
| 22 |
+
},
|
| 23 |
+
"enc_block_1": {
|
| 24 |
+
"mean": -0.005942739546298981,
|
| 25 |
+
"std": 0.10117179155349731,
|
| 26 |
+
"min": -1.1117786169052124,
|
| 27 |
+
"max": 0.7992860078811646,
|
| 28 |
+
"abs_mean": 0.0745168998837471,
|
| 29 |
+
"dead_frac": 1.0172526344831567e-05,
|
| 30 |
+
"sparsity": 0.0,
|
| 31 |
+
"kurtosis": 2.5130233764648438
|
| 32 |
+
},
|
| 33 |
+
"enc_block_2": {
|
| 34 |
+
"mean": -0.021667584776878357,
|
| 35 |
+
"std": 0.11653400212526321,
|
| 36 |
+
"min": -0.7383712530136108,
|
| 37 |
+
"max": 0.6876223683357239,
|
| 38 |
+
"abs_mean": 0.09079861640930176,
|
| 39 |
+
"dead_frac": 8.106231689453125e-06,
|
| 40 |
+
"sparsity": 0.0,
|
| 41 |
+
"kurtosis": 1.041348934173584
|
| 42 |
+
},
|
| 43 |
+
"enc_block_3": {
|
| 44 |
+
"mean": -0.0528937391936779,
|
| 45 |
+
"std": 0.18374653160572052,
|
| 46 |
+
"min": -0.9734799861907959,
|
| 47 |
+
"max": 0.8834015130996704,
|
| 48 |
+
"abs_mean": 0.14933942258358002,
|
| 49 |
+
"dead_frac": 3.814697265625e-06,
|
| 50 |
+
"sparsity": 0.0,
|
| 51 |
+
"kurtosis": 0.4822535514831543
|
| 52 |
+
},
|
| 53 |
+
"enc_out_raw": {
|
| 54 |
+
"mean": -0.17712435126304626,
|
| 55 |
+
"std": 3.6461434364318848,
|
| 56 |
+
"min": -10.889140129089355,
|
| 57 |
+
"max": 11.768145561218262,
|
| 58 |
+
"abs_mean": 2.373218536376953,
|
| 59 |
+
"dead_frac": 1.9073486328125e-06,
|
| 60 |
+
"sparsity": 0.0,
|
| 61 |
+
"kurtosis": 1.0141587257385254
|
| 62 |
+
},
|
| 63 |
+
"cross_attn_0_qkv": {
|
| 64 |
+
"mean": 0.02815081924200058,
|
| 65 |
+
"std": 0.9149454832077026,
|
| 66 |
+
"min": -1.9138216972351074,
|
| 67 |
+
"max": 1.4539501667022705,
|
| 68 |
+
"abs_mean": 0.777122974395752,
|
| 69 |
+
"dead_frac": 0.0,
|
| 70 |
+
"sparsity": 0.0,
|
| 71 |
+
"kurtosis": -0.6490051746368408
|
| 72 |
+
},
|
| 73 |
+
"cross_attn_0_in": {
|
| 74 |
+
"mean": 3.307823657989502,
|
| 75 |
+
"std": 1.028746247291565,
|
| 76 |
+
"min": 1.6925034523010254,
|
| 77 |
+
"max": 4.709893703460693,
|
| 78 |
+
"abs_mean": 3.307823657989502,
|
| 79 |
+
"dead_frac": 0.0,
|
| 80 |
+
"sparsity": 0.0,
|
| 81 |
+
"kurtosis": -1.4075615406036377
|
| 82 |
+
},
|
| 83 |
+
"cross_attn_0_out": {
|
| 84 |
+
"mean": 3.362860679626465,
|
| 85 |
+
"std": 1.0373820066452026,
|
| 86 |
+
"min": 1.7293139696121216,
|
| 87 |
+
"max": 4.793014049530029,
|
| 88 |
+
"abs_mean": 3.362860679626465,
|
| 89 |
+
"dead_frac": 0.0,
|
| 90 |
+
"sparsity": 0.0,
|
| 91 |
+
"kurtosis": -1.3798936605453491
|
| 92 |
+
},
|
| 93 |
+
"cross_attn_1_qkv": {
|
| 94 |
+
"mean": -0.011559374630451202,
|
| 95 |
+
"std": 0.8941014409065247,
|
| 96 |
+
"min": -1.2823809385299683,
|
| 97 |
+
"max": 2.239341974258423,
|
| 98 |
+
"abs_mean": 0.6763687133789062,
|
| 99 |
+
"dead_frac": 0.0,
|
| 100 |
+
"sparsity": 0.0,
|
| 101 |
+
"kurtosis": 0.4769473075866699
|
| 102 |
+
},
|
| 103 |
+
"cross_attn_1_in": {
|
| 104 |
+
"mean": 3.362860679626465,
|
| 105 |
+
"std": 1.0373820066452026,
|
| 106 |
+
"min": 1.7293139696121216,
|
| 107 |
+
"max": 4.793014049530029,
|
| 108 |
+
"abs_mean": 3.362860679626465,
|
| 109 |
+
"dead_frac": 0.0,
|
| 110 |
+
"sparsity": 0.0,
|
| 111 |
+
"kurtosis": -1.3798936605453491
|
| 112 |
+
},
|
| 113 |
+
"cross_attn_1_out": {
|
| 114 |
+
"mean": 3.353649616241455,
|
| 115 |
+
"std": 0.9911406636238098,
|
| 116 |
+
"min": 1.7649903297424316,
|
| 117 |
+
"max": 4.691677093505859,
|
| 118 |
+
"abs_mean": 3.353649616241455,
|
| 119 |
+
"dead_frac": 0.0,
|
| 120 |
+
"sparsity": 0.0,
|
| 121 |
+
"kurtosis": -1.3825546503067017
|
| 122 |
+
},
|
| 123 |
+
"dec_in": {
|
| 124 |
+
"mean": 0.48801693320274353,
|
| 125 |
+
"std": 0.4462810754776001,
|
| 126 |
+
"min": -0.5784251689910889,
|
| 127 |
+
"max": 8.451927185058594,
|
| 128 |
+
"abs_mean": 0.49386128783226013,
|
| 129 |
+
"dead_frac": 0.0,
|
| 130 |
+
"sparsity": 0.0,
|
| 131 |
+
"kurtosis": 195.43594360351562
|
| 132 |
+
},
|
| 133 |
+
"dec_block_0": {
|
| 134 |
+
"mean": -0.1455148309469223,
|
| 135 |
+
"std": 0.26857054233551025,
|
| 136 |
+
"min": -3.289444923400879,
|
| 137 |
+
"max": 4.344697952270508,
|
| 138 |
+
"abs_mean": 0.19442293047904968,
|
| 139 |
+
"dead_frac": 3.1789145396032836e-06,
|
| 140 |
+
"sparsity": 0.0,
|
| 141 |
+
"kurtosis": 59.326595306396484
|
| 142 |
+
},
|
| 143 |
+
"dec_block_1": {
|
| 144 |
+
"mean": -0.017345234751701355,
|
| 145 |
+
"std": 0.1320338398218155,
|
| 146 |
+
"min": -1.80828058719635,
|
| 147 |
+
"max": 1.4308905601501465,
|
| 148 |
+
"abs_mean": 0.09153717756271362,
|
| 149 |
+
"dead_frac": 9.695689186628442e-06,
|
| 150 |
+
"sparsity": 0.0,
|
| 151 |
+
"kurtosis": 13.288253784179688
|
| 152 |
+
},
|
| 153 |
+
"dec_block_2": {
|
| 154 |
+
"mean": -0.030033249408006668,
|
| 155 |
+
"std": 0.13502919673919678,
|
| 156 |
+
"min": -0.9543643593788147,
|
| 157 |
+
"max": 1.3100506067276,
|
| 158 |
+
"abs_mean": 0.09647099673748016,
|
| 159 |
+
"dead_frac": 8.106231689453125e-06,
|
| 160 |
+
"sparsity": 0.0,
|
| 161 |
+
"kurtosis": 12.214755058288574
|
| 162 |
+
},
|
| 163 |
+
"dec_block_3": {
|
| 164 |
+
"mean": -0.07281665503978729,
|
| 165 |
+
"std": 0.165196493268013,
|
| 166 |
+
"min": -1.7998610734939575,
|
| 167 |
+
"max": 2.990985631942749,
|
| 168 |
+
"abs_mean": 0.1313476860523224,
|
| 169 |
+
"dead_frac": 3.973643288190942e-06,
|
| 170 |
+
"sparsity": 0.0,
|
| 171 |
+
"kurtosis": 32.06653594970703
|
| 172 |
+
},
|
| 173 |
+
"dec_out": {
|
| 174 |
+
"mean": 0.009957308880984783,
|
| 175 |
+
"std": 1.2505145072937012,
|
| 176 |
+
"min": -3.9976625442504883,
|
| 177 |
+
"max": 4.018248081207275,
|
| 178 |
+
"abs_mean": 1.0230882167816162,
|
| 179 |
+
"dead_frac": 0.0,
|
| 180 |
+
"sparsity": 0.0,
|
| 181 |
+
"kurtosis": -0.7579162120819092
|
| 182 |
+
},
|
| 183 |
+
"boundary_in": {
|
| 184 |
+
"mean": 0.009957308880984783,
|
| 185 |
+
"std": 1.2505145072937012,
|
| 186 |
+
"min": -3.9976625442504883,
|
| 187 |
+
"max": 4.018248081207275,
|
| 188 |
+
"abs_mean": 1.0230880975723267,
|
| 189 |
+
"dead_frac": 0.0,
|
| 190 |
+
"sparsity": 0.0,
|
| 191 |
+
"kurtosis": -0.7579159736633301
|
| 192 |
+
},
|
| 193 |
+
"boundary_out": {
|
| 194 |
+
"mean": 6.105394277255982e-05,
|
| 195 |
+
"std": 1.2510300874710083,
|
| 196 |
+
"min": -4.003395080566406,
|
| 197 |
+
"max": 4.002503395080566,
|
| 198 |
+
"abs_mean": 1.0234731435775757,
|
| 199 |
+
"dead_frac": 0.0,
|
| 200 |
+
"sparsity": 0.0,
|
| 201 |
+
"kurtosis": -0.75799560546875
|
| 202 |
+
},
|
| 203 |
+
"svd_U": {
|
| 204 |
+
"mean": -0.0015054205432534218,
|
| 205 |
+
"std": 0.1443297415971756,
|
| 206 |
+
"min": -0.5084033012390137,
|
| 207 |
+
"max": 0.5471221208572388,
|
| 208 |
+
"abs_mean": 0.11567305028438568,
|
| 209 |
+
"dead_frac": 6.039937488822034e-06,
|
| 210 |
+
"sparsity": 0.0,
|
| 211 |
+
"kurtosis": 0.46650028228759766
|
| 212 |
+
},
|
| 213 |
+
"svd_S_orig": {
|
| 214 |
+
"mean": 3.307823657989502,
|
| 215 |
+
"std": 1.028746247291565,
|
| 216 |
+
"min": 1.6925034523010254,
|
| 217 |
+
"max": 4.709893703460693,
|
| 218 |
+
"abs_mean": 3.307823657989502,
|
| 219 |
+
"dead_frac": 0.0,
|
| 220 |
+
"sparsity": 0.0,
|
| 221 |
+
"kurtosis": -1.4075615406036377
|
| 222 |
+
},
|
| 223 |
+
"svd_S": {
|
| 224 |
+
"mean": 3.353649616241455,
|
| 225 |
+
"std": 0.9911406636238098,
|
| 226 |
+
"min": 1.7649903297424316,
|
| 227 |
+
"max": 4.691677093505859,
|
| 228 |
+
"abs_mean": 3.353649616241455,
|
| 229 |
+
"dead_frac": 0.0,
|
| 230 |
+
"sparsity": 0.0,
|
| 231 |
+
"kurtosis": -1.3825546503067017
|
| 232 |
+
},
|
| 233 |
+
"svd_Vt": {
|
| 234 |
+
"mean": 0.037834037095308304,
|
| 235 |
+
"std": 0.49856749176979065,
|
| 236 |
+
"min": -0.9999575018882751,
|
| 237 |
+
"max": 0.9589366316795349,
|
| 238 |
+
"abs_mean": 0.37048596143722534,
|
| 239 |
+
"dead_frac": 0.0,
|
| 240 |
+
"sparsity": 0.0,
|
| 241 |
+
"kurtosis": -0.25922727584838867
|
| 242 |
+
},
|
| 243 |
+
"svd_M": {
|
| 244 |
+
"mean": -0.029393270611763,
|
| 245 |
+
"std": 0.4991353750228882,
|
| 246 |
+
"min": -0.9999995231628418,
|
| 247 |
+
"max": 0.9999973177909851,
|
| 248 |
+
"abs_mean": 0.33501312136650085,
|
| 249 |
+
"dead_frac": 5.7220458984375e-06,
|
| 250 |
+
"sparsity": 0.0,
|
| 251 |
+
"kurtosis": 0.24239110946655273
|
| 252 |
+
},
|
| 253 |
+
"recon": {
|
| 254 |
+
"mean": 6.105394277255982e-05,
|
| 255 |
+
"std": 1.2510300874710083,
|
| 256 |
+
"min": -4.003395080566406,
|
| 257 |
+
"max": 4.002503395080566,
|
| 258 |
+
"abs_mean": 1.0234731435775757,
|
| 259 |
+
"dead_frac": 0.0,
|
| 260 |
+
"sparsity": 0.0,
|
| 261 |
+
"kurtosis": -0.75799560546875
|
| 262 |
+
},
|
| 263 |
+
"input": {
|
| 264 |
+
"mean": -2.6453908503754064e-05,
|
| 265 |
+
"std": 1.250998616218567,
|
| 266 |
+
"min": -4.0,
|
| 267 |
+
"max": 4.0,
|
| 268 |
+
"abs_mean": 1.0234488248825073,
|
| 269 |
+
"dead_frac": 1.271565793103946e-06,
|
| 270 |
+
"sparsity": 0.0,
|
| 271 |
+
"kurtosis": -0.7580161094665527
|
| 272 |
+
}
|
| 273 |
+
},
|
| 274 |
+
"svd_bottleneck": {
|
| 275 |
+
"S_mean": [
|
| 276 |
+
4.565850734710693,
|
| 277 |
+
3.915027141571045,
|
| 278 |
+
2.8913984298706055,
|
| 279 |
+
1.859018325805664
|
| 280 |
+
],
|
| 281 |
+
"S_std": [
|
| 282 |
+
0.03792751207947731,
|
| 283 |
+
0.0469575896859169,
|
| 284 |
+
0.05345858260989189,
|
| 285 |
+
0.05418039485812187
|
| 286 |
+
],
|
| 287 |
+
"S_ratio": 2.4560546875,
|
| 288 |
+
"condition_number": 2.4560546875,
|
| 289 |
+
"effective_rank": 1.1468989849090576,
|
| 290 |
+
"recon_error": 1.4800130454409137e-15,
|
| 291 |
+
"U_orthogonality_error": 3.4124291408674278e-15,
|
| 292 |
+
"Vt_orthogonality_error": 6.465710594626638e-16,
|
| 293 |
+
"energy_per_mode": [
|
| 294 |
+
0.43434232473373413,
|
| 295 |
+
0.31936755776405334,
|
| 296 |
+
0.17423005402088165,
|
| 297 |
+
0.07206007838249207
|
| 298 |
+
],
|
| 299 |
+
"sphere_radius_mean": 1.0,
|
| 300 |
+
"sphere_radius_std": 4.2930455634859754e-08
|
| 301 |
+
},
|
| 302 |
+
"cross_attention": {
|
| 303 |
+
"layer_0_delta": {
|
| 304 |
+
"delta_abs_mean": 0.05503721535205841,
|
| 305 |
+
"delta_abs_max": 0.08312034606933594,
|
| 306 |
+
"delta_std": 0.01944689266383648,
|
| 307 |
+
"delta_per_mode": [
|
| 308 |
+
0.08059147000312805,
|
| 309 |
+
0.03247781842947006,
|
| 310 |
+
0.06663240492343903,
|
| 311 |
+
0.040447160601615906
|
| 312 |
+
],
|
| 313 |
+
"alpha_values": [
|
| 314 |
+
0.022388869896531105,
|
| 315 |
+
0.024696987122297287,
|
| 316 |
+
0.02641209028661251,
|
| 317 |
+
0.02643170952796936
|
| 318 |
+
],
|
| 319 |
+
"alpha_mean": 0.02498241513967514,
|
| 320 |
+
"relative_change": 0.0166384968906641,
|
| 321 |
+
"sign_agreement": 1.0
|
| 322 |
+
},
|
| 323 |
+
"layer_0_attention": {
|
| 324 |
+
"q_norm_mean": 2.305610179901123,
|
| 325 |
+
"k_norm_mean": 1.3561913967132568,
|
| 326 |
+
"v_norm_mean": 1.7022989988327026,
|
| 327 |
+
"q_std": 1.1526931524276733,
|
| 328 |
+
"k_std": 0.672618567943573,
|
| 329 |
+
"v_std": 0.8512021899223328,
|
| 330 |
+
"qk_cosine": -0.29407045245170593,
|
| 331 |
+
"qv_cosine": -0.21351489424705505,
|
| 332 |
+
"kv_cosine": 0.7354511022567749
|
| 333 |
+
},
|
| 334 |
+
"layer_1_delta": {
|
| 335 |
+
"delta_abs_mean": 0.04028972610831261,
|
| 336 |
+
"delta_abs_max": 0.10133695602416992,
|
| 337 |
+
"delta_std": 0.05339942127466202,
|
| 338 |
+
"delta_per_mode": [
|
| 339 |
+
0.09837014973163605,
|
| 340 |
+
0.000631974427960813,
|
| 341 |
+
0.02289256826043129,
|
| 342 |
+
0.03926421329379082
|
| 343 |
+
],
|
| 344 |
+
"alpha_values": [
|
| 345 |
+
0.025104766711592674,
|
| 346 |
+
0.02187989465892315,
|
| 347 |
+
0.02244846150279045,
|
| 348 |
+
0.027547497302293777
|
| 349 |
+
],
|
| 350 |
+
"alpha_mean": 0.02424515411257744,
|
| 351 |
+
"relative_change": 0.011980789713561535,
|
| 352 |
+
"sign_agreement": 0.5
|
| 353 |
+
},
|
| 354 |
+
"layer_1_attention": {
|
| 355 |
+
"q_norm_mean": 2.4621071815490723,
|
| 356 |
+
"k_norm_mean": 1.233809471130371,
|
| 357 |
+
"v_norm_mean": 1.4172993898391724,
|
| 358 |
+
"q_std": 1.1312413215637207,
|
| 359 |
+
"k_std": 0.43080684542655945,
|
| 360 |
+
"v_std": 0.7043910026550293,
|
| 361 |
+
"qk_cosine": 0.2318161427974701,
|
| 362 |
+
"qv_cosine": 0.8531496524810791,
|
| 363 |
+
"kv_cosine": 0.6193335056304932
|
| 364 |
+
}
|
| 365 |
+
},
|
| 366 |
+
"enc_dec_symmetry": {
|
| 367 |
+
"block_0_spearman": 0.03686881909989656,
|
| 368 |
+
"block_0_cosine": 0.07803580909967422,
|
| 369 |
+
"block_0_procrustes_error": 8.642137458991783e-07,
|
| 370 |
+
"block_0_procrustes_alignment": 0.9170370101928711,
|
| 371 |
+
"block_1_spearman": 0.009047081100090216,
|
| 372 |
+
"block_1_cosine": 0.030039409175515175,
|
| 373 |
+
"block_1_procrustes_error": 1.0774815564218443e-06,
|
| 374 |
+
"block_1_procrustes_alignment": 0.896563708782196,
|
| 375 |
+
"block_2_spearman": 0.01232992455646543,
|
| 376 |
+
"block_2_cosine": 0.04046924412250519,
|
| 377 |
+
"block_2_procrustes_error": 8.364959853679466e-07,
|
| 378 |
+
"block_2_procrustes_alignment": 0.9196978211402893,
|
| 379 |
+
"block_3_spearman": -0.0008597959765885626,
|
| 380 |
+
"block_3_cosine": 0.16444019973278046,
|
| 381 |
+
"block_3_procrustes_error": 9.134143397204753e-07,
|
| 382 |
+
"block_3_procrustes_alignment": 0.912315309047699
|
| 383 |
+
},
|
| 384 |
+
"information_flow": {
|
| 385 |
+
"input_to_enc_spearman": -0.01943134982072711,
|
| 386 |
+
"M_to_S_compression": 1.0,
|
| 387 |
+
"M_variance": 0.24913613498210907,
|
| 388 |
+
"S_variance": 1.058318853378296,
|
| 389 |
+
"variance_retention": 4.247953878608404,
|
| 390 |
+
"cross_attn_total_delta": 0.05471521615982056,
|
| 391 |
+
"cross_attn_max_delta": 0.09769392013549805,
|
| 392 |
+
"cross_attn_relative_delta": 0.01654115319252014,
|
| 393 |
+
"dec_to_recon_spearman": 0.004384062372986261,
|
| 394 |
+
"end_to_end_spearman": 0.9999988479208689,
|
| 395 |
+
"end_to_end_mse": 7.40700807000394e-06,
|
| 396 |
+
"boundary_delta_mean": 0.00989625509828329,
|
| 397 |
+
"boundary_delta_max": 0.018932104110717773,
|
| 398 |
+
"boundary_relative": 0.009672925807535648
|
| 399 |
+
},
|
| 400 |
+
"cv_stages": {
|
| 401 |
+
"enc_in": {
|
| 402 |
+
"cv": 0.3957862952422528,
|
| 403 |
+
"name": "Encoder input projection",
|
| 404 |
+
"dim": 16
|
| 405 |
+
},
|
| 406 |
+
"enc_block_0": {
|
| 407 |
+
"cv": 0.3340476673895933,
|
| 408 |
+
"name": "Encoder block 0",
|
| 409 |
+
"dim": 16
|
| 410 |
+
},
|
| 411 |
+
"enc_block_1": {
|
| 412 |
+
"cv": 0.3956829632095093,
|
| 413 |
+
"name": "Encoder block 1",
|
| 414 |
+
"dim": 16
|
| 415 |
+
},
|
| 416 |
+
"enc_block_2": {
|
| 417 |
+
"cv": 0.3878920391524174,
|
| 418 |
+
"name": "Encoder block 2",
|
| 419 |
+
"dim": 16
|
| 420 |
+
},
|
| 421 |
+
"enc_block_3": {
|
| 422 |
+
"cv": 0.3079640097284264,
|
| 423 |
+
"name": "Encoder block 3",
|
| 424 |
+
"dim": 16
|
| 425 |
+
},
|
| 426 |
+
"svd_S_orig": {
|
| 427 |
+
"cv": 0.46774789196800154,
|
| 428 |
+
"name": "SVD S (pre cross-attn)",
|
| 429 |
+
"dim": 4
|
| 430 |
+
},
|
| 431 |
+
"svd_S": {
|
| 432 |
+
"cv": 0.46958339733251925,
|
| 433 |
+
"name": "SVD S (post cross-attn)",
|
| 434 |
+
"dim": 4
|
| 435 |
+
},
|
| 436 |
+
"dec_in": {
|
| 437 |
+
"cv": 0.31508316315948587,
|
| 438 |
+
"name": "Decoder input projection",
|
| 439 |
+
"dim": 16
|
| 440 |
+
},
|
| 441 |
+
"dec_block_0": {
|
| 442 |
+
"cv": 0.375718443993368,
|
| 443 |
+
"name": "Decoder block 0",
|
| 444 |
+
"dim": 16
|
| 445 |
+
},
|
| 446 |
+
"dec_block_3": {
|
| 447 |
+
"cv": 0.46716902590523746,
|
| 448 |
+
"name": "Decoder block 3",
|
| 449 |
+
"dim": 16
|
| 450 |
+
}
|
| 451 |
+
},
|
| 452 |
+
"noise_fingerprints": {
|
| 453 |
+
"gaussian": {
|
| 454 |
+
"S_mean": [
|
| 455 |
+
4.551583290100098,
|
| 456 |
+
3.9600980281829834,
|
| 457 |
+
2.967662811279297,
|
| 458 |
+
1.9223911762237549
|
| 459 |
+
],
|
| 460 |
+
"S_std": [
|
| 461 |
+
0.034768905490636826,
|
| 462 |
+
0.035847216844558716,
|
| 463 |
+
0.04485037177801132,
|
| 464 |
+
0.04318368062376976
|
| 465 |
+
],
|
| 466 |
+
"erank": 1.1229482889175415,
|
| 467 |
+
"enc_final_abs_mean": 0.1409558653831482,
|
| 468 |
+
"enc_final_dead_frac": 5.086263172415784e-06,
|
| 469 |
+
"cross_attn_delta": 0.054473843425512314,
|
| 470 |
+
"recon_mse": 2.188036887673661e-06
|
| 471 |
+
},
|
| 472 |
+
"uniform": {
|
| 473 |
+
"S_mean": [
|
| 474 |
+
4.549080848693848,
|
| 475 |
+
4.003808975219727,
|
| 476 |
+
2.929938316345215,
|
| 477 |
+
1.891095519065857
|
| 478 |
+
],
|
| 479 |
+
"S_std": [
|
| 480 |
+
0.026261691004037857,
|
| 481 |
+
0.026727596297860146,
|
| 482 |
+
0.03346379101276398,
|
| 483 |
+
0.0348832905292511
|
| 484 |
+
],
|
| 485 |
+
"erank": 1.0971537828445435,
|
| 486 |
+
"enc_final_abs_mean": 0.11894993484020233,
|
| 487 |
+
"enc_final_dead_frac": 5.086263172415784e-06,
|
| 488 |
+
"cross_attn_delta": 0.05392440780997276,
|
| 489 |
+
"recon_mse": 8.884973681233532e-07
|
| 490 |
+
},
|
| 491 |
+
"pink": {
|
| 492 |
+
"S_mean": [
|
| 493 |
+
4.485807418823242,
|
| 494 |
+
3.9797024726867676,
|
| 495 |
+
2.974792003631592,
|
| 496 |
+
2.0227270126342773
|
| 497 |
+
],
|
| 498 |
+
"S_std": [
|
| 499 |
+
0.060695406049489975,
|
| 500 |
+
0.06107177957892418,
|
| 501 |
+
0.06326989084482193,
|
| 502 |
+
0.19789165258407593
|
| 503 |
+
],
|
| 504 |
+
"erank": 1.2274385690689087,
|
| 505 |
+
"enc_final_abs_mean": 0.11895599216222763,
|
| 506 |
+
"enc_final_dead_frac": 0.0,
|
| 507 |
+
"cross_attn_delta": 0.05545634776353836,
|
| 508 |
+
"recon_mse": 3.906632173311664e-07
|
| 509 |
+
},
|
| 510 |
+
"salt_pepper": {
|
| 511 |
+
"S_mean": [
|
| 512 |
+
4.531198024749756,
|
| 513 |
+
3.8916804790496826,
|
| 514 |
+
3.0370736122131348,
|
| 515 |
+
2.0111701488494873
|
| 516 |
+
],
|
| 517 |
+
"S_std": [
|
| 518 |
+
0.04507261514663696,
|
| 519 |
+
0.049419280141592026,
|
| 520 |
+
0.060413066297769547,
|
| 521 |
+
0.054792631417512894
|
| 522 |
+
],
|
| 523 |
+
"erank": 1.156660556793213,
|
| 524 |
+
"enc_final_abs_mean": 0.18618455529212952,
|
| 525 |
+
"enc_final_dead_frac": 2.543131586207892e-06,
|
| 526 |
+
"cross_attn_delta": 0.055769048631191254,
|
| 527 |
+
"recon_mse": 2.9938106308691204e-05
|
| 528 |
+
},
|
| 529 |
+
"cauchy": {
|
| 530 |
+
"S_mean": [
|
| 531 |
+
4.539364337921143,
|
| 532 |
+
3.90649676322937,
|
| 533 |
+
3.0197839736938477,
|
| 534 |
+
1.9870045185089111
|
| 535 |
+
],
|
| 536 |
+
"S_std": [
|
| 537 |
+
0.04201872646808624,
|
| 538 |
+
0.04622576758265495,
|
| 539 |
+
0.055099062621593475,
|
| 540 |
+
0.052384763956069946
|
| 541 |
+
],
|
| 542 |
+
"erank": 1.1480085849761963,
|
| 543 |
+
"enc_final_abs_mean": 0.1741541475057602,
|
| 544 |
+
"enc_final_dead_frac": 3.1789145396032836e-06,
|
| 545 |
+
"cross_attn_delta": 0.055427126586437225,
|
| 546 |
+
"recon_mse": 1.4438321159104817e-05
|
| 547 |
+
}
|
| 548 |
+
},
|
| 549 |
+
"weights": {
|
| 550 |
+
"enc_in.weight": {
|
| 551 |
+
"shape": [
|
| 552 |
+
384,
|
| 553 |
+
48
|
| 554 |
+
],
|
| 555 |
+
"norm": 4.736795425415039,
|
| 556 |
+
"mean": -3.101239417446777e-05,
|
| 557 |
+
"std": 0.03489072620868683,
|
| 558 |
+
"abs_mean": 0.02599160559475422,
|
| 559 |
+
"sparsity": 0.0,
|
| 560 |
+
"condition": 1.6303043365478516,
|
| 561 |
+
"erank": 47.647403717041016
|
| 562 |
+
},
|
| 563 |
+
"enc_in.bias": {
|
| 564 |
+
"shape": [
|
| 565 |
+
384
|
| 566 |
+
],
|
| 567 |
+
"norm": 3.09242582321167,
|
| 568 |
+
"mean": 0.11399959027767181,
|
| 569 |
+
"std": 0.10926610231399536,
|
| 570 |
+
"abs_mean": 0.12932059168815613,
|
| 571 |
+
"sparsity": 0.0
|
| 572 |
+
},
|
| 573 |
+
"enc_blocks.0.0.weight": {
|
| 574 |
+
"shape": [
|
| 575 |
+
384
|
| 576 |
+
],
|
| 577 |
+
"norm": 12.654324531555176,
|
| 578 |
+
"mean": 0.6359113454818726,
|
| 579 |
+
"std": 0.1125166043639183,
|
| 580 |
+
"abs_mean": 0.6359113454818726,
|
| 581 |
+
"sparsity": 0.0
|
| 582 |
+
},
|
| 583 |
+
"enc_blocks.0.0.bias": {
|
| 584 |
+
"shape": [
|
| 585 |
+
384
|
| 586 |
+
],
|
| 587 |
+
"norm": 1.056090235710144,
|
| 588 |
+
"mean": 0.0009981549810618162,
|
| 589 |
+
"std": 0.053954433649778366,
|
| 590 |
+
"abs_mean": 0.04389415308833122,
|
| 591 |
+
"sparsity": 0.0
|
| 592 |
+
},
|
| 593 |
+
"enc_blocks.0.1.weight": {
|
| 594 |
+
"shape": [
|
| 595 |
+
384,
|
| 596 |
+
384
|
| 597 |
+
],
|
| 598 |
+
"norm": 7.409005165100098,
|
| 599 |
+
"mean": -0.0002259344037156552,
|
| 600 |
+
"std": 0.019293026998639107,
|
| 601 |
+
"abs_mean": 0.0151827996596694,
|
| 602 |
+
"sparsity": 5.425347262644209e-05,
|
| 603 |
+
"condition": 5300.60693359375,
|
| 604 |
+
"erank": 257.93896484375
|
| 605 |
+
},
|
| 606 |
+
"enc_blocks.0.1.bias": {
|
| 607 |
+
"shape": [
|
| 608 |
+
384
|
| 609 |
+
],
|
| 610 |
+
"norm": 0.9071546792984009,
|
| 611 |
+
"mean": 0.02974475547671318,
|
| 612 |
+
"std": 0.03551873937249184,
|
| 613 |
+
"abs_mean": 0.03694774955511093,
|
| 614 |
+
"sparsity": 0.0
|
| 615 |
+
},
|
| 616 |
+
"enc_blocks.0.3.weight": {
|
| 617 |
+
"shape": [
|
| 618 |
+
384,
|
| 619 |
+
384
|
| 620 |
+
],
|
| 621 |
+
"norm": 13.219942092895508,
|
| 622 |
+
"mean": 0.0017413782188668847,
|
| 623 |
+
"std": 0.034382980316877365,
|
| 624 |
+
"abs_mean": 0.028007401153445244,
|
| 625 |
+
"sparsity": 2.7126736313221045e-05,
|
| 626 |
+
"condition": 9943.1591796875,
|
| 627 |
+
"erank": 285.1506042480469
|
| 628 |
+
},
|
| 629 |
+
"enc_blocks.0.3.bias": {
|
| 630 |
+
"shape": [
|
| 631 |
+
384
|
| 632 |
+
],
|
| 633 |
+
"norm": 0.8092156052589417,
|
| 634 |
+
"mean": -0.009931594133377075,
|
| 635 |
+
"std": 0.040135327726602554,
|
| 636 |
+
"abs_mean": 0.032900743186473846,
|
| 637 |
+
"sparsity": 0.0
|
| 638 |
+
},
|
| 639 |
+
"enc_blocks.1.0.weight": {
|
| 640 |
+
"shape": [
|
| 641 |
+
384
|
| 642 |
+
],
|
| 643 |
+
"norm": 16.65181541442871,
|
| 644 |
+
"mean": 0.8478428721427917,
|
| 645 |
+
"std": 0.057114146649837494,
|
| 646 |
+
"abs_mean": 0.8478428721427917,
|
| 647 |
+
"sparsity": 0.0
|
| 648 |
+
},
|
| 649 |
+
"enc_blocks.1.0.bias": {
|
| 650 |
+
"shape": [
|
| 651 |
+
384
|
| 652 |
+
],
|
| 653 |
+
"norm": 0.6272527575492859,
|
| 654 |
+
"mean": -5.119678098708391e-05,
|
| 655 |
+
"std": 0.03205107897520065,
|
| 656 |
+
"abs_mean": 0.024098535999655724,
|
| 657 |
+
"sparsity": 0.0
|
| 658 |
+
},
|
| 659 |
+
"enc_blocks.1.1.weight": {
|
| 660 |
+
"shape": [
|
| 661 |
+
384,
|
| 662 |
+
384
|
| 663 |
+
],
|
| 664 |
+
"norm": 11.865377426147461,
|
| 665 |
+
"mean": 0.0005295192240737379,
|
| 666 |
+
"std": 0.030894985422492027,
|
| 667 |
+
"abs_mean": 0.02438214048743248,
|
| 668 |
+
"sparsity": 4.069010537932627e-05,
|
| 669 |
+
"condition": 3584.115478515625,
|
| 670 |
+
"erank": 272.6521301269531
|
| 671 |
+
},
|
| 672 |
+
"enc_blocks.1.1.bias": {
|
| 673 |
+
"shape": [
|
| 674 |
+
384
|
| 675 |
+
],
|
| 676 |
+
"norm": 0.6861883997917175,
|
| 677 |
+
"mean": 0.006015029735863209,
|
| 678 |
+
"std": 0.03454142436385155,
|
| 679 |
+
"abs_mean": 0.029296232387423515,
|
| 680 |
+
"sparsity": 0.0
|
| 681 |
+
},
|
| 682 |
+
"enc_blocks.1.3.weight": {
|
| 683 |
+
"shape": [
|
| 684 |
+
384,
|
| 685 |
+
384
|
| 686 |
+
],
|
| 687 |
+
"norm": 13.247173309326172,
|
| 688 |
+
"mean": 0.0008856668719090521,
|
| 689 |
+
"std": 0.034486591815948486,
|
| 690 |
+
"abs_mean": 0.027842184528708458,
|
| 691 |
+
"sparsity": 2.7126736313221045e-05,
|
| 692 |
+
"condition": 7903.28125,
|
| 693 |
+
"erank": 281.22845458984375
|
| 694 |
+
},
|
| 695 |
+
"enc_blocks.1.3.bias": {
|
| 696 |
+
"shape": [
|
| 697 |
+
384
|
| 698 |
+
],
|
| 699 |
+
"norm": 0.7264137268066406,
|
| 700 |
+
"mean": -0.010470348410308361,
|
| 701 |
+
"std": 0.03560663014650345,
|
| 702 |
+
"abs_mean": 0.029817825183272362,
|
| 703 |
+
"sparsity": 0.0
|
| 704 |
+
},
|
| 705 |
+
"enc_blocks.2.0.weight": {
|
| 706 |
+
"shape": [
|
| 707 |
+
384
|
| 708 |
+
],
|
| 709 |
+
"norm": 17.554519653320312,
|
| 710 |
+
"mean": 0.8946146965026855,
|
| 711 |
+
"std": 0.046616703271865845,
|
| 712 |
+
"abs_mean": 0.8946146965026855,
|
| 713 |
+
"sparsity": 0.0
|
| 714 |
+
},
|
| 715 |
+
"enc_blocks.2.0.bias": {
|
| 716 |
+
"shape": [
|
| 717 |
+
384
|
| 718 |
+
],
|
| 719 |
+
"norm": 0.638079822063446,
|
| 720 |
+
"mean": -0.0017956134397536516,
|
| 721 |
+
"std": 0.032554741948843,
|
| 722 |
+
"abs_mean": 0.024456802755594254,
|
| 723 |
+
"sparsity": 0.0
|
| 724 |
+
},
|
| 725 |
+
"enc_blocks.2.1.weight": {
|
| 726 |
+
"shape": [
|
| 727 |
+
384,
|
| 728 |
+
384
|
| 729 |
+
],
|
| 730 |
+
"norm": 12.878968238830566,
|
| 731 |
+
"mean": 0.0004422464990057051,
|
| 732 |
+
"std": 0.03353617712855339,
|
| 733 |
+
"abs_mean": 0.02666916884481907,
|
| 734 |
+
"sparsity": 6.781684078305261e-06,
|
| 735 |
+
"condition": 1223.2880859375,
|
| 736 |
+
"erank": 267.52203369140625
|
| 737 |
+
},
|
| 738 |
+
"enc_blocks.2.1.bias": {
|
| 739 |
+
"shape": [
|
| 740 |
+
384
|
| 741 |
+
],
|
| 742 |
+
"norm": 0.6446398496627808,
|
| 743 |
+
"mean": 0.0021967978682368994,
|
| 744 |
+
"std": 0.03286603093147278,
|
| 745 |
+
"abs_mean": 0.02753588557243347,
|
| 746 |
+
"sparsity": 0.0
|
| 747 |
+
},
|
| 748 |
+
"enc_blocks.2.3.weight": {
|
| 749 |
+
"shape": [
|
| 750 |
+
384,
|
| 751 |
+
384
|
| 752 |
+
],
|
| 753 |
+
"norm": 12.749981880187988,
|
| 754 |
+
"mean": 0.0010546775301918387,
|
| 755 |
+
"std": 0.03318643569946289,
|
| 756 |
+
"abs_mean": 0.02681606449186802,
|
| 757 |
+
"sparsity": 4.069010537932627e-05,
|
| 758 |
+
"condition": 11286.2783203125,
|
| 759 |
+
"erank": 287.71466064453125
|
| 760 |
+
},
|
| 761 |
+
"enc_blocks.2.3.bias": {
|
| 762 |
+
"shape": [
|
| 763 |
+
384
|
| 764 |
+
],
|
| 765 |
+
"norm": 0.7675585746765137,
|
| 766 |
+
"mean": -0.011421765200793743,
|
| 767 |
+
"std": 0.03751590475440025,
|
| 768 |
+
"abs_mean": 0.03255150094628334,
|
| 769 |
+
"sparsity": 0.0
|
| 770 |
+
},
|
| 771 |
+
"enc_blocks.3.0.weight": {
|
| 772 |
+
"shape": [
|
| 773 |
+
384
|
| 774 |
+
],
|
| 775 |
+
"norm": 17.64955711364746,
|
| 776 |
+
"mean": 0.8998109698295593,
|
| 777 |
+
"std": 0.039498452097177505,
|
| 778 |
+
"abs_mean": 0.8998109698295593,
|
| 779 |
+
"sparsity": 0.0
|
| 780 |
+
},
|
| 781 |
+
"enc_blocks.3.0.bias": {
|
| 782 |
+
"shape": [
|
| 783 |
+
384
|
| 784 |
+
],
|
| 785 |
+
"norm": 0.632440984249115,
|
| 786 |
+
"mean": -0.001403431873768568,
|
| 787 |
+
"std": 0.0322856567800045,
|
| 788 |
+
"abs_mean": 0.02486751601099968,
|
| 789 |
+
"sparsity": 0.0
|
| 790 |
+
},
|
| 791 |
+
"enc_blocks.3.1.weight": {
|
| 792 |
+
"shape": [
|
| 793 |
+
384,
|
| 794 |
+
384
|
| 795 |
+
],
|
| 796 |
+
"norm": 12.882952690124512,
|
| 797 |
+
"mean": 0.0005309268599376082,
|
| 798 |
+
"std": 0.03354526683688164,
|
| 799 |
+
"abs_mean": 0.026780350133776665,
|
| 800 |
+
"sparsity": 2.0345052689663135e-05,
|
| 801 |
+
"condition": 3306.011474609375,
|
| 802 |
+
"erank": 277.59588623046875
|
| 803 |
+
},
|
| 804 |
+
"enc_blocks.3.1.bias": {
|
| 805 |
+
"shape": [
|
| 806 |
+
384
|
| 807 |
+
],
|
| 808 |
+
"norm": 0.6859570741653442,
|
| 809 |
+
"mean": -9.370347470394336e-06,
|
| 810 |
+
"std": 0.03505076840519905,
|
| 811 |
+
"abs_mean": 0.0292180385440588,
|
| 812 |
+
"sparsity": 0.0
|
| 813 |
+
},
|
| 814 |
+
"enc_blocks.3.3.weight": {
|
| 815 |
+
"shape": [
|
| 816 |
+
384,
|
| 817 |
+
384
|
| 818 |
+
],
|
| 819 |
+
"norm": 12.485105514526367,
|
| 820 |
+
"mean": 0.0012131760595366359,
|
| 821 |
+
"std": 0.03249076381325722,
|
| 822 |
+
"abs_mean": 0.02636081352829933,
|
| 823 |
+
"sparsity": 2.7126736313221045e-05,
|
| 824 |
+
"condition": 6738.20654296875,
|
| 825 |
+
"erank": 288.5359802246094
|
| 826 |
+
},
|
| 827 |
+
"enc_blocks.3.3.bias": {
|
| 828 |
+
"shape": [
|
| 829 |
+
384
|
| 830 |
+
],
|
| 831 |
+
"norm": 0.7744740843772888,
|
| 832 |
+
"mean": -0.014685760252177715,
|
| 833 |
+
"std": 0.036740291863679886,
|
| 834 |
+
"abs_mean": 0.03180073946714401,
|
| 835 |
+
"sparsity": 0.0
|
| 836 |
+
},
|
| 837 |
+
"enc_out.weight": {
|
| 838 |
+
"shape": [
|
| 839 |
+
192,
|
| 840 |
+
384
|
| 841 |
+
],
|
| 842 |
+
"norm": 15.931266784667969,
|
| 843 |
+
"mean": 0.00025461785844527185,
|
| 844 |
+
"std": 0.05867227911949158,
|
| 845 |
+
"abs_mean": 0.04659826308488846,
|
| 846 |
+
"sparsity": 1.3563368156610522e-05,
|
| 847 |
+
"condition": 12.063321113586426,
|
| 848 |
+
"erank": 180.5324249267578
|
| 849 |
+
},
|
| 850 |
+
"enc_out.bias": {
|
| 851 |
+
"shape": [
|
| 852 |
+
192
|
| 853 |
+
],
|
| 854 |
+
"norm": 0.7678680419921875,
|
| 855 |
+
"mean": -0.0035826812963932753,
|
| 856 |
+
"std": 0.05544474720954895,
|
| 857 |
+
"abs_mean": 0.04248465597629547,
|
| 858 |
+
"sparsity": 0.0
|
| 859 |
+
},
|
| 860 |
+
"dec_in.weight": {
|
| 861 |
+
"shape": [
|
| 862 |
+
384,
|
| 863 |
+
192
|
| 864 |
+
],
|
| 865 |
+
"norm": 17.036605834960938,
|
| 866 |
+
"mean": -0.00020357051107566804,
|
| 867 |
+
"std": 0.06274331361055374,
|
| 868 |
+
"abs_mean": 0.050236720591783524,
|
| 869 |
+
"sparsity": 0.0,
|
| 870 |
+
"condition": 11.275330543518066,
|
| 871 |
+
"erank": 160.26626586914062
|
| 872 |
+
},
|
| 873 |
+
"dec_in.bias": {
|
| 874 |
+
"shape": [
|
| 875 |
+
384
|
| 876 |
+
],
|
| 877 |
+
"norm": 0.838572084903717,
|
| 878 |
+
"mean": 0.006184441968798637,
|
| 879 |
+
"std": 0.04239920154213905,
|
| 880 |
+
"abs_mean": 0.03614982217550278,
|
| 881 |
+
"sparsity": 0.0
|
| 882 |
+
},
|
| 883 |
+
"dec_blocks.0.0.weight": {
|
| 884 |
+
"shape": [
|
| 885 |
+
384
|
| 886 |
+
],
|
| 887 |
+
"norm": 17.526269912719727,
|
| 888 |
+
"mean": 0.8938546180725098,
|
| 889 |
+
"std": 0.03080192767083645,
|
| 890 |
+
"abs_mean": 0.8938546180725098,
|
| 891 |
+
"sparsity": 0.0
|
| 892 |
+
},
|
| 893 |
+
"dec_blocks.0.0.bias": {
|
| 894 |
+
"shape": [
|
| 895 |
+
384
|
| 896 |
+
],
|
| 897 |
+
"norm": 0.6757148504257202,
|
| 898 |
+
"mean": 0.02418457716703415,
|
| 899 |
+
"std": 0.02461141161620617,
|
| 900 |
+
"abs_mean": 0.027816174551844597,
|
| 901 |
+
"sparsity": 0.0
|
| 902 |
+
},
|
| 903 |
+
"dec_blocks.0.1.weight": {
|
| 904 |
+
"shape": [
|
| 905 |
+
384,
|
| 906 |
+
384
|
| 907 |
+
],
|
| 908 |
+
"norm": 11.316617012023926,
|
| 909 |
+
"mean": 0.001807661959901452,
|
| 910 |
+
"std": 0.029414964839816093,
|
| 911 |
+
"abs_mean": 0.023787135258316994,
|
| 912 |
+
"sparsity": 2.7126736313221045e-05,
|
| 913 |
+
"condition": 671915.0625,
|
| 914 |
+
"erank": 289.99749755859375
|
| 915 |
+
},
|
| 916 |
+
"dec_blocks.0.1.bias": {
|
| 917 |
+
"shape": [
|
| 918 |
+
384
|
| 919 |
+
],
|
| 920 |
+
"norm": 0.58186274766922,
|
| 921 |
+
"mean": 0.005265340209007263,
|
| 922 |
+
"std": 0.02926061488687992,
|
| 923 |
+
"abs_mean": 0.024605482816696167,
|
| 924 |
+
"sparsity": 0.0
|
| 925 |
+
},
|
| 926 |
+
"dec_blocks.0.3.weight": {
|
| 927 |
+
"shape": [
|
| 928 |
+
384,
|
| 929 |
+
384
|
| 930 |
+
],
|
| 931 |
+
"norm": 15.586650848388672,
|
| 932 |
+
"mean": -0.0036952048540115356,
|
| 933 |
+
"std": 0.040421824902296066,
|
| 934 |
+
"abs_mean": 0.03299005329608917,
|
| 935 |
+
"sparsity": 0.0,
|
| 936 |
+
"condition": 2613.58642578125,
|
| 937 |
+
"erank": 273.7778625488281
|
| 938 |
+
},
|
| 939 |
+
"dec_blocks.0.3.bias": {
|
| 940 |
+
"shape": [
|
| 941 |
+
384
|
| 942 |
+
],
|
| 943 |
+
"norm": 0.60248863697052,
|
| 944 |
+
"mean": -0.005198275670409203,
|
| 945 |
+
"std": 0.030342521145939827,
|
| 946 |
+
"abs_mean": 0.02475673519074917,
|
| 947 |
+
"sparsity": 0.0
|
| 948 |
+
},
|
| 949 |
+
"dec_blocks.1.0.weight": {
|
| 950 |
+
"shape": [
|
| 951 |
+
384
|
| 952 |
+
],
|
| 953 |
+
"norm": 17.223556518554688,
|
| 954 |
+
"mean": 0.8784023523330688,
|
| 955 |
+
"std": 0.030661659315228462,
|
| 956 |
+
"abs_mean": 0.8784023523330688,
|
| 957 |
+
"sparsity": 0.0
|
| 958 |
+
},
|
| 959 |
+
"dec_blocks.1.0.bias": {
|
| 960 |
+
"shape": [
|
| 961 |
+
384
|
| 962 |
+
],
|
| 963 |
+
"norm": 0.6037180423736572,
|
| 964 |
+
"mean": 0.01788325607776642,
|
| 965 |
+
"std": 0.02511945739388466,
|
| 966 |
+
"abs_mean": 0.02397087588906288,
|
| 967 |
+
"sparsity": 0.0
|
| 968 |
+
},
|
| 969 |
+
"dec_blocks.1.1.weight": {
|
| 970 |
+
"shape": [
|
| 971 |
+
384,
|
| 972 |
+
384
|
| 973 |
+
],
|
| 974 |
+
"norm": 12.594478607177734,
|
| 975 |
+
"mean": -0.00029621124849654734,
|
| 976 |
+
"std": 0.03279689699411392,
|
| 977 |
+
"abs_mean": 0.026519903913140297,
|
| 978 |
+
"sparsity": 2.0345052689663135e-05,
|
| 979 |
+
"condition": 1060.1834716796875,
|
| 980 |
+
"erank": 293.8656311035156
|
| 981 |
+
},
|
| 982 |
+
"dec_blocks.1.1.bias": {
|
| 983 |
+
"shape": [
|
| 984 |
+
384
|
| 985 |
+
],
|
| 986 |
+
"norm": 0.6123263239860535,
|
| 987 |
+
"mean": 0.0017201153095811605,
|
| 988 |
+
"std": 0.03124096989631653,
|
| 989 |
+
"abs_mean": 0.025975652039051056,
|
| 990 |
+
"sparsity": 0.0
|
| 991 |
+
},
|
| 992 |
+
"dec_blocks.1.3.weight": {
|
| 993 |
+
"shape": [
|
| 994 |
+
384,
|
| 995 |
+
384
|
| 996 |
+
],
|
| 997 |
+
"norm": 12.478222846984863,
|
| 998 |
+
"mean": 0.00040824111783877015,
|
| 999 |
+
"std": 0.03249291703104973,
|
| 1000 |
+
"abs_mean": 0.02622552216053009,
|
| 1001 |
+
"sparsity": 6.781684078305261e-06,
|
| 1002 |
+
"condition": 4037.880126953125,
|
| 1003 |
+
"erank": 287.6106872558594
|
| 1004 |
+
},
|
| 1005 |
+
"dec_blocks.1.3.bias": {
|
| 1006 |
+
"shape": [
|
| 1007 |
+
384
|
| 1008 |
+
],
|
| 1009 |
+
"norm": 0.5644272565841675,
|
| 1010 |
+
"mean": -0.008861962705850601,
|
| 1011 |
+
"std": 0.02744188904762268,
|
| 1012 |
+
"abs_mean": 0.023942431434988976,
|
| 1013 |
+
"sparsity": 0.0
|
| 1014 |
+
},
|
| 1015 |
+
"dec_blocks.2.0.weight": {
|
| 1016 |
+
"shape": [
|
| 1017 |
+
384
|
| 1018 |
+
],
|
| 1019 |
+
"norm": 17.322954177856445,
|
| 1020 |
+
"mean": 0.8833889961242676,
|
| 1021 |
+
"std": 0.033128079026937485,
|
| 1022 |
+
"abs_mean": 0.8833889961242676,
|
| 1023 |
+
"sparsity": 0.0
|
| 1024 |
+
},
|
| 1025 |
+
"dec_blocks.2.0.bias": {
|
| 1026 |
+
"shape": [
|
| 1027 |
+
384
|
| 1028 |
+
],
|
| 1029 |
+
"norm": 0.6046607494354248,
|
| 1030 |
+
"mean": 0.01801062747836113,
|
| 1031 |
+
"std": 0.025087399408221245,
|
| 1032 |
+
"abs_mean": 0.02382083795964718,
|
| 1033 |
+
"sparsity": 0.0
|
| 1034 |
+
},
|
| 1035 |
+
"dec_blocks.2.1.weight": {
|
| 1036 |
+
"shape": [
|
| 1037 |
+
384,
|
| 1038 |
+
384
|
| 1039 |
+
],
|
| 1040 |
+
"norm": 13.23556900024414,
|
| 1041 |
+
"mean": -0.00037671197787858546,
|
| 1042 |
+
"std": 0.03446568548679352,
|
| 1043 |
+
"abs_mean": 0.027716848999261856,
|
| 1044 |
+
"sparsity": 2.0345052689663135e-05,
|
| 1045 |
+
"condition": 23743.208984375,
|
| 1046 |
+
"erank": 286.2240905761719
|
| 1047 |
+
},
|
| 1048 |
+
"dec_blocks.2.1.bias": {
|
| 1049 |
+
"shape": [
|
| 1050 |
+
384
|
| 1051 |
+
],
|
| 1052 |
+
"norm": 0.6294949650764465,
|
| 1053 |
+
"mean": 0.00018974603153765202,
|
| 1054 |
+
"std": 0.03216513246297836,
|
| 1055 |
+
"abs_mean": 0.026205262169241905,
|
| 1056 |
+
"sparsity": 0.0
|
| 1057 |
+
},
|
| 1058 |
+
"dec_blocks.2.3.weight": {
|
| 1059 |
+
"shape": [
|
| 1060 |
+
384,
|
| 1061 |
+
384
|
| 1062 |
+
],
|
| 1063 |
+
"norm": 12.366257667541504,
|
| 1064 |
+
"mean": 0.0003269152366556227,
|
| 1065 |
+
"std": 0.03220224380493164,
|
| 1066 |
+
"abs_mean": 0.026165293529629707,
|
| 1067 |
+
"sparsity": 4.747178900288418e-05,
|
| 1068 |
+
"condition": 9688.5419921875,
|
| 1069 |
+
"erank": 295.5987548828125
|
| 1070 |
+
},
|
| 1071 |
+
"dec_blocks.2.3.bias": {
|
| 1072 |
+
"shape": [
|
| 1073 |
+
384
|
| 1074 |
+
],
|
| 1075 |
+
"norm": 0.5544984340667725,
|
| 1076 |
+
"mean": -0.007604988291859627,
|
| 1077 |
+
"std": 0.027291078120470047,
|
| 1078 |
+
"abs_mean": 0.023679127916693687,
|
| 1079 |
+
"sparsity": 0.0
|
| 1080 |
+
},
|
| 1081 |
+
"dec_blocks.3.0.weight": {
|
| 1082 |
+
"shape": [
|
| 1083 |
+
384
|
| 1084 |
+
],
|
| 1085 |
+
"norm": 17.596338272094727,
|
| 1086 |
+
"mean": 0.896987795829773,
|
| 1087 |
+
"std": 0.0418156273663044,
|
| 1088 |
+
"abs_mean": 0.896987795829773,
|
| 1089 |
+
"sparsity": 0.0
|
| 1090 |
+
},
|
| 1091 |
+
"dec_blocks.3.0.bias": {
|
| 1092 |
+
"shape": [
|
| 1093 |
+
384
|
| 1094 |
+
],
|
| 1095 |
+
"norm": 0.576328456401825,
|
| 1096 |
+
"mean": 0.015175910666584969,
|
| 1097 |
+
"std": 0.025225672870874405,
|
| 1098 |
+
"abs_mean": 0.021809693425893784,
|
| 1099 |
+
"sparsity": 0.0
|
| 1100 |
+
},
|
| 1101 |
+
"dec_blocks.3.1.weight": {
|
| 1102 |
+
"shape": [
|
| 1103 |
+
384,
|
| 1104 |
+
384
|
| 1105 |
+
],
|
| 1106 |
+
"norm": 13.386054039001465,
|
| 1107 |
+
"mean": 7.282180740730837e-05,
|
| 1108 |
+
"std": 0.0348595567047596,
|
| 1109 |
+
"abs_mean": 0.027967920526862144,
|
| 1110 |
+
"sparsity": 4.069010537932627e-05,
|
| 1111 |
+
"condition": 71744.9921875,
|
| 1112 |
+
"erank": 278.9849548339844
|
| 1113 |
+
},
|
| 1114 |
+
"dec_blocks.3.1.bias": {
|
| 1115 |
+
"shape": [
|
| 1116 |
+
384
|
| 1117 |
+
],
|
| 1118 |
+
"norm": 0.6271045804023743,
|
| 1119 |
+
"mean": 0.0018653826555237174,
|
| 1120 |
+
"std": 0.03198906406760216,
|
| 1121 |
+
"abs_mean": 0.02620757184922695,
|
| 1122 |
+
"sparsity": 0.0
|
| 1123 |
+
},
|
| 1124 |
+
"dec_blocks.3.3.weight": {
|
| 1125 |
+
"shape": [
|
| 1126 |
+
384,
|
| 1127 |
+
384
|
| 1128 |
+
],
|
| 1129 |
+
"norm": 12.783568382263184,
|
| 1130 |
+
"mean": 0.0009830869967117906,
|
| 1131 |
+
"std": 0.033276136964559555,
|
| 1132 |
+
"abs_mean": 0.02756834402680397,
|
| 1133 |
+
"sparsity": 2.7126736313221045e-05,
|
| 1134 |
+
"condition": 817.364013671875,
|
| 1135 |
+
"erank": 301.7668762207031
|
| 1136 |
+
},
|
| 1137 |
+
"dec_blocks.3.3.bias": {
|
| 1138 |
+
"shape": [
|
| 1139 |
+
384
|
| 1140 |
+
],
|
| 1141 |
+
"norm": 0.6337453126907349,
|
| 1142 |
+
"mean": -0.013539277017116547,
|
| 1143 |
+
"std": 0.029408499598503113,
|
| 1144 |
+
"abs_mean": 0.026898186653852463,
|
| 1145 |
+
"sparsity": 0.0
|
| 1146 |
+
},
|
| 1147 |
+
"dec_out.weight": {
|
| 1148 |
+
"shape": [
|
| 1149 |
+
48,
|
| 1150 |
+
384
|
| 1151 |
+
],
|
| 1152 |
+
"norm": 7.669042587280273,
|
| 1153 |
+
"mean": 0.0010714484378695488,
|
| 1154 |
+
"std": 0.05647920444607735,
|
| 1155 |
+
"abs_mean": 0.04642903059720993,
|
| 1156 |
+
"sparsity": 0.0,
|
| 1157 |
+
"condition": 1.529896855354309,
|
| 1158 |
+
"erank": 47.82491683959961
|
| 1159 |
+
},
|
| 1160 |
+
"dec_out.bias": {
|
| 1161 |
+
"shape": [
|
| 1162 |
+
48
|
| 1163 |
+
],
|
| 1164 |
+
"norm": 0.20354627072811127,
|
| 1165 |
+
"mean": -0.011547963134944439,
|
| 1166 |
+
"std": 0.027300545945763588,
|
| 1167 |
+
"abs_mean": 0.025418907403945923,
|
| 1168 |
+
"sparsity": 0.0
|
| 1169 |
+
},
|
| 1170 |
+
"cross_attn.0.alpha_logits": {
|
| 1171 |
+
"shape": [
|
| 1172 |
+
4
|
| 1173 |
+
],
|
| 1174 |
+
"norm": 3.9009289741516113,
|
| 1175 |
+
"mean": -1.9489340782165527,
|
| 1176 |
+
"std": 0.08920314908027649,
|
| 1177 |
+
"abs_mean": 1.9489340782165527,
|
| 1178 |
+
"sparsity": 0.0
|
| 1179 |
+
},
|
| 1180 |
+
"cross_attn.0.qkv.weight": {
|
| 1181 |
+
"shape": [
|
| 1182 |
+
12,
|
| 1183 |
+
4
|
| 1184 |
+
],
|
| 1185 |
+
"norm": 2.395477771759033,
|
| 1186 |
+
"mean": 0.004947615321725607,
|
| 1187 |
+
"std": 0.3493805527687073,
|
| 1188 |
+
"abs_mean": 0.300910621881485,
|
| 1189 |
+
"sparsity": 0.0,
|
| 1190 |
+
"condition": 1.6833115816116333,
|
| 1191 |
+
"erank": 3.929398536682129
|
| 1192 |
+
},
|
| 1193 |
+
"cross_attn.0.qkv.bias": {
|
| 1194 |
+
"shape": [
|
| 1195 |
+
12
|
| 1196 |
+
],
|
| 1197 |
+
"norm": 1.140745997428894,
|
| 1198 |
+
"mean": -0.15099439024925232,
|
| 1199 |
+
"std": 0.305660218000412,
|
| 1200 |
+
"abs_mean": 0.2782840132713318,
|
| 1201 |
+
"sparsity": 0.0
|
| 1202 |
+
},
|
| 1203 |
+
"cross_attn.0.out_proj.weight": {
|
| 1204 |
+
"shape": [
|
| 1205 |
+
4,
|
| 1206 |
+
4
|
| 1207 |
+
],
|
| 1208 |
+
"norm": 1.3285354375839233,
|
| 1209 |
+
"mean": 0.05088873207569122,
|
| 1210 |
+
"std": 0.338976114988327,
|
| 1211 |
+
"abs_mean": 0.27125734090805054,
|
| 1212 |
+
"sparsity": 0.0,
|
| 1213 |
+
"condition": 64.55738830566406,
|
| 1214 |
+
"erank": 2.6695845127105713
|
| 1215 |
+
},
|
| 1216 |
+
"cross_attn.0.out_proj.bias": {
|
| 1217 |
+
"shape": [
|
| 1218 |
+
4
|
| 1219 |
+
],
|
| 1220 |
+
"norm": 0.38879814743995667,
|
| 1221 |
+
"mean": 0.053185347467660904,
|
| 1222 |
+
"std": 0.21590836346149445,
|
| 1223 |
+
"abs_mean": 0.1815299093723297,
|
| 1224 |
+
"sparsity": 0.0
|
| 1225 |
+
},
|
| 1226 |
+
"cross_attn.0.norm.weight": {
|
| 1227 |
+
"shape": [
|
| 1228 |
+
4
|
| 1229 |
+
],
|
| 1230 |
+
"norm": 2.1059398651123047,
|
| 1231 |
+
"mean": 1.0422992706298828,
|
| 1232 |
+
"std": 0.1726568192243576,
|
| 1233 |
+
"abs_mean": 1.0422992706298828,
|
| 1234 |
+
"sparsity": 0.0
|
| 1235 |
+
},
|
| 1236 |
+
"cross_attn.0.norm.bias": {
|
| 1237 |
+
"shape": [
|
| 1238 |
+
4
|
| 1239 |
+
],
|
| 1240 |
+
"norm": 0.35794734954833984,
|
| 1241 |
+
"mean": -0.07177595794200897,
|
| 1242 |
+
"std": 0.18931381404399872,
|
| 1243 |
+
"abs_mean": 0.16136272251605988,
|
| 1244 |
+
"sparsity": 0.0
|
| 1245 |
+
},
|
| 1246 |
+
"cross_attn.1.alpha_logits": {
|
| 1247 |
+
"shape": [
|
| 1248 |
+
4
|
| 1249 |
+
],
|
| 1250 |
+
"norm": 3.9756813049316406,
|
| 1251 |
+
"mean": -1.9850671291351318,
|
| 1252 |
+
"std": 0.1212119609117508,
|
| 1253 |
+
"abs_mean": 1.9850671291351318,
|
| 1254 |
+
"sparsity": 0.0
|
| 1255 |
+
},
|
| 1256 |
+
"cross_attn.1.qkv.weight": {
|
| 1257 |
+
"shape": [
|
| 1258 |
+
12,
|
| 1259 |
+
4
|
| 1260 |
+
],
|
| 1261 |
+
"norm": 2.1153385639190674,
|
| 1262 |
+
"mean": 0.03604736924171448,
|
| 1263 |
+
"std": 0.30639585852622986,
|
| 1264 |
+
"abs_mean": 0.23021213710308075,
|
| 1265 |
+
"sparsity": 0.0,
|
| 1266 |
+
"condition": 2.0000216960906982,
|
| 1267 |
+
"erank": 3.8778748512268066
|
| 1268 |
+
},
|
| 1269 |
+
"cross_attn.1.qkv.bias": {
|
| 1270 |
+
"shape": [
|
| 1271 |
+
12
|
| 1272 |
+
],
|
| 1273 |
+
"norm": 1.3504735231399536,
|
| 1274 |
+
"mean": 0.04980164021253586,
|
| 1275 |
+
"std": 0.40384697914123535,
|
| 1276 |
+
"abs_mean": 0.31919822096824646,
|
| 1277 |
+
"sparsity": 0.0
|
| 1278 |
+
},
|
| 1279 |
+
"cross_attn.1.out_proj.weight": {
|
| 1280 |
+
"shape": [
|
| 1281 |
+
4,
|
| 1282 |
+
4
|
| 1283 |
+
],
|
| 1284 |
+
"norm": 1.544046401977539,
|
| 1285 |
+
"mean": -0.02773943543434143,
|
| 1286 |
+
"std": 0.3976403474807739,
|
| 1287 |
+
"abs_mean": 0.3338635563850403,
|
| 1288 |
+
"sparsity": 0.0,
|
| 1289 |
+
"condition": 6.554041385650635,
|
| 1290 |
+
"erank": 3.265773296356201
|
| 1291 |
+
},
|
| 1292 |
+
"cross_attn.1.out_proj.bias": {
|
| 1293 |
+
"shape": [
|
| 1294 |
+
4
|
| 1295 |
+
],
|
| 1296 |
+
"norm": 0.5558180809020996,
|
| 1297 |
+
"mean": -0.07216920703649521,
|
| 1298 |
+
"std": 0.30989256501197815,
|
| 1299 |
+
"abs_mean": 0.25283655524253845,
|
| 1300 |
+
"sparsity": 0.0
|
| 1301 |
+
},
|
| 1302 |
+
"cross_attn.1.norm.weight": {
|
| 1303 |
+
"shape": [
|
| 1304 |
+
4
|
| 1305 |
+
],
|
| 1306 |
+
"norm": 1.9561911821365356,
|
| 1307 |
+
"mean": 0.9712949395179749,
|
| 1308 |
+
"std": 0.13295160233974457,
|
| 1309 |
+
"abs_mean": 0.9712949395179749,
|
| 1310 |
+
"sparsity": 0.0
|
| 1311 |
+
},
|
| 1312 |
+
"cross_attn.1.norm.bias": {
|
| 1313 |
+
"shape": [
|
| 1314 |
+
4
|
| 1315 |
+
],
|
| 1316 |
+
"norm": 0.2207387089729309,
|
| 1317 |
+
"mean": -0.0339832603931427,
|
| 1318 |
+
"std": 0.12125197798013687,
|
| 1319 |
+
"abs_mean": 0.09095096588134766,
|
| 1320 |
+
"sparsity": 0.0
|
| 1321 |
+
},
|
| 1322 |
+
"boundary_smooth.net.0.weight": {
|
| 1323 |
+
"shape": [
|
| 1324 |
+
8,
|
| 1325 |
+
3,
|
| 1326 |
+
3,
|
| 1327 |
+
3
|
| 1328 |
+
],
|
| 1329 |
+
"norm": 0.048899564892053604,
|
| 1330 |
+
"mean": -3.9799830119591206e-05,
|
| 1331 |
+
"std": 0.0033346840646117926,
|
| 1332 |
+
"abs_mean": 0.0023165682796388865,
|
| 1333 |
+
"sparsity": 0.0
|
| 1334 |
+
},
|
| 1335 |
+
"boundary_smooth.net.0.bias": {
|
| 1336 |
+
"shape": [
|
| 1337 |
+
8
|
| 1338 |
+
],
|
| 1339 |
+
"norm": 0.0003608279803302139,
|
| 1340 |
+
"mean": -3.4329681511735544e-05,
|
| 1341 |
+
"std": 0.0001313493849011138,
|
| 1342 |
+
"abs_mean": 0.0001100103254429996,
|
| 1343 |
+
"sparsity": 0.0
|
| 1344 |
+
},
|
| 1345 |
+
"boundary_smooth.net.2.weight": {
|
| 1346 |
+
"shape": [
|
| 1347 |
+
3,
|
| 1348 |
+
8,
|
| 1349 |
+
3,
|
| 1350 |
+
3
|
| 1351 |
+
],
|
| 1352 |
+
"norm": 0.05223501846194267,
|
| 1353 |
+
"mean": -3.519888559821993e-05,
|
| 1354 |
+
"std": 0.0035622238647192717,
|
| 1355 |
+
"abs_mean": 0.002480539958924055,
|
| 1356 |
+
"sparsity": 0.0
|
| 1357 |
+
},
|
| 1358 |
+
"boundary_smooth.net.2.bias": {
|
| 1359 |
+
"shape": [
|
| 1360 |
+
3
|
| 1361 |
+
],
|
| 1362 |
+
"norm": 0.019741201773285866,
|
| 1363 |
+
"mean": -0.009897474199533463,
|
| 1364 |
+
"std": 0.006922249216586351,
|
| 1365 |
+
"abs_mean": 0.009897474199533463,
|
| 1366 |
+
"sparsity": 0.0
|
| 1367 |
+
}
|
| 1368 |
+
}
|
| 1369 |
+
}
|