Upload folder using huggingface_hub
Browse files- .gitattributes +6 -0
- diagnostics/per_layer_iter1000.json +2603 -0
- diagnostics/per_layer_iter1500.json +2603 -0
- diagnostics/per_layer_iter2000.json +2603 -0
- diagnostics/per_layer_iter2500.json +2603 -0
- diagnostics/per_layer_iter3000.json +2603 -0
- diagnostics/per_layer_iter500.json +2603 -0
- diagnostics/per_t_loss_iter1000.json +42 -0
- diagnostics/per_t_loss_iter1500.json +42 -0
- diagnostics/per_t_loss_iter2000.json +42 -0
- diagnostics/per_t_loss_iter2500.json +42 -0
- diagnostics/per_t_loss_iter3000.json +42 -0
- diagnostics/per_t_loss_iter500.json +42 -0
- diagnostics/samples_iter1000.png +3 -0
- diagnostics/samples_iter1500.png +3 -0
- diagnostics/samples_iter2000.png +3 -0
- diagnostics/samples_iter2500.png +3 -0
- diagnostics/samples_iter3000.png +3 -0
- diagnostics/samples_iter500.png +3 -0
- diagnostics/scalars.csv +152 -0
- diagnostics/tensorboard/events.out.tfevents.1781963285.instance-2asqjmcx-main.1034.0 +3 -0
.gitattributes
CHANGED
|
@@ -33,3 +33,9 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
|
|
| 33 |
*.zip filter=lfs diff=lfs merge=lfs -text
|
| 34 |
*.zst filter=lfs diff=lfs merge=lfs -text
|
| 35 |
*tfevents* filter=lfs diff=lfs merge=lfs -text
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 33 |
*.zip filter=lfs diff=lfs merge=lfs -text
|
| 34 |
*.zst filter=lfs diff=lfs merge=lfs -text
|
| 35 |
*tfevents* filter=lfs diff=lfs merge=lfs -text
|
| 36 |
+
diagnostics/samples_iter1000.png filter=lfs diff=lfs merge=lfs -text
|
| 37 |
+
diagnostics/samples_iter1500.png filter=lfs diff=lfs merge=lfs -text
|
| 38 |
+
diagnostics/samples_iter2000.png filter=lfs diff=lfs merge=lfs -text
|
| 39 |
+
diagnostics/samples_iter2500.png filter=lfs diff=lfs merge=lfs -text
|
| 40 |
+
diagnostics/samples_iter3000.png filter=lfs diff=lfs merge=lfs -text
|
| 41 |
+
diagnostics/samples_iter500.png filter=lfs diff=lfs merge=lfs -text
|
diagnostics/per_layer_iter1000.json
ADDED
|
@@ -0,0 +1,2603 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"iter": 1000,
|
| 3 |
+
"per_layer": {
|
| 4 |
+
"time_mlp.1.weight": {
|
| 5 |
+
"w_norm": 18.69807243347168,
|
| 6 |
+
"g_norm": 0.0460389219224453,
|
| 7 |
+
"shape": [
|
| 8 |
+
1024,
|
| 9 |
+
256
|
| 10 |
+
]
|
| 11 |
+
},
|
| 12 |
+
"time_mlp.1.bias": {
|
| 13 |
+
"w_norm": 1.1588034629821777,
|
| 14 |
+
"g_norm": 0.004071285482496023,
|
| 15 |
+
"shape": [
|
| 16 |
+
1024
|
| 17 |
+
]
|
| 18 |
+
},
|
| 19 |
+
"time_mlp.3.weight": {
|
| 20 |
+
"w_norm": 9.50298023223877,
|
| 21 |
+
"g_norm": 0.018244950100779533,
|
| 22 |
+
"shape": [
|
| 23 |
+
256,
|
| 24 |
+
1024
|
| 25 |
+
]
|
| 26 |
+
},
|
| 27 |
+
"time_mlp.3.bias": {
|
| 28 |
+
"w_norm": 0.2909029722213745,
|
| 29 |
+
"g_norm": 0.005334504880011082,
|
| 30 |
+
"shape": [
|
| 31 |
+
256
|
| 32 |
+
]
|
| 33 |
+
},
|
| 34 |
+
"input_conv.weight": {
|
| 35 |
+
"w_norm": 7.937385082244873,
|
| 36 |
+
"g_norm": 0.04928673058748245,
|
| 37 |
+
"shape": [
|
| 38 |
+
192,
|
| 39 |
+
6,
|
| 40 |
+
3,
|
| 41 |
+
3
|
| 42 |
+
]
|
| 43 |
+
},
|
| 44 |
+
"input_conv.bias": {
|
| 45 |
+
"w_norm": 1.1879003047943115,
|
| 46 |
+
"g_norm": 0.06297601759433746,
|
| 47 |
+
"shape": [
|
| 48 |
+
192
|
| 49 |
+
]
|
| 50 |
+
},
|
| 51 |
+
"enc_blocks.0.norm1.weight": {
|
| 52 |
+
"w_norm": 13.822595596313477,
|
| 53 |
+
"g_norm": 0.016159730032086372,
|
| 54 |
+
"shape": [
|
| 55 |
+
192
|
| 56 |
+
]
|
| 57 |
+
},
|
| 58 |
+
"enc_blocks.0.norm1.bias": {
|
| 59 |
+
"w_norm": 0.06115815043449402,
|
| 60 |
+
"g_norm": 0.024567998945713043,
|
| 61 |
+
"shape": [
|
| 62 |
+
192
|
| 63 |
+
]
|
| 64 |
+
},
|
| 65 |
+
"enc_blocks.0.conv1.weight": {
|
| 66 |
+
"w_norm": 8.222108840942383,
|
| 67 |
+
"g_norm": 0.1748814433813095,
|
| 68 |
+
"shape": [
|
| 69 |
+
192,
|
| 70 |
+
192,
|
| 71 |
+
3,
|
| 72 |
+
3
|
| 73 |
+
]
|
| 74 |
+
},
|
| 75 |
+
"enc_blocks.0.conv1.bias": {
|
| 76 |
+
"w_norm": 0.20211568474769592,
|
| 77 |
+
"g_norm": 0.01814374327659607,
|
| 78 |
+
"shape": [
|
| 79 |
+
192
|
| 80 |
+
]
|
| 81 |
+
},
|
| 82 |
+
"enc_blocks.0.t_proj.weight": {
|
| 83 |
+
"w_norm": 8.027454376220703,
|
| 84 |
+
"g_norm": 0.008248274214565754,
|
| 85 |
+
"shape": [
|
| 86 |
+
192,
|
| 87 |
+
256
|
| 88 |
+
]
|
| 89 |
+
},
|
| 90 |
+
"enc_blocks.0.t_proj.bias": {
|
| 91 |
+
"w_norm": 0.5023583173751831,
|
| 92 |
+
"g_norm": 0.018131105229258537,
|
| 93 |
+
"shape": [
|
| 94 |
+
192
|
| 95 |
+
]
|
| 96 |
+
},
|
| 97 |
+
"enc_blocks.0.norm2.weight": {
|
| 98 |
+
"w_norm": 13.845124244689941,
|
| 99 |
+
"g_norm": 0.006467544473707676,
|
| 100 |
+
"shape": [
|
| 101 |
+
192
|
| 102 |
+
]
|
| 103 |
+
},
|
| 104 |
+
"enc_blocks.0.norm2.bias": {
|
| 105 |
+
"w_norm": 0.035126347094774246,
|
| 106 |
+
"g_norm": 0.00743205426260829,
|
| 107 |
+
"shape": [
|
| 108 |
+
192
|
| 109 |
+
]
|
| 110 |
+
},
|
| 111 |
+
"enc_blocks.0.conv2.weight": {
|
| 112 |
+
"w_norm": 8.169148445129395,
|
| 113 |
+
"g_norm": 0.07441551238298416,
|
| 114 |
+
"shape": [
|
| 115 |
+
192,
|
| 116 |
+
192,
|
| 117 |
+
3,
|
| 118 |
+
3
|
| 119 |
+
]
|
| 120 |
+
},
|
| 121 |
+
"enc_blocks.0.conv2.bias": {
|
| 122 |
+
"w_norm": 0.1873043030500412,
|
| 123 |
+
"g_norm": 0.005247351247817278,
|
| 124 |
+
"shape": [
|
| 125 |
+
192
|
| 126 |
+
]
|
| 127 |
+
},
|
| 128 |
+
"enc_blocks.1.norm1.weight": {
|
| 129 |
+
"w_norm": 13.851595878601074,
|
| 130 |
+
"g_norm": 3.0715102184331045e-05,
|
| 131 |
+
"shape": [
|
| 132 |
+
192
|
| 133 |
+
]
|
| 134 |
+
},
|
| 135 |
+
"enc_blocks.1.norm1.bias": {
|
| 136 |
+
"w_norm": 0.019507592543959618,
|
| 137 |
+
"g_norm": 2.6044943297165446e-05,
|
| 138 |
+
"shape": [
|
| 139 |
+
192
|
| 140 |
+
]
|
| 141 |
+
},
|
| 142 |
+
"enc_blocks.1.conv1.weight": {
|
| 143 |
+
"w_norm": 8.134106636047363,
|
| 144 |
+
"g_norm": 0.0014839934883639216,
|
| 145 |
+
"shape": [
|
| 146 |
+
192,
|
| 147 |
+
192,
|
| 148 |
+
3,
|
| 149 |
+
3
|
| 150 |
+
]
|
| 151 |
+
},
|
| 152 |
+
"enc_blocks.1.conv1.bias": {
|
| 153 |
+
"w_norm": 0.19206438958644867,
|
| 154 |
+
"g_norm": 6.173425208544359e-05,
|
| 155 |
+
"shape": [
|
| 156 |
+
192
|
| 157 |
+
]
|
| 158 |
+
},
|
| 159 |
+
"enc_blocks.1.t_proj.weight": {
|
| 160 |
+
"w_norm": 8.007999420166016,
|
| 161 |
+
"g_norm": 3.269246008130722e-05,
|
| 162 |
+
"shape": [
|
| 163 |
+
192,
|
| 164 |
+
256
|
| 165 |
+
]
|
| 166 |
+
},
|
| 167 |
+
"enc_blocks.1.t_proj.bias": {
|
| 168 |
+
"w_norm": 0.5217474102973938,
|
| 169 |
+
"g_norm": 6.174553709570318e-05,
|
| 170 |
+
"shape": [
|
| 171 |
+
192
|
| 172 |
+
]
|
| 173 |
+
},
|
| 174 |
+
"enc_blocks.1.norm2.weight": {
|
| 175 |
+
"w_norm": 13.828197479248047,
|
| 176 |
+
"g_norm": 0.00014807471598032862,
|
| 177 |
+
"shape": [
|
| 178 |
+
192
|
| 179 |
+
]
|
| 180 |
+
},
|
| 181 |
+
"enc_blocks.1.norm2.bias": {
|
| 182 |
+
"w_norm": 0.031187444925308228,
|
| 183 |
+
"g_norm": 0.00011828799324575812,
|
| 184 |
+
"shape": [
|
| 185 |
+
192
|
| 186 |
+
]
|
| 187 |
+
},
|
| 188 |
+
"enc_blocks.1.conv2.weight": {
|
| 189 |
+
"w_norm": 8.091903686523438,
|
| 190 |
+
"g_norm": 0.004907071590423584,
|
| 191 |
+
"shape": [
|
| 192 |
+
192,
|
| 193 |
+
192,
|
| 194 |
+
3,
|
| 195 |
+
3
|
| 196 |
+
]
|
| 197 |
+
},
|
| 198 |
+
"enc_blocks.1.conv2.bias": {
|
| 199 |
+
"w_norm": 0.20746883749961853,
|
| 200 |
+
"g_norm": 0.0002124477905454114,
|
| 201 |
+
"shape": [
|
| 202 |
+
192
|
| 203 |
+
]
|
| 204 |
+
},
|
| 205 |
+
"enc_blocks.2.norm1.weight": {
|
| 206 |
+
"w_norm": 13.850759506225586,
|
| 207 |
+
"g_norm": 9.877112461254e-05,
|
| 208 |
+
"shape": [
|
| 209 |
+
192
|
| 210 |
+
]
|
| 211 |
+
},
|
| 212 |
+
"enc_blocks.2.norm1.bias": {
|
| 213 |
+
"w_norm": 0.027128858491778374,
|
| 214 |
+
"g_norm": 9.442120062885806e-05,
|
| 215 |
+
"shape": [
|
| 216 |
+
192
|
| 217 |
+
]
|
| 218 |
+
},
|
| 219 |
+
"enc_blocks.2.conv1.weight": {
|
| 220 |
+
"w_norm": 11.504433631896973,
|
| 221 |
+
"g_norm": 0.00176367093808949,
|
| 222 |
+
"shape": [
|
| 223 |
+
384,
|
| 224 |
+
192,
|
| 225 |
+
3,
|
| 226 |
+
3
|
| 227 |
+
]
|
| 228 |
+
},
|
| 229 |
+
"enc_blocks.2.conv1.bias": {
|
| 230 |
+
"w_norm": 0.2661369740962982,
|
| 231 |
+
"g_norm": 7.727985212113708e-05,
|
| 232 |
+
"shape": [
|
| 233 |
+
384
|
| 234 |
+
]
|
| 235 |
+
},
|
| 236 |
+
"enc_blocks.2.t_proj.weight": {
|
| 237 |
+
"w_norm": 11.387003898620605,
|
| 238 |
+
"g_norm": 4.298925705370493e-05,
|
| 239 |
+
"shape": [
|
| 240 |
+
384,
|
| 241 |
+
256
|
| 242 |
+
]
|
| 243 |
+
},
|
| 244 |
+
"enc_blocks.2.t_proj.bias": {
|
| 245 |
+
"w_norm": 0.7206763625144958,
|
| 246 |
+
"g_norm": 7.726851617917418e-05,
|
| 247 |
+
"shape": [
|
| 248 |
+
384
|
| 249 |
+
]
|
| 250 |
+
},
|
| 251 |
+
"enc_blocks.2.norm2.weight": {
|
| 252 |
+
"w_norm": 19.513442993164062,
|
| 253 |
+
"g_norm": 6.003796079312451e-05,
|
| 254 |
+
"shape": [
|
| 255 |
+
384
|
| 256 |
+
]
|
| 257 |
+
},
|
| 258 |
+
"enc_blocks.2.norm2.bias": {
|
| 259 |
+
"w_norm": 0.07231348007917404,
|
| 260 |
+
"g_norm": 5.830789086758159e-05,
|
| 261 |
+
"shape": [
|
| 262 |
+
384
|
| 263 |
+
]
|
| 264 |
+
},
|
| 265 |
+
"enc_blocks.2.conv2.weight": {
|
| 266 |
+
"w_norm": 11.476949691772461,
|
| 267 |
+
"g_norm": 0.002501599956303835,
|
| 268 |
+
"shape": [
|
| 269 |
+
384,
|
| 270 |
+
384,
|
| 271 |
+
3,
|
| 272 |
+
3
|
| 273 |
+
]
|
| 274 |
+
},
|
| 275 |
+
"enc_blocks.2.conv2.bias": {
|
| 276 |
+
"w_norm": 0.2027990072965622,
|
| 277 |
+
"g_norm": 0.00010226813901681453,
|
| 278 |
+
"shape": [
|
| 279 |
+
384
|
| 280 |
+
]
|
| 281 |
+
},
|
| 282 |
+
"enc_blocks.2.skip.weight": {
|
| 283 |
+
"w_norm": 11.265475273132324,
|
| 284 |
+
"g_norm": 0.0029775125440210104,
|
| 285 |
+
"shape": [
|
| 286 |
+
384,
|
| 287 |
+
192,
|
| 288 |
+
1,
|
| 289 |
+
1
|
| 290 |
+
]
|
| 291 |
+
},
|
| 292 |
+
"enc_blocks.2.skip.bias": {
|
| 293 |
+
"w_norm": 0.8253466486930847,
|
| 294 |
+
"g_norm": 0.00010226813901681453,
|
| 295 |
+
"shape": [
|
| 296 |
+
384
|
| 297 |
+
]
|
| 298 |
+
},
|
| 299 |
+
"enc_blocks.3.norm1.weight": {
|
| 300 |
+
"w_norm": 19.597091674804688,
|
| 301 |
+
"g_norm": 7.468698640877847e-06,
|
| 302 |
+
"shape": [
|
| 303 |
+
384
|
| 304 |
+
]
|
| 305 |
+
},
|
| 306 |
+
"enc_blocks.3.norm1.bias": {
|
| 307 |
+
"w_norm": 0.03287740424275398,
|
| 308 |
+
"g_norm": 5.63786943530431e-06,
|
| 309 |
+
"shape": [
|
| 310 |
+
384
|
| 311 |
+
]
|
| 312 |
+
},
|
| 313 |
+
"enc_blocks.3.conv1.weight": {
|
| 314 |
+
"w_norm": 11.572396278381348,
|
| 315 |
+
"g_norm": 0.00018376416119281203,
|
| 316 |
+
"shape": [
|
| 317 |
+
384,
|
| 318 |
+
384,
|
| 319 |
+
3,
|
| 320 |
+
3
|
| 321 |
+
]
|
| 322 |
+
},
|
| 323 |
+
"enc_blocks.3.conv1.bias": {
|
| 324 |
+
"w_norm": 0.19801898300647736,
|
| 325 |
+
"g_norm": 5.174441412236774e-06,
|
| 326 |
+
"shape": [
|
| 327 |
+
384
|
| 328 |
+
]
|
| 329 |
+
},
|
| 330 |
+
"enc_blocks.3.t_proj.weight": {
|
| 331 |
+
"w_norm": 11.330994606018066,
|
| 332 |
+
"g_norm": 2.8422682589734904e-06,
|
| 333 |
+
"shape": [
|
| 334 |
+
384,
|
| 335 |
+
256
|
| 336 |
+
]
|
| 337 |
+
},
|
| 338 |
+
"enc_blocks.3.t_proj.bias": {
|
| 339 |
+
"w_norm": 0.690720796585083,
|
| 340 |
+
"g_norm": 5.173218596610241e-06,
|
| 341 |
+
"shape": [
|
| 342 |
+
384
|
| 343 |
+
]
|
| 344 |
+
},
|
| 345 |
+
"enc_blocks.3.norm2.weight": {
|
| 346 |
+
"w_norm": 19.522457122802734,
|
| 347 |
+
"g_norm": 1.5316471035475843e-05,
|
| 348 |
+
"shape": [
|
| 349 |
+
384
|
| 350 |
+
]
|
| 351 |
+
},
|
| 352 |
+
"enc_blocks.3.norm2.bias": {
|
| 353 |
+
"w_norm": 0.07057540118694305,
|
| 354 |
+
"g_norm": 7.84625080996193e-06,
|
| 355 |
+
"shape": [
|
| 356 |
+
384
|
| 357 |
+
]
|
| 358 |
+
},
|
| 359 |
+
"enc_blocks.3.conv2.weight": {
|
| 360 |
+
"w_norm": 11.416537284851074,
|
| 361 |
+
"g_norm": 0.0004853545979131013,
|
| 362 |
+
"shape": [
|
| 363 |
+
384,
|
| 364 |
+
384,
|
| 365 |
+
3,
|
| 366 |
+
3
|
| 367 |
+
]
|
| 368 |
+
},
|
| 369 |
+
"enc_blocks.3.conv2.bias": {
|
| 370 |
+
"w_norm": 0.19048988819122314,
|
| 371 |
+
"g_norm": 1.8962364265462384e-05,
|
| 372 |
+
"shape": [
|
| 373 |
+
384
|
| 374 |
+
]
|
| 375 |
+
},
|
| 376 |
+
"enc_blocks.4.norm1.weight": {
|
| 377 |
+
"w_norm": 19.59347915649414,
|
| 378 |
+
"g_norm": 3.5858311093761586e-06,
|
| 379 |
+
"shape": [
|
| 380 |
+
384
|
| 381 |
+
]
|
| 382 |
+
},
|
| 383 |
+
"enc_blocks.4.norm1.bias": {
|
| 384 |
+
"w_norm": 0.03173157572746277,
|
| 385 |
+
"g_norm": 4.146143055550056e-06,
|
| 386 |
+
"shape": [
|
| 387 |
+
384
|
| 388 |
+
]
|
| 389 |
+
},
|
| 390 |
+
"enc_blocks.4.conv1.weight": {
|
| 391 |
+
"w_norm": 16.279399871826172,
|
| 392 |
+
"g_norm": 0.0001275187823921442,
|
| 393 |
+
"shape": [
|
| 394 |
+
768,
|
| 395 |
+
384,
|
| 396 |
+
3,
|
| 397 |
+
3
|
| 398 |
+
]
|
| 399 |
+
},
|
| 400 |
+
"enc_blocks.4.conv1.bias": {
|
| 401 |
+
"w_norm": 0.28021085262298584,
|
| 402 |
+
"g_norm": 4.764061486639548e-06,
|
| 403 |
+
"shape": [
|
| 404 |
+
768
|
| 405 |
+
]
|
| 406 |
+
},
|
| 407 |
+
"enc_blocks.4.t_proj.weight": {
|
| 408 |
+
"w_norm": 16.028104782104492,
|
| 409 |
+
"g_norm": 2.406797420917428e-06,
|
| 410 |
+
"shape": [
|
| 411 |
+
768,
|
| 412 |
+
256
|
| 413 |
+
]
|
| 414 |
+
},
|
| 415 |
+
"enc_blocks.4.t_proj.bias": {
|
| 416 |
+
"w_norm": 0.990105926990509,
|
| 417 |
+
"g_norm": 4.763509878102923e-06,
|
| 418 |
+
"shape": [
|
| 419 |
+
768
|
| 420 |
+
]
|
| 421 |
+
},
|
| 422 |
+
"enc_blocks.4.norm2.weight": {
|
| 423 |
+
"w_norm": 27.602861404418945,
|
| 424 |
+
"g_norm": 6.712813046760857e-06,
|
| 425 |
+
"shape": [
|
| 426 |
+
768
|
| 427 |
+
]
|
| 428 |
+
},
|
| 429 |
+
"enc_blocks.4.norm2.bias": {
|
| 430 |
+
"w_norm": 0.10398496687412262,
|
| 431 |
+
"g_norm": 7.564138741145143e-06,
|
| 432 |
+
"shape": [
|
| 433 |
+
768
|
| 434 |
+
]
|
| 435 |
+
},
|
| 436 |
+
"enc_blocks.4.conv2.weight": {
|
| 437 |
+
"w_norm": 16.15690803527832,
|
| 438 |
+
"g_norm": 0.00038142167613841593,
|
| 439 |
+
"shape": [
|
| 440 |
+
768,
|
| 441 |
+
768,
|
| 442 |
+
3,
|
| 443 |
+
3
|
| 444 |
+
]
|
| 445 |
+
},
|
| 446 |
+
"enc_blocks.4.conv2.bias": {
|
| 447 |
+
"w_norm": 0.19474577903747559,
|
| 448 |
+
"g_norm": 1.245525709236972e-05,
|
| 449 |
+
"shape": [
|
| 450 |
+
768
|
| 451 |
+
]
|
| 452 |
+
},
|
| 453 |
+
"enc_blocks.4.skip.weight": {
|
| 454 |
+
"w_norm": 15.938165664672852,
|
| 455 |
+
"g_norm": 0.0006917706923559308,
|
| 456 |
+
"shape": [
|
| 457 |
+
768,
|
| 458 |
+
384,
|
| 459 |
+
1,
|
| 460 |
+
1
|
| 461 |
+
]
|
| 462 |
+
},
|
| 463 |
+
"enc_blocks.4.skip.bias": {
|
| 464 |
+
"w_norm": 0.8077543377876282,
|
| 465 |
+
"g_norm": 1.245525709236972e-05,
|
| 466 |
+
"shape": [
|
| 467 |
+
768
|
| 468 |
+
]
|
| 469 |
+
},
|
| 470 |
+
"enc_blocks.5.norm1.weight": {
|
| 471 |
+
"w_norm": 27.713071823120117,
|
| 472 |
+
"g_norm": 8.311205306199554e-07,
|
| 473 |
+
"shape": [
|
| 474 |
+
768
|
| 475 |
+
]
|
| 476 |
+
},
|
| 477 |
+
"enc_blocks.5.norm1.bias": {
|
| 478 |
+
"w_norm": 0.05238156393170357,
|
| 479 |
+
"g_norm": 1.1990509847237263e-06,
|
| 480 |
+
"shape": [
|
| 481 |
+
768
|
| 482 |
+
]
|
| 483 |
+
},
|
| 484 |
+
"enc_blocks.5.conv1.weight": {
|
| 485 |
+
"w_norm": 16.445384979248047,
|
| 486 |
+
"g_norm": 5.676236833096482e-05,
|
| 487 |
+
"shape": [
|
| 488 |
+
768,
|
| 489 |
+
768,
|
| 490 |
+
3,
|
| 491 |
+
3
|
| 492 |
+
]
|
| 493 |
+
},
|
| 494 |
+
"enc_blocks.5.conv1.bias": {
|
| 495 |
+
"w_norm": 0.20182201266288757,
|
| 496 |
+
"g_norm": 1.51737867781776e-06,
|
| 497 |
+
"shape": [
|
| 498 |
+
768
|
| 499 |
+
]
|
| 500 |
+
},
|
| 501 |
+
"enc_blocks.5.t_proj.weight": {
|
| 502 |
+
"w_norm": 15.997159957885742,
|
| 503 |
+
"g_norm": 7.934851851132407e-07,
|
| 504 |
+
"shape": [
|
| 505 |
+
768,
|
| 506 |
+
256
|
| 507 |
+
]
|
| 508 |
+
},
|
| 509 |
+
"enc_blocks.5.t_proj.bias": {
|
| 510 |
+
"w_norm": 0.9971441030502319,
|
| 511 |
+
"g_norm": 1.5173413885349873e-06,
|
| 512 |
+
"shape": [
|
| 513 |
+
768
|
| 514 |
+
]
|
| 515 |
+
},
|
| 516 |
+
"enc_blocks.5.norm2.weight": {
|
| 517 |
+
"w_norm": 27.63616943359375,
|
| 518 |
+
"g_norm": 2.516604354241281e-06,
|
| 519 |
+
"shape": [
|
| 520 |
+
768
|
| 521 |
+
]
|
| 522 |
+
},
|
| 523 |
+
"enc_blocks.5.norm2.bias": {
|
| 524 |
+
"w_norm": 0.08157779276371002,
|
| 525 |
+
"g_norm": 3.0129847345961025e-06,
|
| 526 |
+
"shape": [
|
| 527 |
+
768
|
| 528 |
+
]
|
| 529 |
+
},
|
| 530 |
+
"enc_blocks.5.conv2.weight": {
|
| 531 |
+
"w_norm": 16.266035079956055,
|
| 532 |
+
"g_norm": 0.00022640718088950962,
|
| 533 |
+
"shape": [
|
| 534 |
+
768,
|
| 535 |
+
768,
|
| 536 |
+
3,
|
| 537 |
+
3
|
| 538 |
+
]
|
| 539 |
+
},
|
| 540 |
+
"enc_blocks.5.conv2.bias": {
|
| 541 |
+
"w_norm": 0.19263482093811035,
|
| 542 |
+
"g_norm": 7.249552709254203e-06,
|
| 543 |
+
"shape": [
|
| 544 |
+
768
|
| 545 |
+
]
|
| 546 |
+
},
|
| 547 |
+
"enc_attns.3.norm.weight": {
|
| 548 |
+
"w_norm": 19.5748291015625,
|
| 549 |
+
"g_norm": 7.831244147382677e-06,
|
| 550 |
+
"shape": [
|
| 551 |
+
384
|
| 552 |
+
]
|
| 553 |
+
},
|
| 554 |
+
"enc_attns.3.norm.bias": {
|
| 555 |
+
"w_norm": 0.041913531720638275,
|
| 556 |
+
"g_norm": 7.98945166025078e-06,
|
| 557 |
+
"shape": [
|
| 558 |
+
384
|
| 559 |
+
]
|
| 560 |
+
},
|
| 561 |
+
"enc_attns.3.qkv.weight": {
|
| 562 |
+
"w_norm": 19.660964965820312,
|
| 563 |
+
"g_norm": 0.00023820465139579028,
|
| 564 |
+
"shape": [
|
| 565 |
+
1152,
|
| 566 |
+
384,
|
| 567 |
+
1,
|
| 568 |
+
1
|
| 569 |
+
]
|
| 570 |
+
},
|
| 571 |
+
"enc_attns.3.qkv.bias": {
|
| 572 |
+
"w_norm": 1.0083452463150024,
|
| 573 |
+
"g_norm": 1.2227897968841717e-05,
|
| 574 |
+
"shape": [
|
| 575 |
+
1152
|
| 576 |
+
]
|
| 577 |
+
},
|
| 578 |
+
"enc_attns.3.proj.weight": {
|
| 579 |
+
"w_norm": 11.306562423706055,
|
| 580 |
+
"g_norm": 0.00021490734070539474,
|
| 581 |
+
"shape": [
|
| 582 |
+
384,
|
| 583 |
+
384,
|
| 584 |
+
1,
|
| 585 |
+
1
|
| 586 |
+
]
|
| 587 |
+
},
|
| 588 |
+
"enc_attns.3.proj.bias": {
|
| 589 |
+
"w_norm": 0.5660514235496521,
|
| 590 |
+
"g_norm": 1.9022467313334346e-05,
|
| 591 |
+
"shape": [
|
| 592 |
+
384
|
| 593 |
+
]
|
| 594 |
+
},
|
| 595 |
+
"enc_attns.4.norm.weight": {
|
| 596 |
+
"w_norm": 27.66777229309082,
|
| 597 |
+
"g_norm": 2.031427129622898e-06,
|
| 598 |
+
"shape": [
|
| 599 |
+
768
|
| 600 |
+
]
|
| 601 |
+
},
|
| 602 |
+
"enc_attns.4.norm.bias": {
|
| 603 |
+
"w_norm": 0.034273792058229446,
|
| 604 |
+
"g_norm": 4.666352651838679e-06,
|
| 605 |
+
"shape": [
|
| 606 |
+
768
|
| 607 |
+
]
|
| 608 |
+
},
|
| 609 |
+
"enc_attns.4.qkv.weight": {
|
| 610 |
+
"w_norm": 27.864606857299805,
|
| 611 |
+
"g_norm": 8.733879076316953e-05,
|
| 612 |
+
"shape": [
|
| 613 |
+
2304,
|
| 614 |
+
768,
|
| 615 |
+
1,
|
| 616 |
+
1
|
| 617 |
+
]
|
| 618 |
+
},
|
| 619 |
+
"enc_attns.4.qkv.bias": {
|
| 620 |
+
"w_norm": 1.0126442909240723,
|
| 621 |
+
"g_norm": 7.294429906323785e-06,
|
| 622 |
+
"shape": [
|
| 623 |
+
2304
|
| 624 |
+
]
|
| 625 |
+
},
|
| 626 |
+
"enc_attns.4.proj.weight": {
|
| 627 |
+
"w_norm": 15.98458194732666,
|
| 628 |
+
"g_norm": 7.8769022366032e-05,
|
| 629 |
+
"shape": [
|
| 630 |
+
768,
|
| 631 |
+
768,
|
| 632 |
+
1,
|
| 633 |
+
1
|
| 634 |
+
]
|
| 635 |
+
},
|
| 636 |
+
"enc_attns.4.proj.bias": {
|
| 637 |
+
"w_norm": 0.5737269520759583,
|
| 638 |
+
"g_norm": 1.2380953194224276e-05,
|
| 639 |
+
"shape": [
|
| 640 |
+
768
|
| 641 |
+
]
|
| 642 |
+
},
|
| 643 |
+
"enc_attns.5.norm.weight": {
|
| 644 |
+
"w_norm": 27.681488037109375,
|
| 645 |
+
"g_norm": 1.4364586604642682e-06,
|
| 646 |
+
"shape": [
|
| 647 |
+
768
|
| 648 |
+
]
|
| 649 |
+
},
|
| 650 |
+
"enc_attns.5.norm.bias": {
|
| 651 |
+
"w_norm": 0.03207901865243912,
|
| 652 |
+
"g_norm": 2.4433634280285332e-06,
|
| 653 |
+
"shape": [
|
| 654 |
+
768
|
| 655 |
+
]
|
| 656 |
+
},
|
| 657 |
+
"enc_attns.5.qkv.weight": {
|
| 658 |
+
"w_norm": 27.812726974487305,
|
| 659 |
+
"g_norm": 6.495189882116392e-05,
|
| 660 |
+
"shape": [
|
| 661 |
+
2304,
|
| 662 |
+
768,
|
| 663 |
+
1,
|
| 664 |
+
1
|
| 665 |
+
]
|
| 666 |
+
},
|
| 667 |
+
"enc_attns.5.qkv.bias": {
|
| 668 |
+
"w_norm": 0.9992081522941589,
|
| 669 |
+
"g_norm": 4.105900643480709e-06,
|
| 670 |
+
"shape": [
|
| 671 |
+
2304
|
| 672 |
+
]
|
| 673 |
+
},
|
| 674 |
+
"enc_attns.5.proj.weight": {
|
| 675 |
+
"w_norm": 16.01068687438965,
|
| 676 |
+
"g_norm": 7.401613402180374e-05,
|
| 677 |
+
"shape": [
|
| 678 |
+
768,
|
| 679 |
+
768,
|
| 680 |
+
1,
|
| 681 |
+
1
|
| 682 |
+
]
|
| 683 |
+
},
|
| 684 |
+
"enc_attns.5.proj.bias": {
|
| 685 |
+
"w_norm": 0.5965924859046936,
|
| 686 |
+
"g_norm": 7.23396624380257e-06,
|
| 687 |
+
"shape": [
|
| 688 |
+
768
|
| 689 |
+
]
|
| 690 |
+
},
|
| 691 |
+
"enc_downs.0.op.weight": {
|
| 692 |
+
"w_norm": 8.119194030761719,
|
| 693 |
+
"g_norm": 0.00474424147978425,
|
| 694 |
+
"shape": [
|
| 695 |
+
192,
|
| 696 |
+
192,
|
| 697 |
+
3,
|
| 698 |
+
3
|
| 699 |
+
]
|
| 700 |
+
},
|
| 701 |
+
"enc_downs.0.op.bias": {
|
| 702 |
+
"w_norm": 0.18655341863632202,
|
| 703 |
+
"g_norm": 0.0002124822058249265,
|
| 704 |
+
"shape": [
|
| 705 |
+
192
|
| 706 |
+
]
|
| 707 |
+
},
|
| 708 |
+
"enc_downs.1.op.weight": {
|
| 709 |
+
"w_norm": 8.064340591430664,
|
| 710 |
+
"g_norm": 0.00629104720428586,
|
| 711 |
+
"shape": [
|
| 712 |
+
192,
|
| 713 |
+
192,
|
| 714 |
+
3,
|
| 715 |
+
3
|
| 716 |
+
]
|
| 717 |
+
},
|
| 718 |
+
"enc_downs.1.op.bias": {
|
| 719 |
+
"w_norm": 0.19401651620864868,
|
| 720 |
+
"g_norm": 0.00010197861411143094,
|
| 721 |
+
"shape": [
|
| 722 |
+
192
|
| 723 |
+
]
|
| 724 |
+
},
|
| 725 |
+
"enc_downs.2.op.weight": {
|
| 726 |
+
"w_norm": 11.412195205688477,
|
| 727 |
+
"g_norm": 0.0016165502602234483,
|
| 728 |
+
"shape": [
|
| 729 |
+
384,
|
| 730 |
+
384,
|
| 731 |
+
3,
|
| 732 |
+
3
|
| 733 |
+
]
|
| 734 |
+
},
|
| 735 |
+
"enc_downs.2.op.bias": {
|
| 736 |
+
"w_norm": 0.18826229870319366,
|
| 737 |
+
"g_norm": 1.9014922145288438e-05,
|
| 738 |
+
"shape": [
|
| 739 |
+
384
|
| 740 |
+
]
|
| 741 |
+
},
|
| 742 |
+
"enc_downs.3.op.weight": {
|
| 743 |
+
"w_norm": 11.37773609161377,
|
| 744 |
+
"g_norm": 0.0009675542241893709,
|
| 745 |
+
"shape": [
|
| 746 |
+
384,
|
| 747 |
+
384,
|
| 748 |
+
3,
|
| 749 |
+
3
|
| 750 |
+
]
|
| 751 |
+
},
|
| 752 |
+
"enc_downs.3.op.bias": {
|
| 753 |
+
"w_norm": 0.19311626255512238,
|
| 754 |
+
"g_norm": 8.070419426076114e-06,
|
| 755 |
+
"shape": [
|
| 756 |
+
384
|
| 757 |
+
]
|
| 758 |
+
},
|
| 759 |
+
"enc_downs.4.op.weight": {
|
| 760 |
+
"w_norm": 16.181838989257812,
|
| 761 |
+
"g_norm": 0.0011415004264563322,
|
| 762 |
+
"shape": [
|
| 763 |
+
768,
|
| 764 |
+
768,
|
| 765 |
+
3,
|
| 766 |
+
3
|
| 767 |
+
]
|
| 768 |
+
},
|
| 769 |
+
"enc_downs.4.op.bias": {
|
| 770 |
+
"w_norm": 0.19675636291503906,
|
| 771 |
+
"g_norm": 7.221314263006207e-06,
|
| 772 |
+
"shape": [
|
| 773 |
+
768
|
| 774 |
+
]
|
| 775 |
+
},
|
| 776 |
+
"mid1.norm1.weight": {
|
| 777 |
+
"w_norm": 27.711105346679688,
|
| 778 |
+
"g_norm": 1.734056468194467e-06,
|
| 779 |
+
"shape": [
|
| 780 |
+
768
|
| 781 |
+
]
|
| 782 |
+
},
|
| 783 |
+
"mid1.norm1.bias": {
|
| 784 |
+
"w_norm": 0.04435775801539421,
|
| 785 |
+
"g_norm": 1.6670099967086571e-06,
|
| 786 |
+
"shape": [
|
| 787 |
+
768
|
| 788 |
+
]
|
| 789 |
+
},
|
| 790 |
+
"mid1.conv1.weight": {
|
| 791 |
+
"w_norm": 16.44451332092285,
|
| 792 |
+
"g_norm": 6.236001354409382e-05,
|
| 793 |
+
"shape": [
|
| 794 |
+
768,
|
| 795 |
+
768,
|
| 796 |
+
3,
|
| 797 |
+
3
|
| 798 |
+
]
|
| 799 |
+
},
|
| 800 |
+
"mid1.conv1.bias": {
|
| 801 |
+
"w_norm": 0.20257723331451416,
|
| 802 |
+
"g_norm": 1.5994581872291747e-06,
|
| 803 |
+
"shape": [
|
| 804 |
+
768
|
| 805 |
+
]
|
| 806 |
+
},
|
| 807 |
+
"mid1.t_proj.weight": {
|
| 808 |
+
"w_norm": 16.007741928100586,
|
| 809 |
+
"g_norm": 8.396062298743345e-07,
|
| 810 |
+
"shape": [
|
| 811 |
+
768,
|
| 812 |
+
256
|
| 813 |
+
]
|
| 814 |
+
},
|
| 815 |
+
"mid1.t_proj.bias": {
|
| 816 |
+
"w_norm": 1.0318254232406616,
|
| 817 |
+
"g_norm": 1.5996096180970198e-06,
|
| 818 |
+
"shape": [
|
| 819 |
+
768
|
| 820 |
+
]
|
| 821 |
+
},
|
| 822 |
+
"mid1.norm2.weight": {
|
| 823 |
+
"w_norm": 27.638864517211914,
|
| 824 |
+
"g_norm": 2.9384184472291963e-06,
|
| 825 |
+
"shape": [
|
| 826 |
+
768
|
| 827 |
+
]
|
| 828 |
+
},
|
| 829 |
+
"mid1.norm2.bias": {
|
| 830 |
+
"w_norm": 0.08424504101276398,
|
| 831 |
+
"g_norm": 3.2435036700917408e-06,
|
| 832 |
+
"shape": [
|
| 833 |
+
768
|
| 834 |
+
]
|
| 835 |
+
},
|
| 836 |
+
"mid1.conv2.weight": {
|
| 837 |
+
"w_norm": 16.264646530151367,
|
| 838 |
+
"g_norm": 0.00015367653395514935,
|
| 839 |
+
"shape": [
|
| 840 |
+
768,
|
| 841 |
+
768,
|
| 842 |
+
3,
|
| 843 |
+
3
|
| 844 |
+
]
|
| 845 |
+
},
|
| 846 |
+
"mid1.conv2.bias": {
|
| 847 |
+
"w_norm": 0.19507506489753723,
|
| 848 |
+
"g_norm": 5.045553280069726e-06,
|
| 849 |
+
"shape": [
|
| 850 |
+
768
|
| 851 |
+
]
|
| 852 |
+
},
|
| 853 |
+
"mid_attn.norm.weight": {
|
| 854 |
+
"w_norm": 27.687448501586914,
|
| 855 |
+
"g_norm": 1.162366174867202e-06,
|
| 856 |
+
"shape": [
|
| 857 |
+
768
|
| 858 |
+
]
|
| 859 |
+
},
|
| 860 |
+
"mid_attn.norm.bias": {
|
| 861 |
+
"w_norm": 0.029944565147161484,
|
| 862 |
+
"g_norm": 1.7182350120492629e-06,
|
| 863 |
+
"shape": [
|
| 864 |
+
768
|
| 865 |
+
]
|
| 866 |
+
},
|
| 867 |
+
"mid_attn.qkv.weight": {
|
| 868 |
+
"w_norm": 27.804889678955078,
|
| 869 |
+
"g_norm": 5.264316496322863e-05,
|
| 870 |
+
"shape": [
|
| 871 |
+
2304,
|
| 872 |
+
768,
|
| 873 |
+
1,
|
| 874 |
+
1
|
| 875 |
+
]
|
| 876 |
+
},
|
| 877 |
+
"mid_attn.qkv.bias": {
|
| 878 |
+
"w_norm": 0.9861231446266174,
|
| 879 |
+
"g_norm": 2.871294555006898e-06,
|
| 880 |
+
"shape": [
|
| 881 |
+
2304
|
| 882 |
+
]
|
| 883 |
+
},
|
| 884 |
+
"mid_attn.proj.weight": {
|
| 885 |
+
"w_norm": 16.009910583496094,
|
| 886 |
+
"g_norm": 5.811886876472272e-05,
|
| 887 |
+
"shape": [
|
| 888 |
+
768,
|
| 889 |
+
768,
|
| 890 |
+
1,
|
| 891 |
+
1
|
| 892 |
+
]
|
| 893 |
+
},
|
| 894 |
+
"mid_attn.proj.bias": {
|
| 895 |
+
"w_norm": 0.576531708240509,
|
| 896 |
+
"g_norm": 5.044004410592606e-06,
|
| 897 |
+
"shape": [
|
| 898 |
+
768
|
| 899 |
+
]
|
| 900 |
+
},
|
| 901 |
+
"mid2.norm1.weight": {
|
| 902 |
+
"w_norm": 27.714092254638672,
|
| 903 |
+
"g_norm": 2.8204319733049488e-06,
|
| 904 |
+
"shape": [
|
| 905 |
+
768
|
| 906 |
+
]
|
| 907 |
+
},
|
| 908 |
+
"mid2.norm1.bias": {
|
| 909 |
+
"w_norm": 0.04821281507611275,
|
| 910 |
+
"g_norm": 2.875601012419793e-06,
|
| 911 |
+
"shape": [
|
| 912 |
+
768
|
| 913 |
+
]
|
| 914 |
+
},
|
| 915 |
+
"mid2.conv1.weight": {
|
| 916 |
+
"w_norm": 16.43076515197754,
|
| 917 |
+
"g_norm": 7.036297029117122e-05,
|
| 918 |
+
"shape": [
|
| 919 |
+
768,
|
| 920 |
+
768,
|
| 921 |
+
3,
|
| 922 |
+
3
|
| 923 |
+
]
|
| 924 |
+
},
|
| 925 |
+
"mid2.conv1.bias": {
|
| 926 |
+
"w_norm": 0.19785825908184052,
|
| 927 |
+
"g_norm": 1.8724236952039064e-06,
|
| 928 |
+
"shape": [
|
| 929 |
+
768
|
| 930 |
+
]
|
| 931 |
+
},
|
| 932 |
+
"mid2.t_proj.weight": {
|
| 933 |
+
"w_norm": 15.99503231048584,
|
| 934 |
+
"g_norm": 9.8669033832266e-07,
|
| 935 |
+
"shape": [
|
| 936 |
+
768,
|
| 937 |
+
256
|
| 938 |
+
]
|
| 939 |
+
},
|
| 940 |
+
"mid2.t_proj.bias": {
|
| 941 |
+
"w_norm": 0.9966008067131042,
|
| 942 |
+
"g_norm": 1.8730396504906821e-06,
|
| 943 |
+
"shape": [
|
| 944 |
+
768
|
| 945 |
+
]
|
| 946 |
+
},
|
| 947 |
+
"mid2.norm2.weight": {
|
| 948 |
+
"w_norm": 27.629024505615234,
|
| 949 |
+
"g_norm": 4.045460173074389e-06,
|
| 950 |
+
"shape": [
|
| 951 |
+
768
|
| 952 |
+
]
|
| 953 |
+
},
|
| 954 |
+
"mid2.norm2.bias": {
|
| 955 |
+
"w_norm": 0.09099399298429489,
|
| 956 |
+
"g_norm": 3.6871888369205408e-06,
|
| 957 |
+
"shape": [
|
| 958 |
+
768
|
| 959 |
+
]
|
| 960 |
+
},
|
| 961 |
+
"mid2.conv2.weight": {
|
| 962 |
+
"w_norm": 16.222124099731445,
|
| 963 |
+
"g_norm": 0.00016647143638692796,
|
| 964 |
+
"shape": [
|
| 965 |
+
768,
|
| 966 |
+
768,
|
| 967 |
+
3,
|
| 968 |
+
3
|
| 969 |
+
]
|
| 970 |
+
},
|
| 971 |
+
"mid2.conv2.bias": {
|
| 972 |
+
"w_norm": 0.189836323261261,
|
| 973 |
+
"g_norm": 5.019630407332443e-06,
|
| 974 |
+
"shape": [
|
| 975 |
+
768
|
| 976 |
+
]
|
| 977 |
+
},
|
| 978 |
+
"dec_blocks.0.norm1.weight": {
|
| 979 |
+
"w_norm": 39.19388198852539,
|
| 980 |
+
"g_norm": 1.2530965705082053e-06,
|
| 981 |
+
"shape": [
|
| 982 |
+
1536
|
| 983 |
+
]
|
| 984 |
+
},
|
| 985 |
+
"dec_blocks.0.norm1.bias": {
|
| 986 |
+
"w_norm": 0.05807539448142052,
|
| 987 |
+
"g_norm": 1.1799430694736657e-06,
|
| 988 |
+
"shape": [
|
| 989 |
+
1536
|
| 990 |
+
]
|
| 991 |
+
},
|
| 992 |
+
"dec_blocks.0.conv1.weight": {
|
| 993 |
+
"w_norm": 16.877586364746094,
|
| 994 |
+
"g_norm": 7.809244561940432e-05,
|
| 995 |
+
"shape": [
|
| 996 |
+
768,
|
| 997 |
+
1536,
|
| 998 |
+
3,
|
| 999 |
+
3
|
| 1000 |
+
]
|
| 1001 |
+
},
|
| 1002 |
+
"dec_blocks.0.conv1.bias": {
|
| 1003 |
+
"w_norm": 0.14163629710674286,
|
| 1004 |
+
"g_norm": 1.5247343299051863e-06,
|
| 1005 |
+
"shape": [
|
| 1006 |
+
768
|
| 1007 |
+
]
|
| 1008 |
+
},
|
| 1009 |
+
"dec_blocks.0.t_proj.weight": {
|
| 1010 |
+
"w_norm": 16.0123348236084,
|
| 1011 |
+
"g_norm": 7.94084598965128e-07,
|
| 1012 |
+
"shape": [
|
| 1013 |
+
768,
|
| 1014 |
+
256
|
| 1015 |
+
]
|
| 1016 |
+
},
|
| 1017 |
+
"dec_blocks.0.t_proj.bias": {
|
| 1018 |
+
"w_norm": 1.0203391313552856,
|
| 1019 |
+
"g_norm": 1.525260472590162e-06,
|
| 1020 |
+
"shape": [
|
| 1021 |
+
768
|
| 1022 |
+
]
|
| 1023 |
+
},
|
| 1024 |
+
"dec_blocks.0.norm2.weight": {
|
| 1025 |
+
"w_norm": 27.59062385559082,
|
| 1026 |
+
"g_norm": 2.9507796170946676e-06,
|
| 1027 |
+
"shape": [
|
| 1028 |
+
768
|
| 1029 |
+
]
|
| 1030 |
+
},
|
| 1031 |
+
"dec_blocks.0.norm2.bias": {
|
| 1032 |
+
"w_norm": 0.11849326640367508,
|
| 1033 |
+
"g_norm": 4.340486611909e-06,
|
| 1034 |
+
"shape": [
|
| 1035 |
+
768
|
| 1036 |
+
]
|
| 1037 |
+
},
|
| 1038 |
+
"dec_blocks.0.conv2.weight": {
|
| 1039 |
+
"w_norm": 16.1192626953125,
|
| 1040 |
+
"g_norm": 0.00026150690973736346,
|
| 1041 |
+
"shape": [
|
| 1042 |
+
768,
|
| 1043 |
+
768,
|
| 1044 |
+
3,
|
| 1045 |
+
3
|
| 1046 |
+
]
|
| 1047 |
+
},
|
| 1048 |
+
"dec_blocks.0.conv2.bias": {
|
| 1049 |
+
"w_norm": 0.19648997485637665,
|
| 1050 |
+
"g_norm": 1.0889924851653632e-05,
|
| 1051 |
+
"shape": [
|
| 1052 |
+
768
|
| 1053 |
+
]
|
| 1054 |
+
},
|
| 1055 |
+
"dec_blocks.0.skip.weight": {
|
| 1056 |
+
"w_norm": 15.902674674987793,
|
| 1057 |
+
"g_norm": 0.001218110672198236,
|
| 1058 |
+
"shape": [
|
| 1059 |
+
768,
|
| 1060 |
+
1536,
|
| 1061 |
+
1,
|
| 1062 |
+
1
|
| 1063 |
+
]
|
| 1064 |
+
},
|
| 1065 |
+
"dec_blocks.0.skip.bias": {
|
| 1066 |
+
"w_norm": 0.41060250997543335,
|
| 1067 |
+
"g_norm": 1.0889924851653632e-05,
|
| 1068 |
+
"shape": [
|
| 1069 |
+
768
|
| 1070 |
+
]
|
| 1071 |
+
},
|
| 1072 |
+
"dec_blocks.1.norm1.weight": {
|
| 1073 |
+
"w_norm": 39.1891975402832,
|
| 1074 |
+
"g_norm": 9.350132131658029e-06,
|
| 1075 |
+
"shape": [
|
| 1076 |
+
1536
|
| 1077 |
+
]
|
| 1078 |
+
},
|
| 1079 |
+
"dec_blocks.1.norm1.bias": {
|
| 1080 |
+
"w_norm": 0.053357407450675964,
|
| 1081 |
+
"g_norm": 1.0519148418097757e-05,
|
| 1082 |
+
"shape": [
|
| 1083 |
+
1536
|
| 1084 |
+
]
|
| 1085 |
+
},
|
| 1086 |
+
"dec_blocks.1.conv1.weight": {
|
| 1087 |
+
"w_norm": 16.819643020629883,
|
| 1088 |
+
"g_norm": 0.00029541790718212724,
|
| 1089 |
+
"shape": [
|
| 1090 |
+
768,
|
| 1091 |
+
1536,
|
| 1092 |
+
3,
|
| 1093 |
+
3
|
| 1094 |
+
]
|
| 1095 |
+
},
|
| 1096 |
+
"dec_blocks.1.conv1.bias": {
|
| 1097 |
+
"w_norm": 0.13914765417575836,
|
| 1098 |
+
"g_norm": 5.622245680569904e-06,
|
| 1099 |
+
"shape": [
|
| 1100 |
+
768
|
| 1101 |
+
]
|
| 1102 |
+
},
|
| 1103 |
+
"dec_blocks.1.t_proj.weight": {
|
| 1104 |
+
"w_norm": 16.038969039916992,
|
| 1105 |
+
"g_norm": 2.8538920560094994e-06,
|
| 1106 |
+
"shape": [
|
| 1107 |
+
768,
|
| 1108 |
+
256
|
| 1109 |
+
]
|
| 1110 |
+
},
|
| 1111 |
+
"dec_blocks.1.t_proj.bias": {
|
| 1112 |
+
"w_norm": 0.9858460426330566,
|
| 1113 |
+
"g_norm": 5.6221151680801995e-06,
|
| 1114 |
+
"shape": [
|
| 1115 |
+
768
|
| 1116 |
+
]
|
| 1117 |
+
},
|
| 1118 |
+
"dec_blocks.1.norm2.weight": {
|
| 1119 |
+
"w_norm": 27.590665817260742,
|
| 1120 |
+
"g_norm": 9.302767466579098e-06,
|
| 1121 |
+
"shape": [
|
| 1122 |
+
768
|
| 1123 |
+
]
|
| 1124 |
+
},
|
| 1125 |
+
"dec_blocks.1.norm2.bias": {
|
| 1126 |
+
"w_norm": 0.12908512353897095,
|
| 1127 |
+
"g_norm": 1.4394856407307088e-05,
|
| 1128 |
+
"shape": [
|
| 1129 |
+
768
|
| 1130 |
+
]
|
| 1131 |
+
},
|
| 1132 |
+
"dec_blocks.1.conv2.weight": {
|
| 1133 |
+
"w_norm": 16.1109619140625,
|
| 1134 |
+
"g_norm": 0.0006568268872797489,
|
| 1135 |
+
"shape": [
|
| 1136 |
+
768,
|
| 1137 |
+
768,
|
| 1138 |
+
3,
|
| 1139 |
+
3
|
| 1140 |
+
]
|
| 1141 |
+
},
|
| 1142 |
+
"dec_blocks.1.conv2.bias": {
|
| 1143 |
+
"w_norm": 0.19586755335330963,
|
| 1144 |
+
"g_norm": 2.2840289602754638e-05,
|
| 1145 |
+
"shape": [
|
| 1146 |
+
768
|
| 1147 |
+
]
|
| 1148 |
+
},
|
| 1149 |
+
"dec_blocks.1.skip.weight": {
|
| 1150 |
+
"w_norm": 15.894036293029785,
|
| 1151 |
+
"g_norm": 0.0016062397044152021,
|
| 1152 |
+
"shape": [
|
| 1153 |
+
768,
|
| 1154 |
+
1536,
|
| 1155 |
+
1,
|
| 1156 |
+
1
|
| 1157 |
+
]
|
| 1158 |
+
},
|
| 1159 |
+
"dec_blocks.1.skip.bias": {
|
| 1160 |
+
"w_norm": 0.40641382336616516,
|
| 1161 |
+
"g_norm": 2.2840289602754638e-05,
|
| 1162 |
+
"shape": [
|
| 1163 |
+
768
|
| 1164 |
+
]
|
| 1165 |
+
},
|
| 1166 |
+
"dec_blocks.2.norm1.weight": {
|
| 1167 |
+
"w_norm": 33.943851470947266,
|
| 1168 |
+
"g_norm": 1.3375498383538797e-05,
|
| 1169 |
+
"shape": [
|
| 1170 |
+
1152
|
| 1171 |
+
]
|
| 1172 |
+
},
|
| 1173 |
+
"dec_blocks.2.norm1.bias": {
|
| 1174 |
+
"w_norm": 0.05633563548326492,
|
| 1175 |
+
"g_norm": 1.2195808267279062e-05,
|
| 1176 |
+
"shape": [
|
| 1177 |
+
1152
|
| 1178 |
+
]
|
| 1179 |
+
},
|
| 1180 |
+
"dec_blocks.2.conv1.weight": {
|
| 1181 |
+
"w_norm": 11.906632423400879,
|
| 1182 |
+
"g_norm": 0.0004902135115116835,
|
| 1183 |
+
"shape": [
|
| 1184 |
+
384,
|
| 1185 |
+
1152,
|
| 1186 |
+
3,
|
| 1187 |
+
3
|
| 1188 |
+
]
|
| 1189 |
+
},
|
| 1190 |
+
"dec_blocks.2.conv1.bias": {
|
| 1191 |
+
"w_norm": 0.11643993854522705,
|
| 1192 |
+
"g_norm": 1.0101834959641565e-05,
|
| 1193 |
+
"shape": [
|
| 1194 |
+
384
|
| 1195 |
+
]
|
| 1196 |
+
},
|
| 1197 |
+
"dec_blocks.2.t_proj.weight": {
|
| 1198 |
+
"w_norm": 11.322545051574707,
|
| 1199 |
+
"g_norm": 5.009514097764622e-06,
|
| 1200 |
+
"shape": [
|
| 1201 |
+
384,
|
| 1202 |
+
256
|
| 1203 |
+
]
|
| 1204 |
+
},
|
| 1205 |
+
"dec_blocks.2.t_proj.bias": {
|
| 1206 |
+
"w_norm": 0.7094705700874329,
|
| 1207 |
+
"g_norm": 1.0103261047333945e-05,
|
| 1208 |
+
"shape": [
|
| 1209 |
+
384
|
| 1210 |
+
]
|
| 1211 |
+
},
|
| 1212 |
+
"dec_blocks.2.norm2.weight": {
|
| 1213 |
+
"w_norm": 19.51650619506836,
|
| 1214 |
+
"g_norm": 2.3810951461200602e-05,
|
| 1215 |
+
"shape": [
|
| 1216 |
+
384
|
| 1217 |
+
]
|
| 1218 |
+
},
|
| 1219 |
+
"dec_blocks.2.norm2.bias": {
|
| 1220 |
+
"w_norm": 0.085000179708004,
|
| 1221 |
+
"g_norm": 3.0295035685412586e-05,
|
| 1222 |
+
"shape": [
|
| 1223 |
+
384
|
| 1224 |
+
]
|
| 1225 |
+
},
|
| 1226 |
+
"dec_blocks.2.conv2.weight": {
|
| 1227 |
+
"w_norm": 11.389786720275879,
|
| 1228 |
+
"g_norm": 0.0011060911929234862,
|
| 1229 |
+
"shape": [
|
| 1230 |
+
384,
|
| 1231 |
+
384,
|
| 1232 |
+
3,
|
| 1233 |
+
3
|
| 1234 |
+
]
|
| 1235 |
+
},
|
| 1236 |
+
"dec_blocks.2.conv2.bias": {
|
| 1237 |
+
"w_norm": 0.18991783261299133,
|
| 1238 |
+
"g_norm": 5.410888115875423e-05,
|
| 1239 |
+
"shape": [
|
| 1240 |
+
384
|
| 1241 |
+
]
|
| 1242 |
+
},
|
| 1243 |
+
"dec_blocks.2.skip.weight": {
|
| 1244 |
+
"w_norm": 11.173041343688965,
|
| 1245 |
+
"g_norm": 0.0025957406032830477,
|
| 1246 |
+
"shape": [
|
| 1247 |
+
384,
|
| 1248 |
+
1152,
|
| 1249 |
+
1,
|
| 1250 |
+
1
|
| 1251 |
+
]
|
| 1252 |
+
},
|
| 1253 |
+
"dec_blocks.2.skip.bias": {
|
| 1254 |
+
"w_norm": 0.3321686089038849,
|
| 1255 |
+
"g_norm": 5.410888115875423e-05,
|
| 1256 |
+
"shape": [
|
| 1257 |
+
384
|
| 1258 |
+
]
|
| 1259 |
+
},
|
| 1260 |
+
"dec_blocks.3.norm1.weight": {
|
| 1261 |
+
"w_norm": 27.71334457397461,
|
| 1262 |
+
"g_norm": 6.114856660133228e-05,
|
| 1263 |
+
"shape": [
|
| 1264 |
+
768
|
| 1265 |
+
]
|
| 1266 |
+
},
|
| 1267 |
+
"dec_blocks.3.norm1.bias": {
|
| 1268 |
+
"w_norm": 0.05490655079483986,
|
| 1269 |
+
"g_norm": 5.5085638450691476e-05,
|
| 1270 |
+
"shape": [
|
| 1271 |
+
768
|
| 1272 |
+
]
|
| 1273 |
+
},
|
| 1274 |
+
"dec_blocks.3.conv1.weight": {
|
| 1275 |
+
"w_norm": 11.796333312988281,
|
| 1276 |
+
"g_norm": 0.0024704246316105127,
|
| 1277 |
+
"shape": [
|
| 1278 |
+
384,
|
| 1279 |
+
768,
|
| 1280 |
+
3,
|
| 1281 |
+
3
|
| 1282 |
+
]
|
| 1283 |
+
},
|
| 1284 |
+
"dec_blocks.3.conv1.bias": {
|
| 1285 |
+
"w_norm": 0.14261765778064728,
|
| 1286 |
+
"g_norm": 5.7586727052694187e-05,
|
| 1287 |
+
"shape": [
|
| 1288 |
+
384
|
| 1289 |
+
]
|
| 1290 |
+
},
|
| 1291 |
+
"dec_blocks.3.t_proj.weight": {
|
| 1292 |
+
"w_norm": 11.342194557189941,
|
| 1293 |
+
"g_norm": 2.818637221935205e-05,
|
| 1294 |
+
"shape": [
|
| 1295 |
+
384,
|
| 1296 |
+
256
|
| 1297 |
+
]
|
| 1298 |
+
},
|
| 1299 |
+
"dec_blocks.3.t_proj.bias": {
|
| 1300 |
+
"w_norm": 0.7088264226913452,
|
| 1301 |
+
"g_norm": 5.7579698477638885e-05,
|
| 1302 |
+
"shape": [
|
| 1303 |
+
384
|
| 1304 |
+
]
|
| 1305 |
+
},
|
| 1306 |
+
"dec_blocks.3.norm2.weight": {
|
| 1307 |
+
"w_norm": 19.502674102783203,
|
| 1308 |
+
"g_norm": 6.80582452332601e-05,
|
| 1309 |
+
"shape": [
|
| 1310 |
+
384
|
| 1311 |
+
]
|
| 1312 |
+
},
|
| 1313 |
+
"dec_blocks.3.norm2.bias": {
|
| 1314 |
+
"w_norm": 0.08563433587551117,
|
| 1315 |
+
"g_norm": 8.817730122245848e-05,
|
| 1316 |
+
"shape": [
|
| 1317 |
+
384
|
| 1318 |
+
]
|
| 1319 |
+
},
|
| 1320 |
+
"dec_blocks.3.conv2.weight": {
|
| 1321 |
+
"w_norm": 11.395421981811523,
|
| 1322 |
+
"g_norm": 0.004227672703564167,
|
| 1323 |
+
"shape": [
|
| 1324 |
+
384,
|
| 1325 |
+
384,
|
| 1326 |
+
3,
|
| 1327 |
+
3
|
| 1328 |
+
]
|
| 1329 |
+
},
|
| 1330 |
+
"dec_blocks.3.conv2.bias": {
|
| 1331 |
+
"w_norm": 0.19376030564308167,
|
| 1332 |
+
"g_norm": 0.00019759820133913308,
|
| 1333 |
+
"shape": [
|
| 1334 |
+
384
|
| 1335 |
+
]
|
| 1336 |
+
},
|
| 1337 |
+
"dec_blocks.3.skip.weight": {
|
| 1338 |
+
"w_norm": 11.19637680053711,
|
| 1339 |
+
"g_norm": 0.006898099556565285,
|
| 1340 |
+
"shape": [
|
| 1341 |
+
384,
|
| 1342 |
+
768,
|
| 1343 |
+
1,
|
| 1344 |
+
1
|
| 1345 |
+
]
|
| 1346 |
+
},
|
| 1347 |
+
"dec_blocks.3.skip.bias": {
|
| 1348 |
+
"w_norm": 0.40675294399261475,
|
| 1349 |
+
"g_norm": 0.00019759820133913308,
|
| 1350 |
+
"shape": [
|
| 1351 |
+
384
|
| 1352 |
+
]
|
| 1353 |
+
},
|
| 1354 |
+
"dec_blocks.4.norm1.weight": {
|
| 1355 |
+
"w_norm": 23.995790481567383,
|
| 1356 |
+
"g_norm": 7.700704009039328e-05,
|
| 1357 |
+
"shape": [
|
| 1358 |
+
576
|
| 1359 |
+
]
|
| 1360 |
+
},
|
| 1361 |
+
"dec_blocks.4.norm1.bias": {
|
| 1362 |
+
"w_norm": 0.05365603044629097,
|
| 1363 |
+
"g_norm": 7.204065332189202e-05,
|
| 1364 |
+
"shape": [
|
| 1365 |
+
576
|
| 1366 |
+
]
|
| 1367 |
+
},
|
| 1368 |
+
"dec_blocks.4.conv1.weight": {
|
| 1369 |
+
"w_norm": 8.276124000549316,
|
| 1370 |
+
"g_norm": 0.0042354753240942955,
|
| 1371 |
+
"shape": [
|
| 1372 |
+
192,
|
| 1373 |
+
576,
|
| 1374 |
+
3,
|
| 1375 |
+
3
|
| 1376 |
+
]
|
| 1377 |
+
},
|
| 1378 |
+
"dec_blocks.4.conv1.bias": {
|
| 1379 |
+
"w_norm": 0.11667870730161667,
|
| 1380 |
+
"g_norm": 0.0001092359161702916,
|
| 1381 |
+
"shape": [
|
| 1382 |
+
192
|
| 1383 |
+
]
|
| 1384 |
+
},
|
| 1385 |
+
"dec_blocks.4.t_proj.weight": {
|
| 1386 |
+
"w_norm": 8.005839347839355,
|
| 1387 |
+
"g_norm": 5.376513217925094e-05,
|
| 1388 |
+
"shape": [
|
| 1389 |
+
192,
|
| 1390 |
+
256
|
| 1391 |
+
]
|
| 1392 |
+
},
|
| 1393 |
+
"dec_blocks.4.t_proj.bias": {
|
| 1394 |
+
"w_norm": 0.4811122417449951,
|
| 1395 |
+
"g_norm": 0.00010927868424914777,
|
| 1396 |
+
"shape": [
|
| 1397 |
+
192
|
| 1398 |
+
]
|
| 1399 |
+
},
|
| 1400 |
+
"dec_blocks.4.norm2.weight": {
|
| 1401 |
+
"w_norm": 13.81904125213623,
|
| 1402 |
+
"g_norm": 0.000297811086056754,
|
| 1403 |
+
"shape": [
|
| 1404 |
+
192
|
| 1405 |
+
]
|
| 1406 |
+
},
|
| 1407 |
+
"dec_blocks.4.norm2.bias": {
|
| 1408 |
+
"w_norm": 0.03246624395251274,
|
| 1409 |
+
"g_norm": 0.0002632128889672458,
|
| 1410 |
+
"shape": [
|
| 1411 |
+
192
|
| 1412 |
+
]
|
| 1413 |
+
},
|
| 1414 |
+
"dec_blocks.4.conv2.weight": {
|
| 1415 |
+
"w_norm": 8.034357070922852,
|
| 1416 |
+
"g_norm": 0.010364442132413387,
|
| 1417 |
+
"shape": [
|
| 1418 |
+
192,
|
| 1419 |
+
192,
|
| 1420 |
+
3,
|
| 1421 |
+
3
|
| 1422 |
+
]
|
| 1423 |
+
},
|
| 1424 |
+
"dec_blocks.4.conv2.bias": {
|
| 1425 |
+
"w_norm": 0.19388015568256378,
|
| 1426 |
+
"g_norm": 0.0004662639112211764,
|
| 1427 |
+
"shape": [
|
| 1428 |
+
192
|
| 1429 |
+
]
|
| 1430 |
+
},
|
| 1431 |
+
"dec_blocks.4.skip.weight": {
|
| 1432 |
+
"w_norm": 7.931014060974121,
|
| 1433 |
+
"g_norm": 0.013633654452860355,
|
| 1434 |
+
"shape": [
|
| 1435 |
+
192,
|
| 1436 |
+
576,
|
| 1437 |
+
1,
|
| 1438 |
+
1
|
| 1439 |
+
]
|
| 1440 |
+
},
|
| 1441 |
+
"dec_blocks.4.skip.bias": {
|
| 1442 |
+
"w_norm": 0.3468608558177948,
|
| 1443 |
+
"g_norm": 0.0004662639112211764,
|
| 1444 |
+
"shape": [
|
| 1445 |
+
192
|
| 1446 |
+
]
|
| 1447 |
+
},
|
| 1448 |
+
"dec_blocks.5.norm1.weight": {
|
| 1449 |
+
"w_norm": 19.574352264404297,
|
| 1450 |
+
"g_norm": 0.002302073873579502,
|
| 1451 |
+
"shape": [
|
| 1452 |
+
384
|
| 1453 |
+
]
|
| 1454 |
+
},
|
| 1455 |
+
"dec_blocks.5.norm1.bias": {
|
| 1456 |
+
"w_norm": 0.05974191799759865,
|
| 1457 |
+
"g_norm": 0.002887281822040677,
|
| 1458 |
+
"shape": [
|
| 1459 |
+
384
|
| 1460 |
+
]
|
| 1461 |
+
},
|
| 1462 |
+
"dec_blocks.5.conv1.weight": {
|
| 1463 |
+
"w_norm": 8.306647300720215,
|
| 1464 |
+
"g_norm": 0.05203191936016083,
|
| 1465 |
+
"shape": [
|
| 1466 |
+
192,
|
| 1467 |
+
384,
|
| 1468 |
+
3,
|
| 1469 |
+
3
|
| 1470 |
+
]
|
| 1471 |
+
},
|
| 1472 |
+
"dec_blocks.5.conv1.bias": {
|
| 1473 |
+
"w_norm": 0.13815189898014069,
|
| 1474 |
+
"g_norm": 0.00202044565230608,
|
| 1475 |
+
"shape": [
|
| 1476 |
+
192
|
| 1477 |
+
]
|
| 1478 |
+
},
|
| 1479 |
+
"dec_blocks.5.t_proj.weight": {
|
| 1480 |
+
"w_norm": 8.092968940734863,
|
| 1481 |
+
"g_norm": 0.000958150252699852,
|
| 1482 |
+
"shape": [
|
| 1483 |
+
192,
|
| 1484 |
+
256
|
| 1485 |
+
]
|
| 1486 |
+
},
|
| 1487 |
+
"dec_blocks.5.t_proj.bias": {
|
| 1488 |
+
"w_norm": 0.5047914981842041,
|
| 1489 |
+
"g_norm": 0.002020731568336487,
|
| 1490 |
+
"shape": [
|
| 1491 |
+
192
|
| 1492 |
+
]
|
| 1493 |
+
},
|
| 1494 |
+
"dec_blocks.5.norm2.weight": {
|
| 1495 |
+
"w_norm": 13.871760368347168,
|
| 1496 |
+
"g_norm": 0.0022596679627895355,
|
| 1497 |
+
"shape": [
|
| 1498 |
+
192
|
| 1499 |
+
]
|
| 1500 |
+
},
|
| 1501 |
+
"dec_blocks.5.norm2.bias": {
|
| 1502 |
+
"w_norm": 0.030323660001158714,
|
| 1503 |
+
"g_norm": 0.0025028937961906195,
|
| 1504 |
+
"shape": [
|
| 1505 |
+
192
|
| 1506 |
+
]
|
| 1507 |
+
},
|
| 1508 |
+
"dec_blocks.5.conv2.weight": {
|
| 1509 |
+
"w_norm": 8.181824684143066,
|
| 1510 |
+
"g_norm": 0.060832276940345764,
|
| 1511 |
+
"shape": [
|
| 1512 |
+
192,
|
| 1513 |
+
192,
|
| 1514 |
+
3,
|
| 1515 |
+
3
|
| 1516 |
+
]
|
| 1517 |
+
},
|
| 1518 |
+
"dec_blocks.5.conv2.bias": {
|
| 1519 |
+
"w_norm": 0.1982647031545639,
|
| 1520 |
+
"g_norm": 0.00286358711309731,
|
| 1521 |
+
"shape": [
|
| 1522 |
+
192
|
| 1523 |
+
]
|
| 1524 |
+
},
|
| 1525 |
+
"dec_blocks.5.skip.weight": {
|
| 1526 |
+
"w_norm": 8.064077377319336,
|
| 1527 |
+
"g_norm": 0.07279777526855469,
|
| 1528 |
+
"shape": [
|
| 1529 |
+
192,
|
| 1530 |
+
384,
|
| 1531 |
+
1,
|
| 1532 |
+
1
|
| 1533 |
+
]
|
| 1534 |
+
},
|
| 1535 |
+
"dec_blocks.5.skip.bias": {
|
| 1536 |
+
"w_norm": 0.42116665840148926,
|
| 1537 |
+
"g_norm": 0.00286358711309731,
|
| 1538 |
+
"shape": [
|
| 1539 |
+
192
|
| 1540 |
+
]
|
| 1541 |
+
},
|
| 1542 |
+
"dec_attns.0.norm.weight": {
|
| 1543 |
+
"w_norm": 27.669570922851562,
|
| 1544 |
+
"g_norm": 3.4397162380628288e-06,
|
| 1545 |
+
"shape": [
|
| 1546 |
+
768
|
| 1547 |
+
]
|
| 1548 |
+
},
|
| 1549 |
+
"dec_attns.0.norm.bias": {
|
| 1550 |
+
"w_norm": 0.024704044684767723,
|
| 1551 |
+
"g_norm": 4.100290880160173e-06,
|
| 1552 |
+
"shape": [
|
| 1553 |
+
768
|
| 1554 |
+
]
|
| 1555 |
+
},
|
| 1556 |
+
"dec_attns.0.qkv.weight": {
|
| 1557 |
+
"w_norm": 27.809728622436523,
|
| 1558 |
+
"g_norm": 0.00014507168089039624,
|
| 1559 |
+
"shape": [
|
| 1560 |
+
2304,
|
| 1561 |
+
768,
|
| 1562 |
+
1,
|
| 1563 |
+
1
|
| 1564 |
+
]
|
| 1565 |
+
},
|
| 1566 |
+
"dec_attns.0.qkv.bias": {
|
| 1567 |
+
"w_norm": 1.0090502500534058,
|
| 1568 |
+
"g_norm": 6.295892490015831e-06,
|
| 1569 |
+
"shape": [
|
| 1570 |
+
2304
|
| 1571 |
+
]
|
| 1572 |
+
},
|
| 1573 |
+
"dec_attns.0.proj.weight": {
|
| 1574 |
+
"w_norm": 16.01603126525879,
|
| 1575 |
+
"g_norm": 0.0001487144036218524,
|
| 1576 |
+
"shape": [
|
| 1577 |
+
768,
|
| 1578 |
+
768,
|
| 1579 |
+
1,
|
| 1580 |
+
1
|
| 1581 |
+
]
|
| 1582 |
+
},
|
| 1583 |
+
"dec_attns.0.proj.bias": {
|
| 1584 |
+
"w_norm": 0.5765091776847839,
|
| 1585 |
+
"g_norm": 1.0640424989105668e-05,
|
| 1586 |
+
"shape": [
|
| 1587 |
+
768
|
| 1588 |
+
]
|
| 1589 |
+
},
|
| 1590 |
+
"dec_attns.1.norm.weight": {
|
| 1591 |
+
"w_norm": 27.69176483154297,
|
| 1592 |
+
"g_norm": 9.369830877403729e-06,
|
| 1593 |
+
"shape": [
|
| 1594 |
+
768
|
| 1595 |
+
]
|
| 1596 |
+
},
|
| 1597 |
+
"dec_attns.1.norm.bias": {
|
| 1598 |
+
"w_norm": 0.03322061523795128,
|
| 1599 |
+
"g_norm": 1.1450497368059587e-05,
|
| 1600 |
+
"shape": [
|
| 1601 |
+
768
|
| 1602 |
+
]
|
| 1603 |
+
},
|
| 1604 |
+
"dec_attns.1.qkv.weight": {
|
| 1605 |
+
"w_norm": 27.846345901489258,
|
| 1606 |
+
"g_norm": 0.0003313145716674626,
|
| 1607 |
+
"shape": [
|
| 1608 |
+
2304,
|
| 1609 |
+
768,
|
| 1610 |
+
1,
|
| 1611 |
+
1
|
| 1612 |
+
]
|
| 1613 |
+
},
|
| 1614 |
+
"dec_attns.1.qkv.bias": {
|
| 1615 |
+
"w_norm": 1.0051014423370361,
|
| 1616 |
+
"g_norm": 1.5137110494833905e-05,
|
| 1617 |
+
"shape": [
|
| 1618 |
+
2304
|
| 1619 |
+
]
|
| 1620 |
+
},
|
| 1621 |
+
"dec_attns.1.proj.weight": {
|
| 1622 |
+
"w_norm": 16.0462703704834,
|
| 1623 |
+
"g_norm": 0.00032928746077232063,
|
| 1624 |
+
"shape": [
|
| 1625 |
+
768,
|
| 1626 |
+
768,
|
| 1627 |
+
1,
|
| 1628 |
+
1
|
| 1629 |
+
]
|
| 1630 |
+
},
|
| 1631 |
+
"dec_attns.1.proj.bias": {
|
| 1632 |
+
"w_norm": 0.5801531076431274,
|
| 1633 |
+
"g_norm": 2.3453705580323003e-05,
|
| 1634 |
+
"shape": [
|
| 1635 |
+
768
|
| 1636 |
+
]
|
| 1637 |
+
},
|
| 1638 |
+
"dec_attns.2.norm.weight": {
|
| 1639 |
+
"w_norm": 19.559301376342773,
|
| 1640 |
+
"g_norm": 1.526609958091285e-05,
|
| 1641 |
+
"shape": [
|
| 1642 |
+
384
|
| 1643 |
+
]
|
| 1644 |
+
},
|
| 1645 |
+
"dec_attns.2.norm.bias": {
|
| 1646 |
+
"w_norm": 0.01634746603667736,
|
| 1647 |
+
"g_norm": 1.7435884728911333e-05,
|
| 1648 |
+
"shape": [
|
| 1649 |
+
384
|
| 1650 |
+
]
|
| 1651 |
+
},
|
| 1652 |
+
"dec_attns.2.qkv.weight": {
|
| 1653 |
+
"w_norm": 19.649356842041016,
|
| 1654 |
+
"g_norm": 0.0005132962833158672,
|
| 1655 |
+
"shape": [
|
| 1656 |
+
1152,
|
| 1657 |
+
384,
|
| 1658 |
+
1,
|
| 1659 |
+
1
|
| 1660 |
+
]
|
| 1661 |
+
},
|
| 1662 |
+
"dec_attns.2.qkv.bias": {
|
| 1663 |
+
"w_norm": 0.987482488155365,
|
| 1664 |
+
"g_norm": 2.9351922421483323e-05,
|
| 1665 |
+
"shape": [
|
| 1666 |
+
1152
|
| 1667 |
+
]
|
| 1668 |
+
},
|
| 1669 |
+
"dec_attns.2.proj.weight": {
|
| 1670 |
+
"w_norm": 11.306385040283203,
|
| 1671 |
+
"g_norm": 0.0005538715049624443,
|
| 1672 |
+
"shape": [
|
| 1673 |
+
384,
|
| 1674 |
+
384,
|
| 1675 |
+
1,
|
| 1676 |
+
1
|
| 1677 |
+
]
|
| 1678 |
+
},
|
| 1679 |
+
"dec_attns.2.proj.bias": {
|
| 1680 |
+
"w_norm": 0.5871843695640564,
|
| 1681 |
+
"g_norm": 5.271810368867591e-05,
|
| 1682 |
+
"shape": [
|
| 1683 |
+
384
|
| 1684 |
+
]
|
| 1685 |
+
},
|
| 1686 |
+
"dec_ups.0.op.weight": {
|
| 1687 |
+
"w_norm": 16.088212966918945,
|
| 1688 |
+
"g_norm": 0.0013465607771649957,
|
| 1689 |
+
"shape": [
|
| 1690 |
+
768,
|
| 1691 |
+
768,
|
| 1692 |
+
3,
|
| 1693 |
+
3
|
| 1694 |
+
]
|
| 1695 |
+
},
|
| 1696 |
+
"dec_ups.0.op.bias": {
|
| 1697 |
+
"w_norm": 0.2013898640871048,
|
| 1698 |
+
"g_norm": 1.0867960554605816e-05,
|
| 1699 |
+
"shape": [
|
| 1700 |
+
768
|
| 1701 |
+
]
|
| 1702 |
+
},
|
| 1703 |
+
"dec_ups.1.op.weight": {
|
| 1704 |
+
"w_norm": 16.08967399597168,
|
| 1705 |
+
"g_norm": 0.002522860188037157,
|
| 1706 |
+
"shape": [
|
| 1707 |
+
768,
|
| 1708 |
+
768,
|
| 1709 |
+
3,
|
| 1710 |
+
3
|
| 1711 |
+
]
|
| 1712 |
+
},
|
| 1713 |
+
"dec_ups.1.op.bias": {
|
| 1714 |
+
"w_norm": 0.1886485069990158,
|
| 1715 |
+
"g_norm": 2.7130645321449265e-05,
|
| 1716 |
+
"shape": [
|
| 1717 |
+
768
|
| 1718 |
+
]
|
| 1719 |
+
},
|
| 1720 |
+
"dec_ups.2.op.weight": {
|
| 1721 |
+
"w_norm": 11.30615520477295,
|
| 1722 |
+
"g_norm": 0.006385178305208683,
|
| 1723 |
+
"shape": [
|
| 1724 |
+
384,
|
| 1725 |
+
384,
|
| 1726 |
+
3,
|
| 1727 |
+
3
|
| 1728 |
+
]
|
| 1729 |
+
},
|
| 1730 |
+
"dec_ups.2.op.bias": {
|
| 1731 |
+
"w_norm": 0.19624017179012299,
|
| 1732 |
+
"g_norm": 9.016467083711177e-05,
|
| 1733 |
+
"shape": [
|
| 1734 |
+
384
|
| 1735 |
+
]
|
| 1736 |
+
},
|
| 1737 |
+
"dec_ups.3.op.weight": {
|
| 1738 |
+
"w_norm": 11.316301345825195,
|
| 1739 |
+
"g_norm": 0.014089048840105534,
|
| 1740 |
+
"shape": [
|
| 1741 |
+
384,
|
| 1742 |
+
384,
|
| 1743 |
+
3,
|
| 1744 |
+
3
|
| 1745 |
+
]
|
| 1746 |
+
},
|
| 1747 |
+
"dec_ups.3.op.bias": {
|
| 1748 |
+
"w_norm": 0.19712364673614502,
|
| 1749 |
+
"g_norm": 0.00023156803217716515,
|
| 1750 |
+
"shape": [
|
| 1751 |
+
384
|
| 1752 |
+
]
|
| 1753 |
+
},
|
| 1754 |
+
"dec_ups.4.op.weight": {
|
| 1755 |
+
"w_norm": 7.999128818511963,
|
| 1756 |
+
"g_norm": 0.03885931521654129,
|
| 1757 |
+
"shape": [
|
| 1758 |
+
192,
|
| 1759 |
+
192,
|
| 1760 |
+
3,
|
| 1761 |
+
3
|
| 1762 |
+
]
|
| 1763 |
+
},
|
| 1764 |
+
"dec_ups.4.op.bias": {
|
| 1765 |
+
"w_norm": 0.19474607706069946,
|
| 1766 |
+
"g_norm": 0.0006655824836343527,
|
| 1767 |
+
"shape": [
|
| 1768 |
+
192
|
| 1769 |
+
]
|
| 1770 |
+
},
|
| 1771 |
+
"out_norm.weight": {
|
| 1772 |
+
"w_norm": 13.822661399841309,
|
| 1773 |
+
"g_norm": 0.005152558907866478,
|
| 1774 |
+
"shape": [
|
| 1775 |
+
192
|
| 1776 |
+
]
|
| 1777 |
+
},
|
| 1778 |
+
"out_norm.bias": {
|
| 1779 |
+
"w_norm": 0.04736120253801346,
|
| 1780 |
+
"g_norm": 0.008335559628903866,
|
| 1781 |
+
"shape": [
|
| 1782 |
+
192
|
| 1783 |
+
]
|
| 1784 |
+
},
|
| 1785 |
+
"out_conv.weight": {
|
| 1786 |
+
"w_norm": 0.11635566502809525,
|
| 1787 |
+
"g_norm": 0.9671086668968201,
|
| 1788 |
+
"shape": [
|
| 1789 |
+
3,
|
| 1790 |
+
192,
|
| 1791 |
+
3,
|
| 1792 |
+
3
|
| 1793 |
+
]
|
| 1794 |
+
},
|
| 1795 |
+
"out_conv.bias": {
|
| 1796 |
+
"w_norm": 0.007038481067866087,
|
| 1797 |
+
"g_norm": 0.06203581020236015,
|
| 1798 |
+
"shape": [
|
| 1799 |
+
3
|
| 1800 |
+
]
|
| 1801 |
+
}
|
| 1802 |
+
},
|
| 1803 |
+
"activations": {
|
| 1804 |
+
"input_conv": {
|
| 1805 |
+
"mean": 0.0005878935335204005,
|
| 1806 |
+
"std": 0.3478555381298065,
|
| 1807 |
+
"abs_max": 2.53125,
|
| 1808 |
+
"frac_dead": 0.0005414850311353803,
|
| 1809 |
+
"frac_sat": 0.0
|
| 1810 |
+
},
|
| 1811 |
+
"time_mlp.1": {
|
| 1812 |
+
"mean": 0.02260744199156761,
|
| 1813 |
+
"std": 0.27119144797325134,
|
| 1814 |
+
"abs_max": 3.0625,
|
| 1815 |
+
"frac_dead": 0.0,
|
| 1816 |
+
"frac_sat": 0.0
|
| 1817 |
+
},
|
| 1818 |
+
"time_mlp.3": {
|
| 1819 |
+
"mean": 0.019019395112991333,
|
| 1820 |
+
"std": 0.13746215403079987,
|
| 1821 |
+
"abs_max": 0.59765625,
|
| 1822 |
+
"frac_dead": 0.00048828125,
|
| 1823 |
+
"frac_sat": 0.0
|
| 1824 |
+
},
|
| 1825 |
+
"enc_blocks.0.norm1": {
|
| 1826 |
+
"mean": -0.0034768523182719946,
|
| 1827 |
+
"std": 0.9972635507583618,
|
| 1828 |
+
"abs_max": 6.692741870880127,
|
| 1829 |
+
"frac_dead": 9.437401899958786e-07,
|
| 1830 |
+
"frac_sat": 0.0
|
| 1831 |
+
},
|
| 1832 |
+
"enc_blocks.0.conv1": {
|
| 1833 |
+
"mean": -0.03561967611312866,
|
| 1834 |
+
"std": 0.5352197885513306,
|
| 1835 |
+
"abs_max": 7.875,
|
| 1836 |
+
"frac_dead": 6.592604768229648e-05,
|
| 1837 |
+
"frac_sat": 0.0
|
| 1838 |
+
},
|
| 1839 |
+
"enc_blocks.0.t_proj": {
|
| 1840 |
+
"mean": -0.0022988717537373304,
|
| 1841 |
+
"std": 0.10281845182180405,
|
| 1842 |
+
"abs_max": 0.71484375,
|
| 1843 |
+
"frac_dead": 0.0,
|
| 1844 |
+
"frac_sat": 0.0
|
| 1845 |
+
},
|
| 1846 |
+
"enc_blocks.0.norm2": {
|
| 1847 |
+
"mean": -0.0019118194468319416,
|
| 1848 |
+
"std": 0.9995514750480652,
|
| 1849 |
+
"abs_max": 10.405152320861816,
|
| 1850 |
+
"frac_dead": 1.1788474694185425e-06,
|
| 1851 |
+
"frac_sat": 6.622738357719982e-09
|
| 1852 |
+
},
|
| 1853 |
+
"enc_blocks.0.conv2": {
|
| 1854 |
+
"mean": -0.009182874113321304,
|
| 1855 |
+
"std": 0.8037123680114746,
|
| 1856 |
+
"abs_max": 7.125,
|
| 1857 |
+
"frac_dead": 3.254082548664883e-05,
|
| 1858 |
+
"frac_sat": 0.0
|
| 1859 |
+
},
|
| 1860 |
+
"enc_downs.0.op": {
|
| 1861 |
+
"mean": -0.1348954737186432,
|
| 1862 |
+
"std": 1.5058555603027344,
|
| 1863 |
+
"abs_max": 10.5625,
|
| 1864 |
+
"frac_dead": 1.3192494407121558e-05,
|
| 1865 |
+
"frac_sat": 1.3245476715439963e-08
|
| 1866 |
+
},
|
| 1867 |
+
"enc_blocks.1.norm1": {
|
| 1868 |
+
"mean": -0.0008627756033092737,
|
| 1869 |
+
"std": 0.9998546838760376,
|
| 1870 |
+
"abs_max": 6.996696472167969,
|
| 1871 |
+
"frac_dead": 2.914004824106087e-07,
|
| 1872 |
+
"frac_sat": 0.0
|
| 1873 |
+
},
|
| 1874 |
+
"enc_blocks.1.conv1": {
|
| 1875 |
+
"mean": -0.21788886189460754,
|
| 1876 |
+
"std": 1.1535621881484985,
|
| 1877 |
+
"abs_max": 7.21875,
|
| 1878 |
+
"frac_dead": 2.4888251573429443e-05,
|
| 1879 |
+
"frac_sat": 0.0
|
| 1880 |
+
},
|
| 1881 |
+
"enc_blocks.1.t_proj": {
|
| 1882 |
+
"mean": 0.0034813284873962402,
|
| 1883 |
+
"std": 0.06453254073858261,
|
| 1884 |
+
"abs_max": 0.322265625,
|
| 1885 |
+
"frac_dead": 0.0,
|
| 1886 |
+
"frac_sat": 0.0
|
| 1887 |
+
},
|
| 1888 |
+
"enc_blocks.1.norm2": {
|
| 1889 |
+
"mean": -0.0024751583114266396,
|
| 1890 |
+
"std": 0.9978860020637512,
|
| 1891 |
+
"abs_max": 7.0123209953308105,
|
| 1892 |
+
"frac_dead": 5.563099989558395e-07,
|
| 1893 |
+
"frac_sat": 0.0
|
| 1894 |
+
},
|
| 1895 |
+
"enc_blocks.1.conv2": {
|
| 1896 |
+
"mean": 0.009306020103394985,
|
| 1897 |
+
"std": 0.946929931640625,
|
| 1898 |
+
"abs_max": 7.53125,
|
| 1899 |
+
"frac_dead": 4.3299463868606836e-05,
|
| 1900 |
+
"frac_sat": 0.0
|
| 1901 |
+
},
|
| 1902 |
+
"enc_downs.1.op": {
|
| 1903 |
+
"mean": -0.30516207218170166,
|
| 1904 |
+
"std": 2.371323585510254,
|
| 1905 |
+
"abs_max": 13.0625,
|
| 1906 |
+
"frac_dead": 1.1973910659435205e-05,
|
| 1907 |
+
"frac_sat": 6.0664282500511035e-05
|
| 1908 |
+
},
|
| 1909 |
+
"enc_blocks.2.norm1": {
|
| 1910 |
+
"mean": -0.0008903791313059628,
|
| 1911 |
+
"std": 0.9997683763504028,
|
| 1912 |
+
"abs_max": 6.303738117218018,
|
| 1913 |
+
"frac_dead": 1.271565793103946e-06,
|
| 1914 |
+
"frac_sat": 0.0
|
| 1915 |
+
},
|
| 1916 |
+
"enc_blocks.2.conv1": {
|
| 1917 |
+
"mean": -0.265188604593277,
|
| 1918 |
+
"std": 0.748349130153656,
|
| 1919 |
+
"abs_max": 6.1875,
|
| 1920 |
+
"frac_dead": 4.259745401213877e-05,
|
| 1921 |
+
"frac_sat": 0.0
|
| 1922 |
+
},
|
| 1923 |
+
"enc_blocks.2.t_proj": {
|
| 1924 |
+
"mean": -0.0014002572279423475,
|
| 1925 |
+
"std": 0.07900581508874893,
|
| 1926 |
+
"abs_max": 0.53125,
|
| 1927 |
+
"frac_dead": 0.00010850694525288418,
|
| 1928 |
+
"frac_sat": 0.0
|
| 1929 |
+
},
|
| 1930 |
+
"enc_blocks.2.norm2": {
|
| 1931 |
+
"mean": -0.0040403506718575954,
|
| 1932 |
+
"std": 0.9962407946586609,
|
| 1933 |
+
"abs_max": 6.904444694519043,
|
| 1934 |
+
"frac_dead": 6.092919306865952e-07,
|
| 1935 |
+
"frac_sat": 0.0
|
| 1936 |
+
},
|
| 1937 |
+
"enc_blocks.2.conv2": {
|
| 1938 |
+
"mean": 0.13902650773525238,
|
| 1939 |
+
"std": 0.9115206599235535,
|
| 1940 |
+
"abs_max": 9.6875,
|
| 1941 |
+
"frac_dead": 2.763006523309741e-05,
|
| 1942 |
+
"frac_sat": 0.0
|
| 1943 |
+
},
|
| 1944 |
+
"enc_blocks.2.skip": {
|
| 1945 |
+
"mean": 0.06025930121541023,
|
| 1946 |
+
"std": 1.4612077474594116,
|
| 1947 |
+
"abs_max": 11.1875,
|
| 1948 |
+
"frac_dead": 5.809466165374033e-05,
|
| 1949 |
+
"frac_sat": 1.8543667579251633e-07
|
| 1950 |
+
},
|
| 1951 |
+
"enc_downs.2.op": {
|
| 1952 |
+
"mean": 0.23879824578762054,
|
| 1953 |
+
"std": 2.575745105743408,
|
| 1954 |
+
"abs_max": 22.625,
|
| 1955 |
+
"frac_dead": 9.43077975534834e-06,
|
| 1956 |
+
"frac_sat": 0.0032885868567973375
|
| 1957 |
+
},
|
| 1958 |
+
"enc_blocks.3.norm1": {
|
| 1959 |
+
"mean": 0.0005088050966151059,
|
| 1960 |
+
"std": 1.0007959604263306,
|
| 1961 |
+
"abs_max": 8.825421333312988,
|
| 1962 |
+
"frac_dead": 1.8013848830378265e-06,
|
| 1963 |
+
"frac_sat": 0.0
|
| 1964 |
+
},
|
| 1965 |
+
"enc_blocks.3.conv1": {
|
| 1966 |
+
"mean": -0.1672450304031372,
|
| 1967 |
+
"std": 1.5013314485549927,
|
| 1968 |
+
"abs_max": 22.5,
|
| 1969 |
+
"frac_dead": 1.896752291941084e-05,
|
| 1970 |
+
"frac_sat": 0.00022898780298419297
|
| 1971 |
+
},
|
| 1972 |
+
"enc_blocks.3.t_proj": {
|
| 1973 |
+
"mean": -0.0006834938540123403,
|
| 1974 |
+
"std": 0.057969387620687485,
|
| 1975 |
+
"abs_max": 0.2578125,
|
| 1976 |
+
"frac_dead": 0.0,
|
| 1977 |
+
"frac_sat": 0.0
|
| 1978 |
+
},
|
| 1979 |
+
"enc_blocks.3.norm2": {
|
| 1980 |
+
"mean": -0.002198457717895508,
|
| 1981 |
+
"std": 0.997900128364563,
|
| 1982 |
+
"abs_max": 12.679757118225098,
|
| 1983 |
+
"frac_dead": 5.298190899338806e-07,
|
| 1984 |
+
"frac_sat": 3.0729506761417724e-06
|
| 1985 |
+
},
|
| 1986 |
+
"enc_blocks.3.conv2": {
|
| 1987 |
+
"mean": -0.07520721107721329,
|
| 1988 |
+
"std": 1.4643439054489136,
|
| 1989 |
+
"abs_max": 17.75,
|
| 1990 |
+
"frac_dead": 2.0451016098377295e-05,
|
| 1991 |
+
"frac_sat": 0.00010522206866880879
|
| 1992 |
+
},
|
| 1993 |
+
"enc_attns.3.norm": {
|
| 1994 |
+
"mean": 0.00023528767633251846,
|
| 1995 |
+
"std": 0.999700129032135,
|
| 1996 |
+
"abs_max": 9.543044090270996,
|
| 1997 |
+
"frac_dead": 1.059638137235197e-07,
|
| 1998 |
+
"frac_sat": 0.0
|
| 1999 |
+
},
|
| 2000 |
+
"enc_attns.3.qkv": {
|
| 2001 |
+
"mean": -0.003681996837258339,
|
| 2002 |
+
"std": 0.7336653470993042,
|
| 2003 |
+
"abs_max": 11.8125,
|
| 2004 |
+
"frac_dead": 9.278897778131068e-05,
|
| 2005 |
+
"frac_sat": 1.7660636331129353e-07
|
| 2006 |
+
},
|
| 2007 |
+
"enc_attns.3.proj": {
|
| 2008 |
+
"mean": -0.004405897576361895,
|
| 2009 |
+
"std": 0.3329181671142578,
|
| 2010 |
+
"abs_max": 2.453125,
|
| 2011 |
+
"frac_dead": 0.00024106768250931054,
|
| 2012 |
+
"frac_sat": 0.0
|
| 2013 |
+
},
|
| 2014 |
+
"enc_downs.3.op": {
|
| 2015 |
+
"mean": 0.5550792813301086,
|
| 2016 |
+
"std": 3.9987339973449707,
|
| 2017 |
+
"abs_max": 30.375,
|
| 2018 |
+
"frac_dead": 1.0172526344831567e-05,
|
| 2019 |
+
"frac_sat": 0.02196587435901165
|
| 2020 |
+
},
|
| 2021 |
+
"enc_blocks.4.norm1": {
|
| 2022 |
+
"mean": -0.00024249155831057578,
|
| 2023 |
+
"std": 0.9996702671051025,
|
| 2024 |
+
"abs_max": 8.92320728302002,
|
| 2025 |
+
"frac_dead": 1.271565793103946e-06,
|
| 2026 |
+
"frac_sat": 0.0
|
| 2027 |
+
},
|
| 2028 |
+
"enc_blocks.4.conv1": {
|
| 2029 |
+
"mean": -0.22923271358013153,
|
| 2030 |
+
"std": 1.21475088596344,
|
| 2031 |
+
"abs_max": 14.1875,
|
| 2032 |
+
"frac_dead": 2.3100112230167724e-05,
|
| 2033 |
+
"frac_sat": 6.167093670228496e-05
|
| 2034 |
+
},
|
| 2035 |
+
"enc_blocks.4.t_proj": {
|
| 2036 |
+
"mean": -0.0025398412253707647,
|
| 2037 |
+
"std": 0.058452896773815155,
|
| 2038 |
+
"abs_max": 0.322265625,
|
| 2039 |
+
"frac_dead": 0.0,
|
| 2040 |
+
"frac_sat": 0.0
|
| 2041 |
+
},
|
| 2042 |
+
"enc_blocks.4.norm2": {
|
| 2043 |
+
"mean": -0.0034613648895174265,
|
| 2044 |
+
"std": 0.9967631101608276,
|
| 2045 |
+
"abs_max": 10.875402450561523,
|
| 2046 |
+
"frac_dead": 1.271565793103946e-06,
|
| 2047 |
+
"frac_sat": 2.755059085757239e-06
|
| 2048 |
+
},
|
| 2049 |
+
"enc_blocks.4.conv2": {
|
| 2050 |
+
"mean": 0.05035363510251045,
|
| 2051 |
+
"std": 1.1800023317337036,
|
| 2052 |
+
"abs_max": 8.6875,
|
| 2053 |
+
"frac_dead": 1.4623005881730933e-05,
|
| 2054 |
+
"frac_sat": 0.0
|
| 2055 |
+
},
|
| 2056 |
+
"enc_blocks.4.skip": {
|
| 2057 |
+
"mean": 0.18852955102920532,
|
| 2058 |
+
"std": 2.690230131149292,
|
| 2059 |
+
"abs_max": 21.875,
|
| 2060 |
+
"frac_dead": 2.7338663130649365e-05,
|
| 2061 |
+
"frac_sat": 0.0027665032539516687
|
| 2062 |
+
},
|
| 2063 |
+
"enc_attns.4.norm": {
|
| 2064 |
+
"mean": 5.994405910314526e-06,
|
| 2065 |
+
"std": 0.9989033341407776,
|
| 2066 |
+
"abs_max": 9.629432678222656,
|
| 2067 |
+
"frac_dead": 1.4834934063401306e-06,
|
| 2068 |
+
"frac_sat": 0.0
|
| 2069 |
+
},
|
| 2070 |
+
"enc_attns.4.qkv": {
|
| 2071 |
+
"mean": -0.003762299893423915,
|
| 2072 |
+
"std": 0.8487932682037354,
|
| 2073 |
+
"abs_max": 9.25,
|
| 2074 |
+
"frac_dead": 6.668655987596139e-05,
|
| 2075 |
+
"frac_sat": 0.0
|
| 2076 |
+
},
|
| 2077 |
+
"enc_attns.4.proj": {
|
| 2078 |
+
"mean": 0.00023298907035496086,
|
| 2079 |
+
"std": 0.16651789844036102,
|
| 2080 |
+
"abs_max": 1.375,
|
| 2081 |
+
"frac_dead": 0.0002608829236123711,
|
| 2082 |
+
"frac_sat": 0.0
|
| 2083 |
+
},
|
| 2084 |
+
"enc_downs.4.op": {
|
| 2085 |
+
"mean": -0.052241500467061996,
|
| 2086 |
+
"std": 4.246338367462158,
|
| 2087 |
+
"abs_max": 30.375,
|
| 2088 |
+
"frac_dead": 5.086263172415784e-06,
|
| 2089 |
+
"frac_sat": 0.03183576837182045
|
| 2090 |
+
},
|
| 2091 |
+
"enc_blocks.5.norm1": {
|
| 2092 |
+
"mean": 0.0003438827989157289,
|
| 2093 |
+
"std": 1.0007237195968628,
|
| 2094 |
+
"abs_max": 10.009206771850586,
|
| 2095 |
+
"frac_dead": 0.0,
|
| 2096 |
+
"frac_sat": 8.477105097881577e-07
|
| 2097 |
+
},
|
| 2098 |
+
"enc_blocks.5.conv1": {
|
| 2099 |
+
"mean": -0.3997481167316437,
|
| 2100 |
+
"std": 1.9342615604400635,
|
| 2101 |
+
"abs_max": 14.6875,
|
| 2102 |
+
"frac_dead": 1.0172526344831567e-05,
|
| 2103 |
+
"frac_sat": 0.00036282010842114687
|
| 2104 |
+
},
|
| 2105 |
+
"enc_blocks.5.t_proj": {
|
| 2106 |
+
"mean": -0.002505121985450387,
|
| 2107 |
+
"std": 0.057327546179294586,
|
| 2108 |
+
"abs_max": 0.333984375,
|
| 2109 |
+
"frac_dead": 0.00010850694525288418,
|
| 2110 |
+
"frac_sat": 0.0
|
| 2111 |
+
},
|
| 2112 |
+
"enc_blocks.5.norm2": {
|
| 2113 |
+
"mean": -0.0028387042693793774,
|
| 2114 |
+
"std": 0.9975071549415588,
|
| 2115 |
+
"abs_max": 7.06866455078125,
|
| 2116 |
+
"frac_dead": 1.6954210195763153e-06,
|
| 2117 |
+
"frac_sat": 0.0
|
| 2118 |
+
},
|
| 2119 |
+
"enc_blocks.5.conv2": {
|
| 2120 |
+
"mean": 0.09301067143678665,
|
| 2121 |
+
"std": 1.1804240942001343,
|
| 2122 |
+
"abs_max": 6.21875,
|
| 2123 |
+
"frac_dead": 1.610649997019209e-05,
|
| 2124 |
+
"frac_sat": 0.0
|
| 2125 |
+
},
|
| 2126 |
+
"enc_attns.5.norm": {
|
| 2127 |
+
"mean": -4.0283219277625903e-05,
|
| 2128 |
+
"std": 0.9991267919540405,
|
| 2129 |
+
"abs_max": 10.267005920410156,
|
| 2130 |
+
"frac_dead": 2.543131586207892e-06,
|
| 2131 |
+
"frac_sat": 1.6954210195763153e-06
|
| 2132 |
+
},
|
| 2133 |
+
"enc_attns.5.qkv": {
|
| 2134 |
+
"mean": -0.004235548432916403,
|
| 2135 |
+
"std": 0.7168772220611572,
|
| 2136 |
+
"abs_max": 6.96875,
|
| 2137 |
+
"frac_dead": 7.62939453125e-05,
|
| 2138 |
+
"frac_sat": 0.0
|
| 2139 |
+
},
|
| 2140 |
+
"enc_attns.5.proj": {
|
| 2141 |
+
"mean": 0.003645795863121748,
|
| 2142 |
+
"std": 0.28655537962913513,
|
| 2143 |
+
"abs_max": 1.3828125,
|
| 2144 |
+
"frac_dead": 0.00013648139429278672,
|
| 2145 |
+
"frac_sat": 0.0
|
| 2146 |
+
},
|
| 2147 |
+
"mid1.norm1": {
|
| 2148 |
+
"mean": 0.00039271556306630373,
|
| 2149 |
+
"std": 1.0009983777999878,
|
| 2150 |
+
"abs_max": 10.273219108581543,
|
| 2151 |
+
"frac_dead": 0.0,
|
| 2152 |
+
"frac_sat": 1.6954210195763153e-06
|
| 2153 |
+
},
|
| 2154 |
+
"mid1.conv1": {
|
| 2155 |
+
"mean": -0.6475042700767517,
|
| 2156 |
+
"std": 1.9125096797943115,
|
| 2157 |
+
"abs_max": 15.5,
|
| 2158 |
+
"frac_dead": 8.47710543894209e-06,
|
| 2159 |
+
"frac_sat": 0.00037807889748364687
|
| 2160 |
+
},
|
| 2161 |
+
"mid1.t_proj": {
|
| 2162 |
+
"mean": 0.0028974239248782396,
|
| 2163 |
+
"std": 0.05917719379067421,
|
| 2164 |
+
"abs_max": 0.2890625,
|
| 2165 |
+
"frac_dead": 0.0,
|
| 2166 |
+
"frac_sat": 0.0
|
| 2167 |
+
},
|
| 2168 |
+
"mid1.norm2": {
|
| 2169 |
+
"mean": -0.002513372339308262,
|
| 2170 |
+
"std": 0.9980164170265198,
|
| 2171 |
+
"abs_max": 8.224201202392578,
|
| 2172 |
+
"frac_dead": 2.543131586207892e-06,
|
| 2173 |
+
"frac_sat": 0.0
|
| 2174 |
+
},
|
| 2175 |
+
"mid1.conv2": {
|
| 2176 |
+
"mean": 0.044998347759246826,
|
| 2177 |
+
"std": 1.407365322113037,
|
| 2178 |
+
"abs_max": 8.0,
|
| 2179 |
+
"frac_dead": 1.1867947250721045e-05,
|
| 2180 |
+
"frac_sat": 0.0
|
| 2181 |
+
},
|
| 2182 |
+
"mid_attn.norm": {
|
| 2183 |
+
"mean": -3.3670828997856006e-05,
|
| 2184 |
+
"std": 0.9990683794021606,
|
| 2185 |
+
"abs_max": 8.834624290466309,
|
| 2186 |
+
"frac_dead": 1.6954210195763153e-06,
|
| 2187 |
+
"frac_sat": 0.0
|
| 2188 |
+
},
|
| 2189 |
+
"mid_attn.qkv": {
|
| 2190 |
+
"mean": 0.013007380068302155,
|
| 2191 |
+
"std": 0.6966890692710876,
|
| 2192 |
+
"abs_max": 7.1875,
|
| 2193 |
+
"frac_dead": 7.007740350672975e-05,
|
| 2194 |
+
"frac_sat": 0.0
|
| 2195 |
+
},
|
| 2196 |
+
"mid_attn.proj": {
|
| 2197 |
+
"mean": 0.0007060468196868896,
|
| 2198 |
+
"std": 0.3016098439693451,
|
| 2199 |
+
"abs_max": 1.3125,
|
| 2200 |
+
"frac_dead": 0.00014326308155432343,
|
| 2201 |
+
"frac_sat": 0.0
|
| 2202 |
+
},
|
| 2203 |
+
"mid2.norm1": {
|
| 2204 |
+
"mean": 0.0003113076090812683,
|
| 2205 |
+
"std": 1.0007456541061401,
|
| 2206 |
+
"abs_max": 8.782316207885742,
|
| 2207 |
+
"frac_dead": 1.6954210195763153e-06,
|
| 2208 |
+
"frac_sat": 0.0
|
| 2209 |
+
},
|
| 2210 |
+
"mid2.conv1": {
|
| 2211 |
+
"mean": -0.5544661283493042,
|
| 2212 |
+
"std": 1.6419425010681152,
|
| 2213 |
+
"abs_max": 14.875,
|
| 2214 |
+
"frac_dead": 1.610649997019209e-05,
|
| 2215 |
+
"frac_sat": 0.00017886691784951836
|
| 2216 |
+
},
|
| 2217 |
+
"mid2.t_proj": {
|
| 2218 |
+
"mean": 0.0005331606371328235,
|
| 2219 |
+
"std": 0.059308089315891266,
|
| 2220 |
+
"abs_max": 0.341796875,
|
| 2221 |
+
"frac_dead": 0.0,
|
| 2222 |
+
"frac_sat": 0.0
|
| 2223 |
+
},
|
| 2224 |
+
"mid2.norm2": {
|
| 2225 |
+
"mean": -0.003296496579423547,
|
| 2226 |
+
"std": 0.9975892305374146,
|
| 2227 |
+
"abs_max": 8.295001983642578,
|
| 2228 |
+
"frac_dead": 0.0,
|
| 2229 |
+
"frac_sat": 0.0
|
| 2230 |
+
},
|
| 2231 |
+
"mid2.conv2": {
|
| 2232 |
+
"mean": 0.06585688143968582,
|
| 2233 |
+
"std": 1.362164855003357,
|
| 2234 |
+
"abs_max": 8.1875,
|
| 2235 |
+
"frac_dead": 1.4411079064302612e-05,
|
| 2236 |
+
"frac_sat": 0.0
|
| 2237 |
+
},
|
| 2238 |
+
"dec_blocks.0.norm1": {
|
| 2239 |
+
"mean": 0.00014007091522216797,
|
| 2240 |
+
"std": 1.0002217292785645,
|
| 2241 |
+
"abs_max": 10.676835060119629,
|
| 2242 |
+
"frac_dead": 4.238552548940788e-07,
|
| 2243 |
+
"frac_sat": 1.6954210195763153e-06
|
| 2244 |
+
},
|
| 2245 |
+
"dec_blocks.0.conv1": {
|
| 2246 |
+
"mean": -1.6476129293441772,
|
| 2247 |
+
"std": 2.691973924636841,
|
| 2248 |
+
"abs_max": 32.25,
|
| 2249 |
+
"frac_dead": 1.1867947250721045e-05,
|
| 2250 |
+
"frac_sat": 0.02121988870203495
|
| 2251 |
+
},
|
| 2252 |
+
"dec_blocks.0.t_proj": {
|
| 2253 |
+
"mean": -0.001999007537961006,
|
| 2254 |
+
"std": 0.06088656559586525,
|
| 2255 |
+
"abs_max": 0.2734375,
|
| 2256 |
+
"frac_dead": 0.0,
|
| 2257 |
+
"frac_sat": 0.0
|
| 2258 |
+
},
|
| 2259 |
+
"dec_blocks.0.norm2": {
|
| 2260 |
+
"mean": -0.00483663147315383,
|
| 2261 |
+
"std": 0.9973129630088806,
|
| 2262 |
+
"abs_max": 10.574625968933105,
|
| 2263 |
+
"frac_dead": 0.0,
|
| 2264 |
+
"frac_sat": 4.238552719471045e-06
|
| 2265 |
+
},
|
| 2266 |
+
"dec_blocks.0.conv2": {
|
| 2267 |
+
"mean": -0.00604268629103899,
|
| 2268 |
+
"std": 0.5441529154777527,
|
| 2269 |
+
"abs_max": 2.578125,
|
| 2270 |
+
"frac_dead": 3.560383993317373e-05,
|
| 2271 |
+
"frac_sat": 0.0
|
| 2272 |
+
},
|
| 2273 |
+
"dec_blocks.0.skip": {
|
| 2274 |
+
"mean": -0.032212819904088974,
|
| 2275 |
+
"std": 1.982886552810669,
|
| 2276 |
+
"abs_max": 13.25,
|
| 2277 |
+
"frac_dead": 2.1192763597355224e-05,
|
| 2278 |
+
"frac_sat": 0.00012122260523028672
|
| 2279 |
+
},
|
| 2280 |
+
"dec_attns.0.norm": {
|
| 2281 |
+
"mean": -0.00016459905600640923,
|
| 2282 |
+
"std": 0.9984184503555298,
|
| 2283 |
+
"abs_max": 8.354228973388672,
|
| 2284 |
+
"frac_dead": 8.477105097881577e-07,
|
| 2285 |
+
"frac_sat": 0.0
|
| 2286 |
+
},
|
| 2287 |
+
"dec_attns.0.qkv": {
|
| 2288 |
+
"mean": -0.009556476026773453,
|
| 2289 |
+
"std": 0.5968071222305298,
|
| 2290 |
+
"abs_max": 7.625,
|
| 2291 |
+
"frac_dead": 8.67490452947095e-05,
|
| 2292 |
+
"frac_sat": 0.0
|
| 2293 |
+
},
|
| 2294 |
+
"dec_attns.0.proj": {
|
| 2295 |
+
"mean": 0.008083020336925983,
|
| 2296 |
+
"std": 0.35218313336372375,
|
| 2297 |
+
"abs_max": 1.3125,
|
| 2298 |
+
"frac_dead": 0.0001237657415913418,
|
| 2299 |
+
"frac_sat": 0.0
|
| 2300 |
+
},
|
| 2301 |
+
"dec_ups.0.op": {
|
| 2302 |
+
"mean": 0.3276306092739105,
|
| 2303 |
+
"std": 2.0408554077148438,
|
| 2304 |
+
"abs_max": 14.5625,
|
| 2305 |
+
"frac_dead": 9.960598617908545e-06,
|
| 2306 |
+
"frac_sat": 0.0007154676713980734
|
| 2307 |
+
},
|
| 2308 |
+
"dec_blocks.1.norm1": {
|
| 2309 |
+
"mean": -8.537082612747326e-05,
|
| 2310 |
+
"std": 1.0001024007797241,
|
| 2311 |
+
"abs_max": 10.442710876464844,
|
| 2312 |
+
"frac_dead": 6.35782896551973e-07,
|
| 2313 |
+
"frac_sat": 3.178914482759865e-07
|
| 2314 |
+
},
|
| 2315 |
+
"dec_blocks.1.conv1": {
|
| 2316 |
+
"mean": -1.339745283126831,
|
| 2317 |
+
"std": 2.232980728149414,
|
| 2318 |
+
"abs_max": 34.25,
|
| 2319 |
+
"frac_dead": 1.038445407175459e-05,
|
| 2320 |
+
"frac_sat": 0.00516870291903615
|
| 2321 |
+
},
|
| 2322 |
+
"dec_blocks.1.t_proj": {
|
| 2323 |
+
"mean": -0.002125428058207035,
|
| 2324 |
+
"std": 0.05761529505252838,
|
| 2325 |
+
"abs_max": 0.34375,
|
| 2326 |
+
"frac_dead": 0.0,
|
| 2327 |
+
"frac_sat": 0.0
|
| 2328 |
+
},
|
| 2329 |
+
"dec_blocks.1.norm2": {
|
| 2330 |
+
"mean": -0.004831163212656975,
|
| 2331 |
+
"std": 0.995678186416626,
|
| 2332 |
+
"abs_max": 12.500800132751465,
|
| 2333 |
+
"frac_dead": 8.477105097881577e-07,
|
| 2334 |
+
"frac_sat": 6.993611805228284e-06
|
| 2335 |
+
},
|
| 2336 |
+
"dec_blocks.1.conv2": {
|
| 2337 |
+
"mean": 0.0020734877325594425,
|
| 2338 |
+
"std": 0.42178505659103394,
|
| 2339 |
+
"abs_max": 1.859375,
|
| 2340 |
+
"frac_dead": 3.24249267578125e-05,
|
| 2341 |
+
"frac_sat": 0.0
|
| 2342 |
+
},
|
| 2343 |
+
"dec_blocks.1.skip": {
|
| 2344 |
+
"mean": -0.018268505111336708,
|
| 2345 |
+
"std": 1.4295871257781982,
|
| 2346 |
+
"abs_max": 9.125,
|
| 2347 |
+
"frac_dead": 2.5219387680408545e-05,
|
| 2348 |
+
"frac_sat": 0.0
|
| 2349 |
+
},
|
| 2350 |
+
"dec_attns.1.norm": {
|
| 2351 |
+
"mean": 0.00011205342161701992,
|
| 2352 |
+
"std": 0.9994926452636719,
|
| 2353 |
+
"abs_max": 8.612427711486816,
|
| 2354 |
+
"frac_dead": 6.35782896551973e-07,
|
| 2355 |
+
"frac_sat": 0.0
|
| 2356 |
+
},
|
| 2357 |
+
"dec_attns.1.qkv": {
|
| 2358 |
+
"mean": 0.01128898374736309,
|
| 2359 |
+
"std": 0.6574664115905762,
|
| 2360 |
+
"abs_max": 6.59375,
|
| 2361 |
+
"frac_dead": 7.043061486911029e-05,
|
| 2362 |
+
"frac_sat": 0.0
|
| 2363 |
+
},
|
| 2364 |
+
"dec_attns.1.proj": {
|
| 2365 |
+
"mean": 0.0020239518489688635,
|
| 2366 |
+
"std": 0.3653465509414673,
|
| 2367 |
+
"abs_max": 1.3984375,
|
| 2368 |
+
"frac_dead": 0.000118255615234375,
|
| 2369 |
+
"frac_sat": 0.0
|
| 2370 |
+
},
|
| 2371 |
+
"dec_ups.1.op": {
|
| 2372 |
+
"mean": 0.015648458153009415,
|
| 2373 |
+
"std": 1.215781569480896,
|
| 2374 |
+
"abs_max": 8.9375,
|
| 2375 |
+
"frac_dead": 1.340442213404458e-05,
|
| 2376 |
+
"frac_sat": 0.0
|
| 2377 |
+
},
|
| 2378 |
+
"dec_blocks.2.norm1": {
|
| 2379 |
+
"mean": 3.191718496964313e-05,
|
| 2380 |
+
"std": 0.9998952150344849,
|
| 2381 |
+
"abs_max": 9.021163940429688,
|
| 2382 |
+
"frac_dead": 3.8853397654747823e-07,
|
| 2383 |
+
"frac_sat": 0.0
|
| 2384 |
+
},
|
| 2385 |
+
"dec_blocks.2.conv1": {
|
| 2386 |
+
"mean": -0.6205039620399475,
|
| 2387 |
+
"std": 2.467486619949341,
|
| 2388 |
+
"abs_max": 38.75,
|
| 2389 |
+
"frac_dead": 7.417466804326978e-06,
|
| 2390 |
+
"frac_sat": 0.006686422508209944
|
| 2391 |
+
},
|
| 2392 |
+
"dec_blocks.2.t_proj": {
|
| 2393 |
+
"mean": -0.005355479661375284,
|
| 2394 |
+
"std": 0.060733918100595474,
|
| 2395 |
+
"abs_max": 0.271484375,
|
| 2396 |
+
"frac_dead": 0.00010850694525288418,
|
| 2397 |
+
"frac_sat": 0.0
|
| 2398 |
+
},
|
| 2399 |
+
"dec_blocks.2.norm2": {
|
| 2400 |
+
"mean": -0.005140045657753944,
|
| 2401 |
+
"std": 0.9974221587181091,
|
| 2402 |
+
"abs_max": 12.372010231018066,
|
| 2403 |
+
"frac_dead": 3.178914482759865e-07,
|
| 2404 |
+
"frac_sat": 1.600053656147793e-05
|
| 2405 |
+
},
|
| 2406 |
+
"dec_blocks.2.conv2": {
|
| 2407 |
+
"mean": -0.0005674130516126752,
|
| 2408 |
+
"std": 0.5642880797386169,
|
| 2409 |
+
"abs_max": 3.90625,
|
| 2410 |
+
"frac_dead": 2.564324313425459e-05,
|
| 2411 |
+
"frac_sat": 0.0
|
| 2412 |
+
},
|
| 2413 |
+
"dec_blocks.2.skip": {
|
| 2414 |
+
"mean": 0.0454326756298542,
|
| 2415 |
+
"std": 1.3182048797607422,
|
| 2416 |
+
"abs_max": 10.125,
|
| 2417 |
+
"frac_dead": 3.0517578125e-05,
|
| 2418 |
+
"frac_sat": 1.059638137235197e-07
|
| 2419 |
+
},
|
| 2420 |
+
"dec_attns.2.norm": {
|
| 2421 |
+
"mean": 2.561923611210659e-05,
|
| 2422 |
+
"std": 0.9981249570846558,
|
| 2423 |
+
"abs_max": 8.257494926452637,
|
| 2424 |
+
"frac_dead": 5.298190899338806e-07,
|
| 2425 |
+
"frac_sat": 0.0
|
| 2426 |
+
},
|
| 2427 |
+
"dec_attns.2.qkv": {
|
| 2428 |
+
"mean": 0.015855487436056137,
|
| 2429 |
+
"std": 0.5967628955841064,
|
| 2430 |
+
"abs_max": 7.1875,
|
| 2431 |
+
"frac_dead": 0.00010296150867361575,
|
| 2432 |
+
"frac_sat": 0.0
|
| 2433 |
+
},
|
| 2434 |
+
"dec_attns.2.proj": {
|
| 2435 |
+
"mean": -0.0011597300181165338,
|
| 2436 |
+
"std": 0.3089199960231781,
|
| 2437 |
+
"abs_max": 1.1796875,
|
| 2438 |
+
"frac_dead": 0.00023799472546670586,
|
| 2439 |
+
"frac_sat": 0.0
|
| 2440 |
+
},
|
| 2441 |
+
"dec_ups.2.op": {
|
| 2442 |
+
"mean": 0.05098172277212143,
|
| 2443 |
+
"std": 1.2449263334274292,
|
| 2444 |
+
"abs_max": 7.40625,
|
| 2445 |
+
"frac_dead": 1.7510519683128223e-05,
|
| 2446 |
+
"frac_sat": 0.0
|
| 2447 |
+
},
|
| 2448 |
+
"dec_blocks.3.norm1": {
|
| 2449 |
+
"mean": 0.0001633289357414469,
|
| 2450 |
+
"std": 1.0001723766326904,
|
| 2451 |
+
"abs_max": 7.487753391265869,
|
| 2452 |
+
"frac_dead": 9.5367431640625e-07,
|
| 2453 |
+
"frac_sat": 0.0
|
| 2454 |
+
},
|
| 2455 |
+
"dec_blocks.3.conv1": {
|
| 2456 |
+
"mean": -0.9776155948638916,
|
| 2457 |
+
"std": 1.510038137435913,
|
| 2458 |
+
"abs_max": 18.375,
|
| 2459 |
+
"frac_dead": 1.1258654922130518e-05,
|
| 2460 |
+
"frac_sat": 0.00020133124780841172
|
| 2461 |
+
},
|
| 2462 |
+
"dec_blocks.3.t_proj": {
|
| 2463 |
+
"mean": -0.00041362602496519685,
|
| 2464 |
+
"std": 0.06447338312864304,
|
| 2465 |
+
"abs_max": 0.306640625,
|
| 2466 |
+
"frac_dead": 0.0,
|
| 2467 |
+
"frac_sat": 0.0
|
| 2468 |
+
},
|
| 2469 |
+
"dec_blocks.3.norm2": {
|
| 2470 |
+
"mean": -0.004704537335783243,
|
| 2471 |
+
"std": 0.9959903955459595,
|
| 2472 |
+
"abs_max": 11.89158821105957,
|
| 2473 |
+
"frac_dead": 7.417467031700653e-07,
|
| 2474 |
+
"frac_sat": 7.417467031700653e-07
|
| 2475 |
+
},
|
| 2476 |
+
"dec_blocks.3.conv2": {
|
| 2477 |
+
"mean": 0.05290665104985237,
|
| 2478 |
+
"std": 0.4503594934940338,
|
| 2479 |
+
"abs_max": 3.375,
|
| 2480 |
+
"frac_dead": 4.665056985686533e-05,
|
| 2481 |
+
"frac_sat": 0.0
|
| 2482 |
+
},
|
| 2483 |
+
"dec_blocks.3.skip": {
|
| 2484 |
+
"mean": 0.03325294330716133,
|
| 2485 |
+
"std": 1.0477484464645386,
|
| 2486 |
+
"abs_max": 7.46875,
|
| 2487 |
+
"frac_dead": 4.121992242289707e-05,
|
| 2488 |
+
"frac_sat": 0.0
|
| 2489 |
+
},
|
| 2490 |
+
"dec_ups.3.op": {
|
| 2491 |
+
"mean": 0.03040865808725357,
|
| 2492 |
+
"std": 1.2022557258605957,
|
| 2493 |
+
"abs_max": 7.25,
|
| 2494 |
+
"frac_dead": 1.8397966414340772e-05,
|
| 2495 |
+
"frac_sat": 0.0
|
| 2496 |
+
},
|
| 2497 |
+
"dec_blocks.4.norm1": {
|
| 2498 |
+
"mean": 0.00019849088857881725,
|
| 2499 |
+
"std": 1.0003993511199951,
|
| 2500 |
+
"abs_max": 7.175535202026367,
|
| 2501 |
+
"frac_dead": 7.329163622671331e-07,
|
| 2502 |
+
"frac_sat": 0.0
|
| 2503 |
+
},
|
| 2504 |
+
"dec_blocks.4.conv1": {
|
| 2505 |
+
"mean": -0.652289092540741,
|
| 2506 |
+
"std": 1.7954871654510498,
|
| 2507 |
+
"abs_max": 9.375,
|
| 2508 |
+
"frac_dead": 1.0304980605724268e-05,
|
| 2509 |
+
"frac_sat": 0.0
|
| 2510 |
+
},
|
| 2511 |
+
"dec_blocks.4.t_proj": {
|
| 2512 |
+
"mean": -0.00229295389726758,
|
| 2513 |
+
"std": 0.06348087638616562,
|
| 2514 |
+
"abs_max": 0.29296875,
|
| 2515 |
+
"frac_dead": 0.0,
|
| 2516 |
+
"frac_sat": 0.0
|
| 2517 |
+
},
|
| 2518 |
+
"dec_blocks.4.norm2": {
|
| 2519 |
+
"mean": -0.0027420183178037405,
|
| 2520 |
+
"std": 0.9977593421936035,
|
| 2521 |
+
"abs_max": 4.8135457038879395,
|
| 2522 |
+
"frac_dead": 1.086129145733139e-06,
|
| 2523 |
+
"frac_sat": 0.0
|
| 2524 |
+
},
|
| 2525 |
+
"dec_blocks.4.conv2": {
|
| 2526 |
+
"mean": -0.09926741570234299,
|
| 2527 |
+
"std": 0.6462329030036926,
|
| 2528 |
+
"abs_max": 3.515625,
|
| 2529 |
+
"frac_dead": 4.35378824477084e-05,
|
| 2530 |
+
"frac_sat": 0.0
|
| 2531 |
+
},
|
| 2532 |
+
"dec_blocks.4.skip": {
|
| 2533 |
+
"mean": -0.07995840162038803,
|
| 2534 |
+
"std": 1.1788166761398315,
|
| 2535 |
+
"abs_max": 7.59375,
|
| 2536 |
+
"frac_dead": 4.5286284148460254e-05,
|
| 2537 |
+
"frac_sat": 0.0
|
| 2538 |
+
},
|
| 2539 |
+
"dec_ups.4.op": {
|
| 2540 |
+
"mean": -0.2425907403230667,
|
| 2541 |
+
"std": 1.8716286420822144,
|
| 2542 |
+
"abs_max": 14.5625,
|
| 2543 |
+
"frac_dead": 1.8404589354759082e-05,
|
| 2544 |
+
"frac_sat": 0.00025292899226769805
|
| 2545 |
+
},
|
| 2546 |
+
"dec_blocks.5.norm1": {
|
| 2547 |
+
"mean": -0.0022172522731125355,
|
| 2548 |
+
"std": 0.9993937015533447,
|
| 2549 |
+
"abs_max": 8.945484161376953,
|
| 2550 |
+
"frac_dead": 8.245309004450974e-07,
|
| 2551 |
+
"frac_sat": 0.0
|
| 2552 |
+
},
|
| 2553 |
+
"dec_blocks.5.conv1": {
|
| 2554 |
+
"mean": -0.19136612117290497,
|
| 2555 |
+
"std": 1.2564855813980103,
|
| 2556 |
+
"abs_max": 9.3125,
|
| 2557 |
+
"frac_dead": 1.5394554793601856e-05,
|
| 2558 |
+
"frac_sat": 0.0
|
| 2559 |
+
},
|
| 2560 |
+
"dec_blocks.5.t_proj": {
|
| 2561 |
+
"mean": 0.002546150702983141,
|
| 2562 |
+
"std": 0.09909532964229584,
|
| 2563 |
+
"abs_max": 0.60546875,
|
| 2564 |
+
"frac_dead": 0.0,
|
| 2565 |
+
"frac_sat": 0.0
|
| 2566 |
+
},
|
| 2567 |
+
"dec_blocks.5.norm2": {
|
| 2568 |
+
"mean": 0.00035895511973649263,
|
| 2569 |
+
"std": 1.0026216506958008,
|
| 2570 |
+
"abs_max": 10.721990585327148,
|
| 2571 |
+
"frac_dead": 1.6225709487116546e-07,
|
| 2572 |
+
"frac_sat": 4.967053612858763e-08
|
| 2573 |
+
},
|
| 2574 |
+
"dec_blocks.5.conv2": {
|
| 2575 |
+
"mean": -0.31369104981422424,
|
| 2576 |
+
"std": 1.3686388731002808,
|
| 2577 |
+
"abs_max": 13.8125,
|
| 2578 |
+
"frac_dead": 2.8093656510463916e-05,
|
| 2579 |
+
"frac_sat": 1.231829355674563e-05
|
| 2580 |
+
},
|
| 2581 |
+
"dec_blocks.5.skip": {
|
| 2582 |
+
"mean": -0.03304848447442055,
|
| 2583 |
+
"std": 1.2728080749511719,
|
| 2584 |
+
"abs_max": 8.125,
|
| 2585 |
+
"frac_dead": 4.986591011402197e-05,
|
| 2586 |
+
"frac_sat": 0.0
|
| 2587 |
+
},
|
| 2588 |
+
"out_norm": {
|
| 2589 |
+
"mean": -0.0035718195140361786,
|
| 2590 |
+
"std": 0.9979620575904846,
|
| 2591 |
+
"abs_max": 8.589007377624512,
|
| 2592 |
+
"frac_dead": 3.1458006333195954e-07,
|
| 2593 |
+
"frac_sat": 0.0
|
| 2594 |
+
},
|
| 2595 |
+
"out_conv": {
|
| 2596 |
+
"mean": 0.14288969337940216,
|
| 2597 |
+
"std": 0.2213430255651474,
|
| 2598 |
+
"abs_max": 0.91796875,
|
| 2599 |
+
"frac_dead": 1.3139512702764478e-05,
|
| 2600 |
+
"frac_sat": 0.0
|
| 2601 |
+
}
|
| 2602 |
+
}
|
| 2603 |
+
}
|
diagnostics/per_layer_iter1500.json
ADDED
|
@@ -0,0 +1,2603 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"iter": 1500,
|
| 3 |
+
"per_layer": {
|
| 4 |
+
"time_mlp.1.weight": {
|
| 5 |
+
"w_norm": 18.831348419189453,
|
| 6 |
+
"g_norm": 0.10205292701721191,
|
| 7 |
+
"shape": [
|
| 8 |
+
1024,
|
| 9 |
+
256
|
| 10 |
+
]
|
| 11 |
+
},
|
| 12 |
+
"time_mlp.1.bias": {
|
| 13 |
+
"w_norm": 1.1589568853378296,
|
| 14 |
+
"g_norm": 0.009027610532939434,
|
| 15 |
+
"shape": [
|
| 16 |
+
1024
|
| 17 |
+
]
|
| 18 |
+
},
|
| 19 |
+
"time_mlp.3.weight": {
|
| 20 |
+
"w_norm": 9.640146255493164,
|
| 21 |
+
"g_norm": 0.019456155598163605,
|
| 22 |
+
"shape": [
|
| 23 |
+
256,
|
| 24 |
+
1024
|
| 25 |
+
]
|
| 26 |
+
},
|
| 27 |
+
"time_mlp.3.bias": {
|
| 28 |
+
"w_norm": 0.29026785492897034,
|
| 29 |
+
"g_norm": 0.00706234248355031,
|
| 30 |
+
"shape": [
|
| 31 |
+
256
|
| 32 |
+
]
|
| 33 |
+
},
|
| 34 |
+
"input_conv.weight": {
|
| 35 |
+
"w_norm": 7.941542625427246,
|
| 36 |
+
"g_norm": 0.06458809971809387,
|
| 37 |
+
"shape": [
|
| 38 |
+
192,
|
| 39 |
+
6,
|
| 40 |
+
3,
|
| 41 |
+
3
|
| 42 |
+
]
|
| 43 |
+
},
|
| 44 |
+
"input_conv.bias": {
|
| 45 |
+
"w_norm": 1.1898912191390991,
|
| 46 |
+
"g_norm": 0.07269138097763062,
|
| 47 |
+
"shape": [
|
| 48 |
+
192
|
| 49 |
+
]
|
| 50 |
+
},
|
| 51 |
+
"enc_blocks.0.norm1.weight": {
|
| 52 |
+
"w_norm": 13.811203002929688,
|
| 53 |
+
"g_norm": 0.017467472702264786,
|
| 54 |
+
"shape": [
|
| 55 |
+
192
|
| 56 |
+
]
|
| 57 |
+
},
|
| 58 |
+
"enc_blocks.0.norm1.bias": {
|
| 59 |
+
"w_norm": 0.08192380517721176,
|
| 60 |
+
"g_norm": 0.026821866631507874,
|
| 61 |
+
"shape": [
|
| 62 |
+
192
|
| 63 |
+
]
|
| 64 |
+
},
|
| 65 |
+
"enc_blocks.0.conv1.weight": {
|
| 66 |
+
"w_norm": 8.32677936553955,
|
| 67 |
+
"g_norm": 0.17138740420341492,
|
| 68 |
+
"shape": [
|
| 69 |
+
192,
|
| 70 |
+
192,
|
| 71 |
+
3,
|
| 72 |
+
3
|
| 73 |
+
]
|
| 74 |
+
},
|
| 75 |
+
"enc_blocks.0.conv1.bias": {
|
| 76 |
+
"w_norm": 0.20204275846481323,
|
| 77 |
+
"g_norm": 0.017192741855978966,
|
| 78 |
+
"shape": [
|
| 79 |
+
192
|
| 80 |
+
]
|
| 81 |
+
},
|
| 82 |
+
"enc_blocks.0.t_proj.weight": {
|
| 83 |
+
"w_norm": 8.070746421813965,
|
| 84 |
+
"g_norm": 0.010215070098638535,
|
| 85 |
+
"shape": [
|
| 86 |
+
192,
|
| 87 |
+
256
|
| 88 |
+
]
|
| 89 |
+
},
|
| 90 |
+
"enc_blocks.0.t_proj.bias": {
|
| 91 |
+
"w_norm": 0.5025191903114319,
|
| 92 |
+
"g_norm": 0.01719454675912857,
|
| 93 |
+
"shape": [
|
| 94 |
+
192
|
| 95 |
+
]
|
| 96 |
+
},
|
| 97 |
+
"enc_blocks.0.norm2.weight": {
|
| 98 |
+
"w_norm": 13.839881896972656,
|
| 99 |
+
"g_norm": 0.008044346235692501,
|
| 100 |
+
"shape": [
|
| 101 |
+
192
|
| 102 |
+
]
|
| 103 |
+
},
|
| 104 |
+
"enc_blocks.0.norm2.bias": {
|
| 105 |
+
"w_norm": 0.04994768276810646,
|
| 106 |
+
"g_norm": 0.009143427945673466,
|
| 107 |
+
"shape": [
|
| 108 |
+
192
|
| 109 |
+
]
|
| 110 |
+
},
|
| 111 |
+
"enc_blocks.0.conv2.weight": {
|
| 112 |
+
"w_norm": 8.244281768798828,
|
| 113 |
+
"g_norm": 0.08354029059410095,
|
| 114 |
+
"shape": [
|
| 115 |
+
192,
|
| 116 |
+
192,
|
| 117 |
+
3,
|
| 118 |
+
3
|
| 119 |
+
]
|
| 120 |
+
},
|
| 121 |
+
"enc_blocks.0.conv2.bias": {
|
| 122 |
+
"w_norm": 0.1875186562538147,
|
| 123 |
+
"g_norm": 0.005785597022622824,
|
| 124 |
+
"shape": [
|
| 125 |
+
192
|
| 126 |
+
]
|
| 127 |
+
},
|
| 128 |
+
"enc_blocks.1.norm1.weight": {
|
| 129 |
+
"w_norm": 13.85268497467041,
|
| 130 |
+
"g_norm": 7.552062015747651e-05,
|
| 131 |
+
"shape": [
|
| 132 |
+
192
|
| 133 |
+
]
|
| 134 |
+
},
|
| 135 |
+
"enc_blocks.1.norm1.bias": {
|
| 136 |
+
"w_norm": 0.026036987081170082,
|
| 137 |
+
"g_norm": 4.653045107261278e-05,
|
| 138 |
+
"shape": [
|
| 139 |
+
192
|
| 140 |
+
]
|
| 141 |
+
},
|
| 142 |
+
"enc_blocks.1.conv1.weight": {
|
| 143 |
+
"w_norm": 8.178509712219238,
|
| 144 |
+
"g_norm": 0.001746668480336666,
|
| 145 |
+
"shape": [
|
| 146 |
+
192,
|
| 147 |
+
192,
|
| 148 |
+
3,
|
| 149 |
+
3
|
| 150 |
+
]
|
| 151 |
+
},
|
| 152 |
+
"enc_blocks.1.conv1.bias": {
|
| 153 |
+
"w_norm": 0.19200706481933594,
|
| 154 |
+
"g_norm": 5.417833017418161e-05,
|
| 155 |
+
"shape": [
|
| 156 |
+
192
|
| 157 |
+
]
|
| 158 |
+
},
|
| 159 |
+
"enc_blocks.1.t_proj.weight": {
|
| 160 |
+
"w_norm": 8.048393249511719,
|
| 161 |
+
"g_norm": 8.108843758236617e-05,
|
| 162 |
+
"shape": [
|
| 163 |
+
192,
|
| 164 |
+
256
|
| 165 |
+
]
|
| 166 |
+
},
|
| 167 |
+
"enc_blocks.1.t_proj.bias": {
|
| 168 |
+
"w_norm": 0.5217288136482239,
|
| 169 |
+
"g_norm": 5.417064676294103e-05,
|
| 170 |
+
"shape": [
|
| 171 |
+
192
|
| 172 |
+
]
|
| 173 |
+
},
|
| 174 |
+
"enc_blocks.1.norm2.weight": {
|
| 175 |
+
"w_norm": 13.822420120239258,
|
| 176 |
+
"g_norm": 0.00010076359467348084,
|
| 177 |
+
"shape": [
|
| 178 |
+
192
|
| 179 |
+
]
|
| 180 |
+
},
|
| 181 |
+
"enc_blocks.1.norm2.bias": {
|
| 182 |
+
"w_norm": 0.04598449543118477,
|
| 183 |
+
"g_norm": 6.376473902491853e-05,
|
| 184 |
+
"shape": [
|
| 185 |
+
192
|
| 186 |
+
]
|
| 187 |
+
},
|
| 188 |
+
"enc_blocks.1.conv2.weight": {
|
| 189 |
+
"w_norm": 8.135522842407227,
|
| 190 |
+
"g_norm": 0.002183968434110284,
|
| 191 |
+
"shape": [
|
| 192 |
+
192,
|
| 193 |
+
192,
|
| 194 |
+
3,
|
| 195 |
+
3
|
| 196 |
+
]
|
| 197 |
+
},
|
| 198 |
+
"enc_blocks.1.conv2.bias": {
|
| 199 |
+
"w_norm": 0.20787833631038666,
|
| 200 |
+
"g_norm": 0.00010054003359982744,
|
| 201 |
+
"shape": [
|
| 202 |
+
192
|
| 203 |
+
]
|
| 204 |
+
},
|
| 205 |
+
"enc_blocks.2.norm1.weight": {
|
| 206 |
+
"w_norm": 13.844120025634766,
|
| 207 |
+
"g_norm": 2.4686974938958883e-05,
|
| 208 |
+
"shape": [
|
| 209 |
+
192
|
| 210 |
+
]
|
| 211 |
+
},
|
| 212 |
+
"enc_blocks.2.norm1.bias": {
|
| 213 |
+
"w_norm": 0.035012636333703995,
|
| 214 |
+
"g_norm": 1.9267890820628963e-05,
|
| 215 |
+
"shape": [
|
| 216 |
+
192
|
| 217 |
+
]
|
| 218 |
+
},
|
| 219 |
+
"enc_blocks.2.conv1.weight": {
|
| 220 |
+
"w_norm": 11.575699806213379,
|
| 221 |
+
"g_norm": 0.000611058552749455,
|
| 222 |
+
"shape": [
|
| 223 |
+
384,
|
| 224 |
+
192,
|
| 225 |
+
3,
|
| 226 |
+
3
|
| 227 |
+
]
|
| 228 |
+
},
|
| 229 |
+
"enc_blocks.2.conv1.bias": {
|
| 230 |
+
"w_norm": 0.26600784063339233,
|
| 231 |
+
"g_norm": 2.2671565602649935e-05,
|
| 232 |
+
"shape": [
|
| 233 |
+
384
|
| 234 |
+
]
|
| 235 |
+
},
|
| 236 |
+
"enc_blocks.2.t_proj.weight": {
|
| 237 |
+
"w_norm": 11.522205352783203,
|
| 238 |
+
"g_norm": 3.756795194931328e-05,
|
| 239 |
+
"shape": [
|
| 240 |
+
384,
|
| 241 |
+
256
|
| 242 |
+
]
|
| 243 |
+
},
|
| 244 |
+
"enc_blocks.2.t_proj.bias": {
|
| 245 |
+
"w_norm": 0.720262348651886,
|
| 246 |
+
"g_norm": 2.2679385438095778e-05,
|
| 247 |
+
"shape": [
|
| 248 |
+
384
|
| 249 |
+
]
|
| 250 |
+
},
|
| 251 |
+
"enc_blocks.2.norm2.weight": {
|
| 252 |
+
"w_norm": 19.48476791381836,
|
| 253 |
+
"g_norm": 4.690673085860908e-05,
|
| 254 |
+
"shape": [
|
| 255 |
+
384
|
| 256 |
+
]
|
| 257 |
+
},
|
| 258 |
+
"enc_blocks.2.norm2.bias": {
|
| 259 |
+
"w_norm": 0.11410542577505112,
|
| 260 |
+
"g_norm": 2.5380313672940247e-05,
|
| 261 |
+
"shape": [
|
| 262 |
+
384
|
| 263 |
+
]
|
| 264 |
+
},
|
| 265 |
+
"enc_blocks.2.conv2.weight": {
|
| 266 |
+
"w_norm": 11.546863555908203,
|
| 267 |
+
"g_norm": 0.0008082764106802642,
|
| 268 |
+
"shape": [
|
| 269 |
+
384,
|
| 270 |
+
384,
|
| 271 |
+
3,
|
| 272 |
+
3
|
| 273 |
+
]
|
| 274 |
+
},
|
| 275 |
+
"enc_blocks.2.conv2.bias": {
|
| 276 |
+
"w_norm": 0.20199169218540192,
|
| 277 |
+
"g_norm": 2.2719479602528736e-05,
|
| 278 |
+
"shape": [
|
| 279 |
+
384
|
| 280 |
+
]
|
| 281 |
+
},
|
| 282 |
+
"enc_blocks.2.skip.weight": {
|
| 283 |
+
"w_norm": 11.241255760192871,
|
| 284 |
+
"g_norm": 0.0006491548265330493,
|
| 285 |
+
"shape": [
|
| 286 |
+
384,
|
| 287 |
+
192,
|
| 288 |
+
1,
|
| 289 |
+
1
|
| 290 |
+
]
|
| 291 |
+
},
|
| 292 |
+
"enc_blocks.2.skip.bias": {
|
| 293 |
+
"w_norm": 0.8258646726608276,
|
| 294 |
+
"g_norm": 2.2719479602528736e-05,
|
| 295 |
+
"shape": [
|
| 296 |
+
384
|
| 297 |
+
]
|
| 298 |
+
},
|
| 299 |
+
"enc_blocks.3.norm1.weight": {
|
| 300 |
+
"w_norm": 19.59480857849121,
|
| 301 |
+
"g_norm": 4.11441669712076e-06,
|
| 302 |
+
"shape": [
|
| 303 |
+
384
|
| 304 |
+
]
|
| 305 |
+
},
|
| 306 |
+
"enc_blocks.3.norm1.bias": {
|
| 307 |
+
"w_norm": 0.03167359158396721,
|
| 308 |
+
"g_norm": 3.48591424881306e-06,
|
| 309 |
+
"shape": [
|
| 310 |
+
384
|
| 311 |
+
]
|
| 312 |
+
},
|
| 313 |
+
"enc_blocks.3.conv1.weight": {
|
| 314 |
+
"w_norm": 11.656805038452148,
|
| 315 |
+
"g_norm": 0.00011399946379242465,
|
| 316 |
+
"shape": [
|
| 317 |
+
384,
|
| 318 |
+
384,
|
| 319 |
+
3,
|
| 320 |
+
3
|
| 321 |
+
]
|
| 322 |
+
},
|
| 323 |
+
"enc_blocks.3.conv1.bias": {
|
| 324 |
+
"w_norm": 0.1977333277463913,
|
| 325 |
+
"g_norm": 3.2061070669442415e-06,
|
| 326 |
+
"shape": [
|
| 327 |
+
384
|
| 328 |
+
]
|
| 329 |
+
},
|
| 330 |
+
"enc_blocks.3.t_proj.weight": {
|
| 331 |
+
"w_norm": 11.358353614807129,
|
| 332 |
+
"g_norm": 2.2405642994272057e-06,
|
| 333 |
+
"shape": [
|
| 334 |
+
384,
|
| 335 |
+
256
|
| 336 |
+
]
|
| 337 |
+
},
|
| 338 |
+
"enc_blocks.3.t_proj.bias": {
|
| 339 |
+
"w_norm": 0.6917930841445923,
|
| 340 |
+
"g_norm": 3.2053535505838227e-06,
|
| 341 |
+
"shape": [
|
| 342 |
+
384
|
| 343 |
+
]
|
| 344 |
+
},
|
| 345 |
+
"enc_blocks.3.norm2.weight": {
|
| 346 |
+
"w_norm": 19.485275268554688,
|
| 347 |
+
"g_norm": 5.654585947922897e-06,
|
| 348 |
+
"shape": [
|
| 349 |
+
384
|
| 350 |
+
]
|
| 351 |
+
},
|
| 352 |
+
"enc_blocks.3.norm2.bias": {
|
| 353 |
+
"w_norm": 0.0983862653374672,
|
| 354 |
+
"g_norm": 4.409213943290524e-06,
|
| 355 |
+
"shape": [
|
| 356 |
+
384
|
| 357 |
+
]
|
| 358 |
+
},
|
| 359 |
+
"enc_blocks.3.conv2.weight": {
|
| 360 |
+
"w_norm": 11.416952133178711,
|
| 361 |
+
"g_norm": 0.00018496303528081626,
|
| 362 |
+
"shape": [
|
| 363 |
+
384,
|
| 364 |
+
384,
|
| 365 |
+
3,
|
| 366 |
+
3
|
| 367 |
+
]
|
| 368 |
+
},
|
| 369 |
+
"enc_blocks.3.conv2.bias": {
|
| 370 |
+
"w_norm": 0.1904715597629547,
|
| 371 |
+
"g_norm": 6.365587978507392e-06,
|
| 372 |
+
"shape": [
|
| 373 |
+
384
|
| 374 |
+
]
|
| 375 |
+
},
|
| 376 |
+
"enc_blocks.4.norm1.weight": {
|
| 377 |
+
"w_norm": 19.591276168823242,
|
| 378 |
+
"g_norm": 1.3642720659845509e-06,
|
| 379 |
+
"shape": [
|
| 380 |
+
384
|
| 381 |
+
]
|
| 382 |
+
},
|
| 383 |
+
"enc_blocks.4.norm1.bias": {
|
| 384 |
+
"w_norm": 0.03558076173067093,
|
| 385 |
+
"g_norm": 1.0926190725513152e-06,
|
| 386 |
+
"shape": [
|
| 387 |
+
384
|
| 388 |
+
]
|
| 389 |
+
},
|
| 390 |
+
"enc_blocks.4.conv1.weight": {
|
| 391 |
+
"w_norm": 16.351974487304688,
|
| 392 |
+
"g_norm": 4.5458058593794703e-05,
|
| 393 |
+
"shape": [
|
| 394 |
+
768,
|
| 395 |
+
384,
|
| 396 |
+
3,
|
| 397 |
+
3
|
| 398 |
+
]
|
| 399 |
+
},
|
| 400 |
+
"enc_blocks.4.conv1.bias": {
|
| 401 |
+
"w_norm": 0.2806565761566162,
|
| 402 |
+
"g_norm": 7.310076739486249e-07,
|
| 403 |
+
"shape": [
|
| 404 |
+
768
|
| 405 |
+
]
|
| 406 |
+
},
|
| 407 |
+
"enc_blocks.4.t_proj.weight": {
|
| 408 |
+
"w_norm": 16.041704177856445,
|
| 409 |
+
"g_norm": 8.280318866127345e-07,
|
| 410 |
+
"shape": [
|
| 411 |
+
768,
|
| 412 |
+
256
|
| 413 |
+
]
|
| 414 |
+
},
|
| 415 |
+
"enc_blocks.4.t_proj.bias": {
|
| 416 |
+
"w_norm": 0.9907169342041016,
|
| 417 |
+
"g_norm": 7.310062528631533e-07,
|
| 418 |
+
"shape": [
|
| 419 |
+
768
|
| 420 |
+
]
|
| 421 |
+
},
|
| 422 |
+
"enc_blocks.4.norm2.weight": {
|
| 423 |
+
"w_norm": 27.56458282470703,
|
| 424 |
+
"g_norm": 1.4526040104101412e-06,
|
| 425 |
+
"shape": [
|
| 426 |
+
768
|
| 427 |
+
]
|
| 428 |
+
},
|
| 429 |
+
"enc_blocks.4.norm2.bias": {
|
| 430 |
+
"w_norm": 0.1475459188222885,
|
| 431 |
+
"g_norm": 1.0344501788495108e-06,
|
| 432 |
+
"shape": [
|
| 433 |
+
768
|
| 434 |
+
]
|
| 435 |
+
},
|
| 436 |
+
"enc_blocks.4.conv2.weight": {
|
| 437 |
+
"w_norm": 16.167402267456055,
|
| 438 |
+
"g_norm": 7.684766023885459e-05,
|
| 439 |
+
"shape": [
|
| 440 |
+
768,
|
| 441 |
+
768,
|
| 442 |
+
3,
|
| 443 |
+
3
|
| 444 |
+
]
|
| 445 |
+
},
|
| 446 |
+
"enc_blocks.4.conv2.bias": {
|
| 447 |
+
"w_norm": 0.19410400092601776,
|
| 448 |
+
"g_norm": 2.097709511872381e-06,
|
| 449 |
+
"shape": [
|
| 450 |
+
768
|
| 451 |
+
]
|
| 452 |
+
},
|
| 453 |
+
"enc_blocks.4.skip.weight": {
|
| 454 |
+
"w_norm": 15.907557487487793,
|
| 455 |
+
"g_norm": 0.000103915692307055,
|
| 456 |
+
"shape": [
|
| 457 |
+
768,
|
| 458 |
+
384,
|
| 459 |
+
1,
|
| 460 |
+
1
|
| 461 |
+
]
|
| 462 |
+
},
|
| 463 |
+
"enc_blocks.4.skip.bias": {
|
| 464 |
+
"w_norm": 0.8080542683601379,
|
| 465 |
+
"g_norm": 2.097709511872381e-06,
|
| 466 |
+
"shape": [
|
| 467 |
+
768
|
| 468 |
+
]
|
| 469 |
+
},
|
| 470 |
+
"enc_blocks.5.norm1.weight": {
|
| 471 |
+
"w_norm": 27.710412979125977,
|
| 472 |
+
"g_norm": 5.43541773367906e-07,
|
| 473 |
+
"shape": [
|
| 474 |
+
768
|
| 475 |
+
]
|
| 476 |
+
},
|
| 477 |
+
"enc_blocks.5.norm1.bias": {
|
| 478 |
+
"w_norm": 0.049210891127586365,
|
| 479 |
+
"g_norm": 5.415469104264048e-07,
|
| 480 |
+
"shape": [
|
| 481 |
+
768
|
| 482 |
+
]
|
| 483 |
+
},
|
| 484 |
+
"enc_blocks.5.conv1.weight": {
|
| 485 |
+
"w_norm": 16.535066604614258,
|
| 486 |
+
"g_norm": 1.713006349746138e-05,
|
| 487 |
+
"shape": [
|
| 488 |
+
768,
|
| 489 |
+
768,
|
| 490 |
+
3,
|
| 491 |
+
3
|
| 492 |
+
]
|
| 493 |
+
},
|
| 494 |
+
"enc_blocks.5.conv1.bias": {
|
| 495 |
+
"w_norm": 0.20113901793956757,
|
| 496 |
+
"g_norm": 3.853272971809929e-07,
|
| 497 |
+
"shape": [
|
| 498 |
+
768
|
| 499 |
+
]
|
| 500 |
+
},
|
| 501 |
+
"enc_blocks.5.t_proj.weight": {
|
| 502 |
+
"w_norm": 15.998994827270508,
|
| 503 |
+
"g_norm": 3.5397795272729127e-07,
|
| 504 |
+
"shape": [
|
| 505 |
+
768,
|
| 506 |
+
256
|
| 507 |
+
]
|
| 508 |
+
},
|
| 509 |
+
"enc_blocks.5.t_proj.bias": {
|
| 510 |
+
"w_norm": 0.9972907304763794,
|
| 511 |
+
"g_norm": 3.853358805372409e-07,
|
| 512 |
+
"shape": [
|
| 513 |
+
768
|
| 514 |
+
]
|
| 515 |
+
},
|
| 516 |
+
"enc_blocks.5.norm2.weight": {
|
| 517 |
+
"w_norm": 27.60606575012207,
|
| 518 |
+
"g_norm": 5.676409955412964e-07,
|
| 519 |
+
"shape": [
|
| 520 |
+
768
|
| 521 |
+
]
|
| 522 |
+
},
|
| 523 |
+
"enc_blocks.5.norm2.bias": {
|
| 524 |
+
"w_norm": 0.1123356744647026,
|
| 525 |
+
"g_norm": 5.299498866406793e-07,
|
| 526 |
+
"shape": [
|
| 527 |
+
768
|
| 528 |
+
]
|
| 529 |
+
},
|
| 530 |
+
"enc_blocks.5.conv2.weight": {
|
| 531 |
+
"w_norm": 16.294984817504883,
|
| 532 |
+
"g_norm": 3.0194179998943582e-05,
|
| 533 |
+
"shape": [
|
| 534 |
+
768,
|
| 535 |
+
768,
|
| 536 |
+
3,
|
| 537 |
+
3
|
| 538 |
+
]
|
| 539 |
+
},
|
| 540 |
+
"enc_blocks.5.conv2.bias": {
|
| 541 |
+
"w_norm": 0.19272859394550323,
|
| 542 |
+
"g_norm": 7.503235224248783e-07,
|
| 543 |
+
"shape": [
|
| 544 |
+
768
|
| 545 |
+
]
|
| 546 |
+
},
|
| 547 |
+
"enc_attns.3.norm.weight": {
|
| 548 |
+
"w_norm": 19.56330680847168,
|
| 549 |
+
"g_norm": 4.690531568485312e-06,
|
| 550 |
+
"shape": [
|
| 551 |
+
384
|
| 552 |
+
]
|
| 553 |
+
},
|
| 554 |
+
"enc_attns.3.norm.bias": {
|
| 555 |
+
"w_norm": 0.04402874410152435,
|
| 556 |
+
"g_norm": 2.7959677026956342e-06,
|
| 557 |
+
"shape": [
|
| 558 |
+
384
|
| 559 |
+
]
|
| 560 |
+
},
|
| 561 |
+
"enc_attns.3.qkv.weight": {
|
| 562 |
+
"w_norm": 19.674089431762695,
|
| 563 |
+
"g_norm": 0.0001051163490046747,
|
| 564 |
+
"shape": [
|
| 565 |
+
1152,
|
| 566 |
+
384,
|
| 567 |
+
1,
|
| 568 |
+
1
|
| 569 |
+
]
|
| 570 |
+
},
|
| 571 |
+
"enc_attns.3.qkv.bias": {
|
| 572 |
+
"w_norm": 1.0076570510864258,
|
| 573 |
+
"g_norm": 4.560146862786496e-06,
|
| 574 |
+
"shape": [
|
| 575 |
+
1152
|
| 576 |
+
]
|
| 577 |
+
},
|
| 578 |
+
"enc_attns.3.proj.weight": {
|
| 579 |
+
"w_norm": 11.307135581970215,
|
| 580 |
+
"g_norm": 8.055998478084803e-05,
|
| 581 |
+
"shape": [
|
| 582 |
+
384,
|
| 583 |
+
384,
|
| 584 |
+
1,
|
| 585 |
+
1
|
| 586 |
+
]
|
| 587 |
+
},
|
| 588 |
+
"enc_attns.3.proj.bias": {
|
| 589 |
+
"w_norm": 0.5636854767799377,
|
| 590 |
+
"g_norm": 4.890626769338269e-06,
|
| 591 |
+
"shape": [
|
| 592 |
+
384
|
| 593 |
+
]
|
| 594 |
+
},
|
| 595 |
+
"enc_attns.4.norm.weight": {
|
| 596 |
+
"w_norm": 27.659299850463867,
|
| 597 |
+
"g_norm": 1.196475409415143e-06,
|
| 598 |
+
"shape": [
|
| 599 |
+
768
|
| 600 |
+
]
|
| 601 |
+
},
|
| 602 |
+
"enc_attns.4.norm.bias": {
|
| 603 |
+
"w_norm": 0.03487004339694977,
|
| 604 |
+
"g_norm": 7.422240173582395e-07,
|
| 605 |
+
"shape": [
|
| 606 |
+
768
|
| 607 |
+
]
|
| 608 |
+
},
|
| 609 |
+
"enc_attns.4.qkv.weight": {
|
| 610 |
+
"w_norm": 27.882322311401367,
|
| 611 |
+
"g_norm": 4.89087360620033e-05,
|
| 612 |
+
"shape": [
|
| 613 |
+
2304,
|
| 614 |
+
768,
|
| 615 |
+
1,
|
| 616 |
+
1
|
| 617 |
+
]
|
| 618 |
+
},
|
| 619 |
+
"enc_attns.4.qkv.bias": {
|
| 620 |
+
"w_norm": 1.0131044387817383,
|
| 621 |
+
"g_norm": 1.190264356409898e-06,
|
| 622 |
+
"shape": [
|
| 623 |
+
2304
|
| 624 |
+
]
|
| 625 |
+
},
|
| 626 |
+
"enc_attns.4.proj.weight": {
|
| 627 |
+
"w_norm": 15.986047744750977,
|
| 628 |
+
"g_norm": 4.969876681570895e-05,
|
| 629 |
+
"shape": [
|
| 630 |
+
768,
|
| 631 |
+
768,
|
| 632 |
+
1,
|
| 633 |
+
1
|
| 634 |
+
]
|
| 635 |
+
},
|
| 636 |
+
"enc_attns.4.proj.bias": {
|
| 637 |
+
"w_norm": 0.5729395747184753,
|
| 638 |
+
"g_norm": 1.8785445945468382e-06,
|
| 639 |
+
"shape": [
|
| 640 |
+
768
|
| 641 |
+
]
|
| 642 |
+
},
|
| 643 |
+
"enc_attns.5.norm.weight": {
|
| 644 |
+
"w_norm": 27.67577362060547,
|
| 645 |
+
"g_norm": 4.395630526232708e-07,
|
| 646 |
+
"shape": [
|
| 647 |
+
768
|
| 648 |
+
]
|
| 649 |
+
},
|
| 650 |
+
"enc_attns.5.norm.bias": {
|
| 651 |
+
"w_norm": 0.035282380878925323,
|
| 652 |
+
"g_norm": 2.870086461825849e-07,
|
| 653 |
+
"shape": [
|
| 654 |
+
768
|
| 655 |
+
]
|
| 656 |
+
},
|
| 657 |
+
"enc_attns.5.qkv.weight": {
|
| 658 |
+
"w_norm": 27.845542907714844,
|
| 659 |
+
"g_norm": 1.6994632460409775e-05,
|
| 660 |
+
"shape": [
|
| 661 |
+
2304,
|
| 662 |
+
768,
|
| 663 |
+
1,
|
| 664 |
+
1
|
| 665 |
+
]
|
| 666 |
+
},
|
| 667 |
+
"enc_attns.5.qkv.bias": {
|
| 668 |
+
"w_norm": 0.9989751577377319,
|
| 669 |
+
"g_norm": 4.274110381174978e-07,
|
| 670 |
+
"shape": [
|
| 671 |
+
2304
|
| 672 |
+
]
|
| 673 |
+
},
|
| 674 |
+
"enc_attns.5.proj.weight": {
|
| 675 |
+
"w_norm": 16.010074615478516,
|
| 676 |
+
"g_norm": 1.648172110435553e-05,
|
| 677 |
+
"shape": [
|
| 678 |
+
768,
|
| 679 |
+
768,
|
| 680 |
+
1,
|
| 681 |
+
1
|
| 682 |
+
]
|
| 683 |
+
},
|
| 684 |
+
"enc_attns.5.proj.bias": {
|
| 685 |
+
"w_norm": 0.5966185331344604,
|
| 686 |
+
"g_norm": 7.092186251611565e-07,
|
| 687 |
+
"shape": [
|
| 688 |
+
768
|
| 689 |
+
]
|
| 690 |
+
},
|
| 691 |
+
"enc_downs.0.op.weight": {
|
| 692 |
+
"w_norm": 8.130995750427246,
|
| 693 |
+
"g_norm": 0.0033071700017899275,
|
| 694 |
+
"shape": [
|
| 695 |
+
192,
|
| 696 |
+
192,
|
| 697 |
+
3,
|
| 698 |
+
3
|
| 699 |
+
]
|
| 700 |
+
},
|
| 701 |
+
"enc_downs.0.op.bias": {
|
| 702 |
+
"w_norm": 0.18532414734363556,
|
| 703 |
+
"g_norm": 0.00010940373613266274,
|
| 704 |
+
"shape": [
|
| 705 |
+
192
|
| 706 |
+
]
|
| 707 |
+
},
|
| 708 |
+
"enc_downs.1.op.weight": {
|
| 709 |
+
"w_norm": 8.075477600097656,
|
| 710 |
+
"g_norm": 0.0010724548483267426,
|
| 711 |
+
"shape": [
|
| 712 |
+
192,
|
| 713 |
+
192,
|
| 714 |
+
3,
|
| 715 |
+
3
|
| 716 |
+
]
|
| 717 |
+
},
|
| 718 |
+
"enc_downs.1.op.bias": {
|
| 719 |
+
"w_norm": 0.19328613579273224,
|
| 720 |
+
"g_norm": 2.2360094590112567e-05,
|
| 721 |
+
"shape": [
|
| 722 |
+
192
|
| 723 |
+
]
|
| 724 |
+
},
|
| 725 |
+
"enc_downs.2.op.weight": {
|
| 726 |
+
"w_norm": 11.434945106506348,
|
| 727 |
+
"g_norm": 0.000316201476380229,
|
| 728 |
+
"shape": [
|
| 729 |
+
384,
|
| 730 |
+
384,
|
| 731 |
+
3,
|
| 732 |
+
3
|
| 733 |
+
]
|
| 734 |
+
},
|
| 735 |
+
"enc_downs.2.op.bias": {
|
| 736 |
+
"w_norm": 0.18795061111450195,
|
| 737 |
+
"g_norm": 7.141250534914434e-06,
|
| 738 |
+
"shape": [
|
| 739 |
+
384
|
| 740 |
+
]
|
| 741 |
+
},
|
| 742 |
+
"enc_downs.3.op.weight": {
|
| 743 |
+
"w_norm": 11.38402271270752,
|
| 744 |
+
"g_norm": 0.00017456771456636488,
|
| 745 |
+
"shape": [
|
| 746 |
+
384,
|
| 747 |
+
384,
|
| 748 |
+
3,
|
| 749 |
+
3
|
| 750 |
+
]
|
| 751 |
+
},
|
| 752 |
+
"enc_downs.3.op.bias": {
|
| 753 |
+
"w_norm": 0.1932166963815689,
|
| 754 |
+
"g_norm": 1.5802761481609195e-06,
|
| 755 |
+
"shape": [
|
| 756 |
+
384
|
| 757 |
+
]
|
| 758 |
+
},
|
| 759 |
+
"enc_downs.4.op.weight": {
|
| 760 |
+
"w_norm": 16.206579208374023,
|
| 761 |
+
"g_norm": 9.180299821309745e-05,
|
| 762 |
+
"shape": [
|
| 763 |
+
768,
|
| 764 |
+
768,
|
| 765 |
+
3,
|
| 766 |
+
3
|
| 767 |
+
]
|
| 768 |
+
},
|
| 769 |
+
"enc_downs.4.op.bias": {
|
| 770 |
+
"w_norm": 0.19638435542583466,
|
| 771 |
+
"g_norm": 7.905038046374102e-07,
|
| 772 |
+
"shape": [
|
| 773 |
+
768
|
| 774 |
+
]
|
| 775 |
+
},
|
| 776 |
+
"mid1.norm1.weight": {
|
| 777 |
+
"w_norm": 27.71148681640625,
|
| 778 |
+
"g_norm": 2.901707603086834e-07,
|
| 779 |
+
"shape": [
|
| 780 |
+
768
|
| 781 |
+
]
|
| 782 |
+
},
|
| 783 |
+
"mid1.norm1.bias": {
|
| 784 |
+
"w_norm": 0.04520085081458092,
|
| 785 |
+
"g_norm": 3.0601083267356444e-07,
|
| 786 |
+
"shape": [
|
| 787 |
+
768
|
| 788 |
+
]
|
| 789 |
+
},
|
| 790 |
+
"mid1.conv1.weight": {
|
| 791 |
+
"w_norm": 16.51447296142578,
|
| 792 |
+
"g_norm": 8.964741937234066e-06,
|
| 793 |
+
"shape": [
|
| 794 |
+
768,
|
| 795 |
+
768,
|
| 796 |
+
3,
|
| 797 |
+
3
|
| 798 |
+
]
|
| 799 |
+
},
|
| 800 |
+
"mid1.conv1.bias": {
|
| 801 |
+
"w_norm": 0.20231729745864868,
|
| 802 |
+
"g_norm": 2.1629281832247216e-07,
|
| 803 |
+
"shape": [
|
| 804 |
+
768
|
| 805 |
+
]
|
| 806 |
+
},
|
| 807 |
+
"mid1.t_proj.weight": {
|
| 808 |
+
"w_norm": 16.008007049560547,
|
| 809 |
+
"g_norm": 2.066750681706253e-07,
|
| 810 |
+
"shape": [
|
| 811 |
+
768,
|
| 812 |
+
256
|
| 813 |
+
]
|
| 814 |
+
},
|
| 815 |
+
"mid1.t_proj.bias": {
|
| 816 |
+
"w_norm": 1.0318657159805298,
|
| 817 |
+
"g_norm": 2.1627558055570262e-07,
|
| 818 |
+
"shape": [
|
| 819 |
+
768
|
| 820 |
+
]
|
| 821 |
+
},
|
| 822 |
+
"mid1.norm2.weight": {
|
| 823 |
+
"w_norm": 27.610170364379883,
|
| 824 |
+
"g_norm": 2.3066324672527116e-07,
|
| 825 |
+
"shape": [
|
| 826 |
+
768
|
| 827 |
+
]
|
| 828 |
+
},
|
| 829 |
+
"mid1.norm2.bias": {
|
| 830 |
+
"w_norm": 0.12007729709148407,
|
| 831 |
+
"g_norm": 3.1912145459500607e-07,
|
| 832 |
+
"shape": [
|
| 833 |
+
768
|
| 834 |
+
]
|
| 835 |
+
},
|
| 836 |
+
"mid1.conv2.weight": {
|
| 837 |
+
"w_norm": 16.278959274291992,
|
| 838 |
+
"g_norm": 1.5227375115500763e-05,
|
| 839 |
+
"shape": [
|
| 840 |
+
768,
|
| 841 |
+
768,
|
| 842 |
+
3,
|
| 843 |
+
3
|
| 844 |
+
]
|
| 845 |
+
},
|
| 846 |
+
"mid1.conv2.bias": {
|
| 847 |
+
"w_norm": 0.19529534876346588,
|
| 848 |
+
"g_norm": 4.908901587441505e-07,
|
| 849 |
+
"shape": [
|
| 850 |
+
768
|
| 851 |
+
]
|
| 852 |
+
},
|
| 853 |
+
"mid_attn.norm.weight": {
|
| 854 |
+
"w_norm": 27.68242835998535,
|
| 855 |
+
"g_norm": 1.5732706515336758e-07,
|
| 856 |
+
"shape": [
|
| 857 |
+
768
|
| 858 |
+
]
|
| 859 |
+
},
|
| 860 |
+
"mid_attn.norm.bias": {
|
| 861 |
+
"w_norm": 0.03362787514925003,
|
| 862 |
+
"g_norm": 1.7879504810025537e-07,
|
| 863 |
+
"shape": [
|
| 864 |
+
768
|
| 865 |
+
]
|
| 866 |
+
},
|
| 867 |
+
"mid_attn.qkv.weight": {
|
| 868 |
+
"w_norm": 27.83576011657715,
|
| 869 |
+
"g_norm": 6.8316489887365606e-06,
|
| 870 |
+
"shape": [
|
| 871 |
+
2304,
|
| 872 |
+
768,
|
| 873 |
+
1,
|
| 874 |
+
1
|
| 875 |
+
]
|
| 876 |
+
},
|
| 877 |
+
"mid_attn.qkv.bias": {
|
| 878 |
+
"w_norm": 0.9854804277420044,
|
| 879 |
+
"g_norm": 2.807340706567629e-07,
|
| 880 |
+
"shape": [
|
| 881 |
+
2304
|
| 882 |
+
]
|
| 883 |
+
},
|
| 884 |
+
"mid_attn.proj.weight": {
|
| 885 |
+
"w_norm": 16.009660720825195,
|
| 886 |
+
"g_norm": 6.5837921283673495e-06,
|
| 887 |
+
"shape": [
|
| 888 |
+
768,
|
| 889 |
+
768,
|
| 890 |
+
1,
|
| 891 |
+
1
|
| 892 |
+
]
|
| 893 |
+
},
|
| 894 |
+
"mid_attn.proj.bias": {
|
| 895 |
+
"w_norm": 0.5765897631645203,
|
| 896 |
+
"g_norm": 4.809566576113866e-07,
|
| 897 |
+
"shape": [
|
| 898 |
+
768
|
| 899 |
+
]
|
| 900 |
+
},
|
| 901 |
+
"mid2.norm1.weight": {
|
| 902 |
+
"w_norm": 27.716794967651367,
|
| 903 |
+
"g_norm": 1.7296281384915346e-07,
|
| 904 |
+
"shape": [
|
| 905 |
+
768
|
| 906 |
+
]
|
| 907 |
+
},
|
| 908 |
+
"mid2.norm1.bias": {
|
| 909 |
+
"w_norm": 0.05131439492106438,
|
| 910 |
+
"g_norm": 1.5387250584808498e-07,
|
| 911 |
+
"shape": [
|
| 912 |
+
768
|
| 913 |
+
]
|
| 914 |
+
},
|
| 915 |
+
"mid2.conv1.weight": {
|
| 916 |
+
"w_norm": 16.50710105895996,
|
| 917 |
+
"g_norm": 7.962901690916624e-06,
|
| 918 |
+
"shape": [
|
| 919 |
+
768,
|
| 920 |
+
768,
|
| 921 |
+
3,
|
| 922 |
+
3
|
| 923 |
+
]
|
| 924 |
+
},
|
| 925 |
+
"mid2.conv1.bias": {
|
| 926 |
+
"w_norm": 0.19801872968673706,
|
| 927 |
+
"g_norm": 1.5266020625404053e-07,
|
| 928 |
+
"shape": [
|
| 929 |
+
768
|
| 930 |
+
]
|
| 931 |
+
},
|
| 932 |
+
"mid2.t_proj.weight": {
|
| 933 |
+
"w_norm": 15.995810508728027,
|
| 934 |
+
"g_norm": 2.089972923613459e-07,
|
| 935 |
+
"shape": [
|
| 936 |
+
768,
|
| 937 |
+
256
|
| 938 |
+
]
|
| 939 |
+
},
|
| 940 |
+
"mid2.t_proj.bias": {
|
| 941 |
+
"w_norm": 0.9966025352478027,
|
| 942 |
+
"g_norm": 1.5270575204340275e-07,
|
| 943 |
+
"shape": [
|
| 944 |
+
768
|
| 945 |
+
]
|
| 946 |
+
},
|
| 947 |
+
"mid2.norm2.weight": {
|
| 948 |
+
"w_norm": 27.59794807434082,
|
| 949 |
+
"g_norm": 1.9348948399056098e-07,
|
| 950 |
+
"shape": [
|
| 951 |
+
768
|
| 952 |
+
]
|
| 953 |
+
},
|
| 954 |
+
"mid2.norm2.bias": {
|
| 955 |
+
"w_norm": 0.1285128891468048,
|
| 956 |
+
"g_norm": 2.489040582531743e-07,
|
| 957 |
+
"shape": [
|
| 958 |
+
768
|
| 959 |
+
]
|
| 960 |
+
},
|
| 961 |
+
"mid2.conv2.weight": {
|
| 962 |
+
"w_norm": 16.23427963256836,
|
| 963 |
+
"g_norm": 1.610977233212907e-05,
|
| 964 |
+
"shape": [
|
| 965 |
+
768,
|
| 966 |
+
768,
|
| 967 |
+
3,
|
| 968 |
+
3
|
| 969 |
+
]
|
| 970 |
+
},
|
| 971 |
+
"mid2.conv2.bias": {
|
| 972 |
+
"w_norm": 0.18982110917568207,
|
| 973 |
+
"g_norm": 4.8388949380751e-07,
|
| 974 |
+
"shape": [
|
| 975 |
+
768
|
| 976 |
+
]
|
| 977 |
+
},
|
| 978 |
+
"dec_blocks.0.norm1.weight": {
|
| 979 |
+
"w_norm": 39.197998046875,
|
| 980 |
+
"g_norm": 3.1209330586534634e-07,
|
| 981 |
+
"shape": [
|
| 982 |
+
1536
|
| 983 |
+
]
|
| 984 |
+
},
|
| 985 |
+
"dec_blocks.0.norm1.bias": {
|
| 986 |
+
"w_norm": 0.06604019552469254,
|
| 987 |
+
"g_norm": 2.441701383304462e-07,
|
| 988 |
+
"shape": [
|
| 989 |
+
1536
|
| 990 |
+
]
|
| 991 |
+
},
|
| 992 |
+
"dec_blocks.0.conv1.weight": {
|
| 993 |
+
"w_norm": 17.042268753051758,
|
| 994 |
+
"g_norm": 1.324670301983133e-05,
|
| 995 |
+
"shape": [
|
| 996 |
+
768,
|
| 997 |
+
1536,
|
| 998 |
+
3,
|
| 999 |
+
3
|
| 1000 |
+
]
|
| 1001 |
+
},
|
| 1002 |
+
"dec_blocks.0.conv1.bias": {
|
| 1003 |
+
"w_norm": 0.14318080246448517,
|
| 1004 |
+
"g_norm": 1.5126960306588444e-07,
|
| 1005 |
+
"shape": [
|
| 1006 |
+
768
|
| 1007 |
+
]
|
| 1008 |
+
},
|
| 1009 |
+
"dec_blocks.0.t_proj.weight": {
|
| 1010 |
+
"w_norm": 16.01215171813965,
|
| 1011 |
+
"g_norm": 1.9884710411588458e-07,
|
| 1012 |
+
"shape": [
|
| 1013 |
+
768,
|
| 1014 |
+
256
|
| 1015 |
+
]
|
| 1016 |
+
},
|
| 1017 |
+
"dec_blocks.0.t_proj.bias": {
|
| 1018 |
+
"w_norm": 1.0212759971618652,
|
| 1019 |
+
"g_norm": 1.5122640206755023e-07,
|
| 1020 |
+
"shape": [
|
| 1021 |
+
768
|
| 1022 |
+
]
|
| 1023 |
+
},
|
| 1024 |
+
"dec_blocks.0.norm2.weight": {
|
| 1025 |
+
"w_norm": 27.559242248535156,
|
| 1026 |
+
"g_norm": 3.4676895666052587e-07,
|
| 1027 |
+
"shape": [
|
| 1028 |
+
768
|
| 1029 |
+
]
|
| 1030 |
+
},
|
| 1031 |
+
"dec_blocks.0.norm2.bias": {
|
| 1032 |
+
"w_norm": 0.16142374277114868,
|
| 1033 |
+
"g_norm": 4.399703641411179e-07,
|
| 1034 |
+
"shape": [
|
| 1035 |
+
768
|
| 1036 |
+
]
|
| 1037 |
+
},
|
| 1038 |
+
"dec_blocks.0.conv2.weight": {
|
| 1039 |
+
"w_norm": 16.135143280029297,
|
| 1040 |
+
"g_norm": 3.1659994419896975e-05,
|
| 1041 |
+
"shape": [
|
| 1042 |
+
768,
|
| 1043 |
+
768,
|
| 1044 |
+
3,
|
| 1045 |
+
3
|
| 1046 |
+
]
|
| 1047 |
+
},
|
| 1048 |
+
"dec_blocks.0.conv2.bias": {
|
| 1049 |
+
"w_norm": 0.19768120348453522,
|
| 1050 |
+
"g_norm": 1.10852681700635e-06,
|
| 1051 |
+
"shape": [
|
| 1052 |
+
768
|
| 1053 |
+
]
|
| 1054 |
+
},
|
| 1055 |
+
"dec_blocks.0.skip.weight": {
|
| 1056 |
+
"w_norm": 15.865204811096191,
|
| 1057 |
+
"g_norm": 0.00012172277638455853,
|
| 1058 |
+
"shape": [
|
| 1059 |
+
768,
|
| 1060 |
+
1536,
|
| 1061 |
+
1,
|
| 1062 |
+
1
|
| 1063 |
+
]
|
| 1064 |
+
},
|
| 1065 |
+
"dec_blocks.0.skip.bias": {
|
| 1066 |
+
"w_norm": 0.4107856750488281,
|
| 1067 |
+
"g_norm": 1.10852681700635e-06,
|
| 1068 |
+
"shape": [
|
| 1069 |
+
768
|
| 1070 |
+
]
|
| 1071 |
+
},
|
| 1072 |
+
"dec_blocks.1.norm1.weight": {
|
| 1073 |
+
"w_norm": 39.19361877441406,
|
| 1074 |
+
"g_norm": 1.6795871715657995e-06,
|
| 1075 |
+
"shape": [
|
| 1076 |
+
1536
|
| 1077 |
+
]
|
| 1078 |
+
},
|
| 1079 |
+
"dec_blocks.1.norm1.bias": {
|
| 1080 |
+
"w_norm": 0.05957057327032089,
|
| 1081 |
+
"g_norm": 1.083103825294529e-06,
|
| 1082 |
+
"shape": [
|
| 1083 |
+
1536
|
| 1084 |
+
]
|
| 1085 |
+
},
|
| 1086 |
+
"dec_blocks.1.conv1.weight": {
|
| 1087 |
+
"w_norm": 17.069610595703125,
|
| 1088 |
+
"g_norm": 4.9241381930187345e-05,
|
| 1089 |
+
"shape": [
|
| 1090 |
+
768,
|
| 1091 |
+
1536,
|
| 1092 |
+
3,
|
| 1093 |
+
3
|
| 1094 |
+
]
|
| 1095 |
+
},
|
| 1096 |
+
"dec_blocks.1.conv1.bias": {
|
| 1097 |
+
"w_norm": 0.1401885598897934,
|
| 1098 |
+
"g_norm": 5.735878403356764e-07,
|
| 1099 |
+
"shape": [
|
| 1100 |
+
768
|
| 1101 |
+
]
|
| 1102 |
+
},
|
| 1103 |
+
"dec_blocks.1.t_proj.weight": {
|
| 1104 |
+
"w_norm": 16.04375457763672,
|
| 1105 |
+
"g_norm": 8.650154086353723e-07,
|
| 1106 |
+
"shape": [
|
| 1107 |
+
768,
|
| 1108 |
+
256
|
| 1109 |
+
]
|
| 1110 |
+
},
|
| 1111 |
+
"dec_blocks.1.t_proj.bias": {
|
| 1112 |
+
"w_norm": 0.9863389730453491,
|
| 1113 |
+
"g_norm": 5.737327342103526e-07,
|
| 1114 |
+
"shape": [
|
| 1115 |
+
768
|
| 1116 |
+
]
|
| 1117 |
+
},
|
| 1118 |
+
"dec_blocks.1.norm2.weight": {
|
| 1119 |
+
"w_norm": 27.54302215576172,
|
| 1120 |
+
"g_norm": 1.7761914250513655e-06,
|
| 1121 |
+
"shape": [
|
| 1122 |
+
768
|
| 1123 |
+
]
|
| 1124 |
+
},
|
| 1125 |
+
"dec_blocks.1.norm2.bias": {
|
| 1126 |
+
"w_norm": 0.19749881327152252,
|
| 1127 |
+
"g_norm": 2.196136847487651e-06,
|
| 1128 |
+
"shape": [
|
| 1129 |
+
768
|
| 1130 |
+
]
|
| 1131 |
+
},
|
| 1132 |
+
"dec_blocks.1.conv2.weight": {
|
| 1133 |
+
"w_norm": 16.147022247314453,
|
| 1134 |
+
"g_norm": 0.00012046505435137078,
|
| 1135 |
+
"shape": [
|
| 1136 |
+
768,
|
| 1137 |
+
768,
|
| 1138 |
+
3,
|
| 1139 |
+
3
|
| 1140 |
+
]
|
| 1141 |
+
},
|
| 1142 |
+
"dec_blocks.1.conv2.bias": {
|
| 1143 |
+
"w_norm": 0.19546672701835632,
|
| 1144 |
+
"g_norm": 3.8538873923243955e-06,
|
| 1145 |
+
"shape": [
|
| 1146 |
+
768
|
| 1147 |
+
]
|
| 1148 |
+
},
|
| 1149 |
+
"dec_blocks.1.skip.weight": {
|
| 1150 |
+
"w_norm": 15.8510160446167,
|
| 1151 |
+
"g_norm": 0.00025108447880484164,
|
| 1152 |
+
"shape": [
|
| 1153 |
+
768,
|
| 1154 |
+
1536,
|
| 1155 |
+
1,
|
| 1156 |
+
1
|
| 1157 |
+
]
|
| 1158 |
+
},
|
| 1159 |
+
"dec_blocks.1.skip.bias": {
|
| 1160 |
+
"w_norm": 0.4066181778907776,
|
| 1161 |
+
"g_norm": 3.8538873923243955e-06,
|
| 1162 |
+
"shape": [
|
| 1163 |
+
768
|
| 1164 |
+
]
|
| 1165 |
+
},
|
| 1166 |
+
"dec_blocks.2.norm1.weight": {
|
| 1167 |
+
"w_norm": 33.947349548339844,
|
| 1168 |
+
"g_norm": 2.5137094326055376e-06,
|
| 1169 |
+
"shape": [
|
| 1170 |
+
1152
|
| 1171 |
+
]
|
| 1172 |
+
},
|
| 1173 |
+
"dec_blocks.2.norm1.bias": {
|
| 1174 |
+
"w_norm": 0.06189341843128204,
|
| 1175 |
+
"g_norm": 2.2232970877666958e-06,
|
| 1176 |
+
"shape": [
|
| 1177 |
+
1152
|
| 1178 |
+
]
|
| 1179 |
+
},
|
| 1180 |
+
"dec_blocks.2.conv1.weight": {
|
| 1181 |
+
"w_norm": 12.084885597229004,
|
| 1182 |
+
"g_norm": 0.00012408540351316333,
|
| 1183 |
+
"shape": [
|
| 1184 |
+
384,
|
| 1185 |
+
1152,
|
| 1186 |
+
3,
|
| 1187 |
+
3
|
| 1188 |
+
]
|
| 1189 |
+
},
|
| 1190 |
+
"dec_blocks.2.conv1.bias": {
|
| 1191 |
+
"w_norm": 0.1179933249950409,
|
| 1192 |
+
"g_norm": 2.0747602320625447e-06,
|
| 1193 |
+
"shape": [
|
| 1194 |
+
384
|
| 1195 |
+
]
|
| 1196 |
+
},
|
| 1197 |
+
"dec_blocks.2.t_proj.weight": {
|
| 1198 |
+
"w_norm": 11.32999038696289,
|
| 1199 |
+
"g_norm": 3.524335625115782e-06,
|
| 1200 |
+
"shape": [
|
| 1201 |
+
384,
|
| 1202 |
+
256
|
| 1203 |
+
]
|
| 1204 |
+
},
|
| 1205 |
+
"dec_blocks.2.t_proj.bias": {
|
| 1206 |
+
"w_norm": 0.7094293236732483,
|
| 1207 |
+
"g_norm": 2.0747856979141943e-06,
|
| 1208 |
+
"shape": [
|
| 1209 |
+
384
|
| 1210 |
+
]
|
| 1211 |
+
},
|
| 1212 |
+
"dec_blocks.2.norm2.weight": {
|
| 1213 |
+
"w_norm": 19.48737335205078,
|
| 1214 |
+
"g_norm": 3.282286343164742e-06,
|
| 1215 |
+
"shape": [
|
| 1216 |
+
384
|
| 1217 |
+
]
|
| 1218 |
+
},
|
| 1219 |
+
"dec_blocks.2.norm2.bias": {
|
| 1220 |
+
"w_norm": 0.131941020488739,
|
| 1221 |
+
"g_norm": 5.007308573112823e-06,
|
| 1222 |
+
"shape": [
|
| 1223 |
+
384
|
| 1224 |
+
]
|
| 1225 |
+
},
|
| 1226 |
+
"dec_blocks.2.conv2.weight": {
|
| 1227 |
+
"w_norm": 11.395310401916504,
|
| 1228 |
+
"g_norm": 0.00022668339079245925,
|
| 1229 |
+
"shape": [
|
| 1230 |
+
384,
|
| 1231 |
+
384,
|
| 1232 |
+
3,
|
| 1233 |
+
3
|
| 1234 |
+
]
|
| 1235 |
+
},
|
| 1236 |
+
"dec_blocks.2.conv2.bias": {
|
| 1237 |
+
"w_norm": 0.18954412639141083,
|
| 1238 |
+
"g_norm": 1.244637678610161e-05,
|
| 1239 |
+
"shape": [
|
| 1240 |
+
384
|
| 1241 |
+
]
|
| 1242 |
+
},
|
| 1243 |
+
"dec_blocks.2.skip.weight": {
|
| 1244 |
+
"w_norm": 11.1150541305542,
|
| 1245 |
+
"g_norm": 0.0005784353706985712,
|
| 1246 |
+
"shape": [
|
| 1247 |
+
384,
|
| 1248 |
+
1152,
|
| 1249 |
+
1,
|
| 1250 |
+
1
|
| 1251 |
+
]
|
| 1252 |
+
},
|
| 1253 |
+
"dec_blocks.2.skip.bias": {
|
| 1254 |
+
"w_norm": 0.3310128450393677,
|
| 1255 |
+
"g_norm": 1.244637678610161e-05,
|
| 1256 |
+
"shape": [
|
| 1257 |
+
384
|
| 1258 |
+
]
|
| 1259 |
+
},
|
| 1260 |
+
"dec_blocks.3.norm1.weight": {
|
| 1261 |
+
"w_norm": 27.713695526123047,
|
| 1262 |
+
"g_norm": 1.6070203855633736e-05,
|
| 1263 |
+
"shape": [
|
| 1264 |
+
768
|
| 1265 |
+
]
|
| 1266 |
+
},
|
| 1267 |
+
"dec_blocks.3.norm1.bias": {
|
| 1268 |
+
"w_norm": 0.06734566390514374,
|
| 1269 |
+
"g_norm": 1.080029960576212e-05,
|
| 1270 |
+
"shape": [
|
| 1271 |
+
768
|
| 1272 |
+
]
|
| 1273 |
+
},
|
| 1274 |
+
"dec_blocks.3.conv1.weight": {
|
| 1275 |
+
"w_norm": 11.990684509277344,
|
| 1276 |
+
"g_norm": 0.0005816977354697883,
|
| 1277 |
+
"shape": [
|
| 1278 |
+
384,
|
| 1279 |
+
768,
|
| 1280 |
+
3,
|
| 1281 |
+
3
|
| 1282 |
+
]
|
| 1283 |
+
},
|
| 1284 |
+
"dec_blocks.3.conv1.bias": {
|
| 1285 |
+
"w_norm": 0.14317339658737183,
|
| 1286 |
+
"g_norm": 9.035586117533967e-06,
|
| 1287 |
+
"shape": [
|
| 1288 |
+
384
|
| 1289 |
+
]
|
| 1290 |
+
},
|
| 1291 |
+
"dec_blocks.3.t_proj.weight": {
|
| 1292 |
+
"w_norm": 11.34940242767334,
|
| 1293 |
+
"g_norm": 2.0452938770176843e-05,
|
| 1294 |
+
"shape": [
|
| 1295 |
+
384,
|
| 1296 |
+
256
|
| 1297 |
+
]
|
| 1298 |
+
},
|
| 1299 |
+
"dec_blocks.3.t_proj.bias": {
|
| 1300 |
+
"w_norm": 0.7089352011680603,
|
| 1301 |
+
"g_norm": 9.036564733833075e-06,
|
| 1302 |
+
"shape": [
|
| 1303 |
+
384
|
| 1304 |
+
]
|
| 1305 |
+
},
|
| 1306 |
+
"dec_blocks.3.norm2.weight": {
|
| 1307 |
+
"w_norm": 19.466581344604492,
|
| 1308 |
+
"g_norm": 1.5670944776502438e-05,
|
| 1309 |
+
"shape": [
|
| 1310 |
+
384
|
| 1311 |
+
]
|
| 1312 |
+
},
|
| 1313 |
+
"dec_blocks.3.norm2.bias": {
|
| 1314 |
+
"w_norm": 0.1333981603384018,
|
| 1315 |
+
"g_norm": 1.8773333067656495e-05,
|
| 1316 |
+
"shape": [
|
| 1317 |
+
384
|
| 1318 |
+
]
|
| 1319 |
+
},
|
| 1320 |
+
"dec_blocks.3.conv2.weight": {
|
| 1321 |
+
"w_norm": 11.425914764404297,
|
| 1322 |
+
"g_norm": 0.0009791628690436482,
|
| 1323 |
+
"shape": [
|
| 1324 |
+
384,
|
| 1325 |
+
384,
|
| 1326 |
+
3,
|
| 1327 |
+
3
|
| 1328 |
+
]
|
| 1329 |
+
},
|
| 1330 |
+
"dec_blocks.3.conv2.bias": {
|
| 1331 |
+
"w_norm": 0.19313561916351318,
|
| 1332 |
+
"g_norm": 4.651550989365205e-05,
|
| 1333 |
+
"shape": [
|
| 1334 |
+
384
|
| 1335 |
+
]
|
| 1336 |
+
},
|
| 1337 |
+
"dec_blocks.3.skip.weight": {
|
| 1338 |
+
"w_norm": 11.149859428405762,
|
| 1339 |
+
"g_norm": 0.0015515109989792109,
|
| 1340 |
+
"shape": [
|
| 1341 |
+
384,
|
| 1342 |
+
768,
|
| 1343 |
+
1,
|
| 1344 |
+
1
|
| 1345 |
+
]
|
| 1346 |
+
},
|
| 1347 |
+
"dec_blocks.3.skip.bias": {
|
| 1348 |
+
"w_norm": 0.4057067632675171,
|
| 1349 |
+
"g_norm": 4.651550989365205e-05,
|
| 1350 |
+
"shape": [
|
| 1351 |
+
384
|
| 1352 |
+
]
|
| 1353 |
+
},
|
| 1354 |
+
"dec_blocks.4.norm1.weight": {
|
| 1355 |
+
"w_norm": 23.98821258544922,
|
| 1356 |
+
"g_norm": 5.4541109420824796e-05,
|
| 1357 |
+
"shape": [
|
| 1358 |
+
576
|
| 1359 |
+
]
|
| 1360 |
+
},
|
| 1361 |
+
"dec_blocks.4.norm1.bias": {
|
| 1362 |
+
"w_norm": 0.06465596705675125,
|
| 1363 |
+
"g_norm": 4.13734233006835e-05,
|
| 1364 |
+
"shape": [
|
| 1365 |
+
576
|
| 1366 |
+
]
|
| 1367 |
+
},
|
| 1368 |
+
"dec_blocks.4.conv1.weight": {
|
| 1369 |
+
"w_norm": 8.374838829040527,
|
| 1370 |
+
"g_norm": 0.001791031681932509,
|
| 1371 |
+
"shape": [
|
| 1372 |
+
192,
|
| 1373 |
+
576,
|
| 1374 |
+
3,
|
| 1375 |
+
3
|
| 1376 |
+
]
|
| 1377 |
+
},
|
| 1378 |
+
"dec_blocks.4.conv1.bias": {
|
| 1379 |
+
"w_norm": 0.11661636829376221,
|
| 1380 |
+
"g_norm": 4.558874206850305e-05,
|
| 1381 |
+
"shape": [
|
| 1382 |
+
192
|
| 1383 |
+
]
|
| 1384 |
+
},
|
| 1385 |
+
"dec_blocks.4.t_proj.weight": {
|
| 1386 |
+
"w_norm": 8.018696784973145,
|
| 1387 |
+
"g_norm": 6.551636761287227e-05,
|
| 1388 |
+
"shape": [
|
| 1389 |
+
192,
|
| 1390 |
+
256
|
| 1391 |
+
]
|
| 1392 |
+
},
|
| 1393 |
+
"dec_blocks.4.t_proj.bias": {
|
| 1394 |
+
"w_norm": 0.4810430109500885,
|
| 1395 |
+
"g_norm": 4.55839435744565e-05,
|
| 1396 |
+
"shape": [
|
| 1397 |
+
192
|
| 1398 |
+
]
|
| 1399 |
+
},
|
| 1400 |
+
"dec_blocks.4.norm2.weight": {
|
| 1401 |
+
"w_norm": 13.801824569702148,
|
| 1402 |
+
"g_norm": 9.073015098692849e-05,
|
| 1403 |
+
"shape": [
|
| 1404 |
+
192
|
| 1405 |
+
]
|
| 1406 |
+
},
|
| 1407 |
+
"dec_blocks.4.norm2.bias": {
|
| 1408 |
+
"w_norm": 0.05389580875635147,
|
| 1409 |
+
"g_norm": 0.00011636758426902816,
|
| 1410 |
+
"shape": [
|
| 1411 |
+
192
|
| 1412 |
+
]
|
| 1413 |
+
},
|
| 1414 |
+
"dec_blocks.4.conv2.weight": {
|
| 1415 |
+
"w_norm": 8.050837516784668,
|
| 1416 |
+
"g_norm": 0.004189273342490196,
|
| 1417 |
+
"shape": [
|
| 1418 |
+
192,
|
| 1419 |
+
192,
|
| 1420 |
+
3,
|
| 1421 |
+
3
|
| 1422 |
+
]
|
| 1423 |
+
},
|
| 1424 |
+
"dec_blocks.4.conv2.bias": {
|
| 1425 |
+
"w_norm": 0.19405966997146606,
|
| 1426 |
+
"g_norm": 0.0002257868618471548,
|
| 1427 |
+
"shape": [
|
| 1428 |
+
192
|
| 1429 |
+
]
|
| 1430 |
+
},
|
| 1431 |
+
"dec_blocks.4.skip.weight": {
|
| 1432 |
+
"w_norm": 7.899153709411621,
|
| 1433 |
+
"g_norm": 0.004444679711014032,
|
| 1434 |
+
"shape": [
|
| 1435 |
+
192,
|
| 1436 |
+
576,
|
| 1437 |
+
1,
|
| 1438 |
+
1
|
| 1439 |
+
]
|
| 1440 |
+
},
|
| 1441 |
+
"dec_blocks.4.skip.bias": {
|
| 1442 |
+
"w_norm": 0.34669438004493713,
|
| 1443 |
+
"g_norm": 0.0002257868618471548,
|
| 1444 |
+
"shape": [
|
| 1445 |
+
192
|
| 1446 |
+
]
|
| 1447 |
+
},
|
| 1448 |
+
"dec_blocks.5.norm1.weight": {
|
| 1449 |
+
"w_norm": 19.55978775024414,
|
| 1450 |
+
"g_norm": 0.0025432000402361155,
|
| 1451 |
+
"shape": [
|
| 1452 |
+
384
|
| 1453 |
+
]
|
| 1454 |
+
},
|
| 1455 |
+
"dec_blocks.5.norm1.bias": {
|
| 1456 |
+
"w_norm": 0.08432969450950623,
|
| 1457 |
+
"g_norm": 0.0032189551275223494,
|
| 1458 |
+
"shape": [
|
| 1459 |
+
384
|
| 1460 |
+
]
|
| 1461 |
+
},
|
| 1462 |
+
"dec_blocks.5.conv1.weight": {
|
| 1463 |
+
"w_norm": 8.425467491149902,
|
| 1464 |
+
"g_norm": 0.04127212241292,
|
| 1465 |
+
"shape": [
|
| 1466 |
+
192,
|
| 1467 |
+
384,
|
| 1468 |
+
3,
|
| 1469 |
+
3
|
| 1470 |
+
]
|
| 1471 |
+
},
|
| 1472 |
+
"dec_blocks.5.conv1.bias": {
|
| 1473 |
+
"w_norm": 0.13819772005081177,
|
| 1474 |
+
"g_norm": 0.0018730823649093509,
|
| 1475 |
+
"shape": [
|
| 1476 |
+
192
|
| 1477 |
+
]
|
| 1478 |
+
},
|
| 1479 |
+
"dec_blocks.5.t_proj.weight": {
|
| 1480 |
+
"w_norm": 8.143948554992676,
|
| 1481 |
+
"g_norm": 0.0016058351611718535,
|
| 1482 |
+
"shape": [
|
| 1483 |
+
192,
|
| 1484 |
+
256
|
| 1485 |
+
]
|
| 1486 |
+
},
|
| 1487 |
+
"dec_blocks.5.t_proj.bias": {
|
| 1488 |
+
"w_norm": 0.504915714263916,
|
| 1489 |
+
"g_norm": 0.0018739085644483566,
|
| 1490 |
+
"shape": [
|
| 1491 |
+
192
|
| 1492 |
+
]
|
| 1493 |
+
},
|
| 1494 |
+
"dec_blocks.5.norm2.weight": {
|
| 1495 |
+
"w_norm": 13.880186080932617,
|
| 1496 |
+
"g_norm": 0.002454320201650262,
|
| 1497 |
+
"shape": [
|
| 1498 |
+
192
|
| 1499 |
+
]
|
| 1500 |
+
},
|
| 1501 |
+
"dec_blocks.5.norm2.bias": {
|
| 1502 |
+
"w_norm": 0.03555542230606079,
|
| 1503 |
+
"g_norm": 0.002779952948912978,
|
| 1504 |
+
"shape": [
|
| 1505 |
+
192
|
| 1506 |
+
]
|
| 1507 |
+
},
|
| 1508 |
+
"dec_blocks.5.conv2.weight": {
|
| 1509 |
+
"w_norm": 8.265290260314941,
|
| 1510 |
+
"g_norm": 0.053190335631370544,
|
| 1511 |
+
"shape": [
|
| 1512 |
+
192,
|
| 1513 |
+
192,
|
| 1514 |
+
3,
|
| 1515 |
+
3
|
| 1516 |
+
]
|
| 1517 |
+
},
|
| 1518 |
+
"dec_blocks.5.conv2.bias": {
|
| 1519 |
+
"w_norm": 0.1995941400527954,
|
| 1520 |
+
"g_norm": 0.002878693165257573,
|
| 1521 |
+
"shape": [
|
| 1522 |
+
192
|
| 1523 |
+
]
|
| 1524 |
+
},
|
| 1525 |
+
"dec_blocks.5.skip.weight": {
|
| 1526 |
+
"w_norm": 8.083870887756348,
|
| 1527 |
+
"g_norm": 0.050536688417196274,
|
| 1528 |
+
"shape": [
|
| 1529 |
+
192,
|
| 1530 |
+
384,
|
| 1531 |
+
1,
|
| 1532 |
+
1
|
| 1533 |
+
]
|
| 1534 |
+
},
|
| 1535 |
+
"dec_blocks.5.skip.bias": {
|
| 1536 |
+
"w_norm": 0.42189013957977295,
|
| 1537 |
+
"g_norm": 0.002878693165257573,
|
| 1538 |
+
"shape": [
|
| 1539 |
+
192
|
| 1540 |
+
]
|
| 1541 |
+
},
|
| 1542 |
+
"dec_attns.0.norm.weight": {
|
| 1543 |
+
"w_norm": 27.65727424621582,
|
| 1544 |
+
"g_norm": 6.846094038337469e-07,
|
| 1545 |
+
"shape": [
|
| 1546 |
+
768
|
| 1547 |
+
]
|
| 1548 |
+
},
|
| 1549 |
+
"dec_attns.0.norm.bias": {
|
| 1550 |
+
"w_norm": 0.03050125576555729,
|
| 1551 |
+
"g_norm": 4.60043281691469e-07,
|
| 1552 |
+
"shape": [
|
| 1553 |
+
768
|
| 1554 |
+
]
|
| 1555 |
+
},
|
| 1556 |
+
"dec_attns.0.qkv.weight": {
|
| 1557 |
+
"w_norm": 27.862565994262695,
|
| 1558 |
+
"g_norm": 2.4649789338582195e-05,
|
| 1559 |
+
"shape": [
|
| 1560 |
+
2304,
|
| 1561 |
+
768,
|
| 1562 |
+
1,
|
| 1563 |
+
1
|
| 1564 |
+
]
|
| 1565 |
+
},
|
| 1566 |
+
"dec_attns.0.qkv.bias": {
|
| 1567 |
+
"w_norm": 1.0089398622512817,
|
| 1568 |
+
"g_norm": 6.871725872770185e-07,
|
| 1569 |
+
"shape": [
|
| 1570 |
+
2304
|
| 1571 |
+
]
|
| 1572 |
+
},
|
| 1573 |
+
"dec_attns.0.proj.weight": {
|
| 1574 |
+
"w_norm": 16.023706436157227,
|
| 1575 |
+
"g_norm": 2.3888020223239437e-05,
|
| 1576 |
+
"shape": [
|
| 1577 |
+
768,
|
| 1578 |
+
768,
|
| 1579 |
+
1,
|
| 1580 |
+
1
|
| 1581 |
+
]
|
| 1582 |
+
},
|
| 1583 |
+
"dec_attns.0.proj.bias": {
|
| 1584 |
+
"w_norm": 0.5768023133277893,
|
| 1585 |
+
"g_norm": 1.100436406886729e-06,
|
| 1586 |
+
"shape": [
|
| 1587 |
+
768
|
| 1588 |
+
]
|
| 1589 |
+
},
|
| 1590 |
+
"dec_attns.1.norm.weight": {
|
| 1591 |
+
"w_norm": 27.667020797729492,
|
| 1592 |
+
"g_norm": 1.563885575706081e-06,
|
| 1593 |
+
"shape": [
|
| 1594 |
+
768
|
| 1595 |
+
]
|
| 1596 |
+
},
|
| 1597 |
+
"dec_attns.1.norm.bias": {
|
| 1598 |
+
"w_norm": 0.033365633338689804,
|
| 1599 |
+
"g_norm": 1.447783574803907e-06,
|
| 1600 |
+
"shape": [
|
| 1601 |
+
768
|
| 1602 |
+
]
|
| 1603 |
+
},
|
| 1604 |
+
"dec_attns.1.qkv.weight": {
|
| 1605 |
+
"w_norm": 27.891311645507812,
|
| 1606 |
+
"g_norm": 6.250386650208384e-05,
|
| 1607 |
+
"shape": [
|
| 1608 |
+
2304,
|
| 1609 |
+
768,
|
| 1610 |
+
1,
|
| 1611 |
+
1
|
| 1612 |
+
]
|
| 1613 |
+
},
|
| 1614 |
+
"dec_attns.1.qkv.bias": {
|
| 1615 |
+
"w_norm": 1.0041871070861816,
|
| 1616 |
+
"g_norm": 2.3000570763542783e-06,
|
| 1617 |
+
"shape": [
|
| 1618 |
+
2304
|
| 1619 |
+
]
|
| 1620 |
+
},
|
| 1621 |
+
"dec_attns.1.proj.weight": {
|
| 1622 |
+
"w_norm": 16.047632217407227,
|
| 1623 |
+
"g_norm": 5.869360757060349e-05,
|
| 1624 |
+
"shape": [
|
| 1625 |
+
768,
|
| 1626 |
+
768,
|
| 1627 |
+
1,
|
| 1628 |
+
1
|
| 1629 |
+
]
|
| 1630 |
+
},
|
| 1631 |
+
"dec_attns.1.proj.bias": {
|
| 1632 |
+
"w_norm": 0.579659640789032,
|
| 1633 |
+
"g_norm": 3.644436446847976e-06,
|
| 1634 |
+
"shape": [
|
| 1635 |
+
768
|
| 1636 |
+
]
|
| 1637 |
+
},
|
| 1638 |
+
"dec_attns.2.norm.weight": {
|
| 1639 |
+
"w_norm": 19.534183502197266,
|
| 1640 |
+
"g_norm": 4.331167474447284e-06,
|
| 1641 |
+
"shape": [
|
| 1642 |
+
384
|
| 1643 |
+
]
|
| 1644 |
+
},
|
| 1645 |
+
"dec_attns.2.norm.bias": {
|
| 1646 |
+
"w_norm": 0.019406652078032494,
|
| 1647 |
+
"g_norm": 4.485439148993464e-06,
|
| 1648 |
+
"shape": [
|
| 1649 |
+
384
|
| 1650 |
+
]
|
| 1651 |
+
},
|
| 1652 |
+
"dec_attns.2.qkv.weight": {
|
| 1653 |
+
"w_norm": 19.665908813476562,
|
| 1654 |
+
"g_norm": 0.0001103595714084804,
|
| 1655 |
+
"shape": [
|
| 1656 |
+
1152,
|
| 1657 |
+
384,
|
| 1658 |
+
1,
|
| 1659 |
+
1
|
| 1660 |
+
]
|
| 1661 |
+
},
|
| 1662 |
+
"dec_attns.2.qkv.bias": {
|
| 1663 |
+
"w_norm": 0.9858006238937378,
|
| 1664 |
+
"g_norm": 7.38036851544166e-06,
|
| 1665 |
+
"shape": [
|
| 1666 |
+
1152
|
| 1667 |
+
]
|
| 1668 |
+
},
|
| 1669 |
+
"dec_attns.2.proj.weight": {
|
| 1670 |
+
"w_norm": 11.296539306640625,
|
| 1671 |
+
"g_norm": 0.00010973431199090555,
|
| 1672 |
+
"shape": [
|
| 1673 |
+
384,
|
| 1674 |
+
384,
|
| 1675 |
+
1,
|
| 1676 |
+
1
|
| 1677 |
+
]
|
| 1678 |
+
},
|
| 1679 |
+
"dec_attns.2.proj.bias": {
|
| 1680 |
+
"w_norm": 0.5865283012390137,
|
| 1681 |
+
"g_norm": 1.2120593964937143e-05,
|
| 1682 |
+
"shape": [
|
| 1683 |
+
384
|
| 1684 |
+
]
|
| 1685 |
+
},
|
| 1686 |
+
"dec_ups.0.op.weight": {
|
| 1687 |
+
"w_norm": 16.096851348876953,
|
| 1688 |
+
"g_norm": 0.00019765795150306076,
|
| 1689 |
+
"shape": [
|
| 1690 |
+
768,
|
| 1691 |
+
768,
|
| 1692 |
+
3,
|
| 1693 |
+
3
|
| 1694 |
+
]
|
| 1695 |
+
},
|
| 1696 |
+
"dec_ups.0.op.bias": {
|
| 1697 |
+
"w_norm": 0.2011629343032837,
|
| 1698 |
+
"g_norm": 1.6096408899102244e-06,
|
| 1699 |
+
"shape": [
|
| 1700 |
+
768
|
| 1701 |
+
]
|
| 1702 |
+
},
|
| 1703 |
+
"dec_ups.1.op.weight": {
|
| 1704 |
+
"w_norm": 16.08983039855957,
|
| 1705 |
+
"g_norm": 0.0004939130740240216,
|
| 1706 |
+
"shape": [
|
| 1707 |
+
768,
|
| 1708 |
+
768,
|
| 1709 |
+
3,
|
| 1710 |
+
3
|
| 1711 |
+
]
|
| 1712 |
+
},
|
| 1713 |
+
"dec_ups.1.op.bias": {
|
| 1714 |
+
"w_norm": 0.18965400755405426,
|
| 1715 |
+
"g_norm": 5.676467935700202e-06,
|
| 1716 |
+
"shape": [
|
| 1717 |
+
768
|
| 1718 |
+
]
|
| 1719 |
+
},
|
| 1720 |
+
"dec_ups.2.op.weight": {
|
| 1721 |
+
"w_norm": 11.300466537475586,
|
| 1722 |
+
"g_norm": 0.0012292304309085011,
|
| 1723 |
+
"shape": [
|
| 1724 |
+
384,
|
| 1725 |
+
384,
|
| 1726 |
+
3,
|
| 1727 |
+
3
|
| 1728 |
+
]
|
| 1729 |
+
},
|
| 1730 |
+
"dec_ups.2.op.bias": {
|
| 1731 |
+
"w_norm": 0.1965014785528183,
|
| 1732 |
+
"g_norm": 2.0350802515167743e-05,
|
| 1733 |
+
"shape": [
|
| 1734 |
+
384
|
| 1735 |
+
]
|
| 1736 |
+
},
|
| 1737 |
+
"dec_ups.3.op.weight": {
|
| 1738 |
+
"w_norm": 11.314447402954102,
|
| 1739 |
+
"g_norm": 0.0029695341363549232,
|
| 1740 |
+
"shape": [
|
| 1741 |
+
384,
|
| 1742 |
+
384,
|
| 1743 |
+
3,
|
| 1744 |
+
3
|
| 1745 |
+
]
|
| 1746 |
+
},
|
| 1747 |
+
"dec_ups.3.op.bias": {
|
| 1748 |
+
"w_norm": 0.19801823794841766,
|
| 1749 |
+
"g_norm": 8.36284234537743e-05,
|
| 1750 |
+
"shape": [
|
| 1751 |
+
384
|
| 1752 |
+
]
|
| 1753 |
+
},
|
| 1754 |
+
"dec_ups.4.op.weight": {
|
| 1755 |
+
"w_norm": 7.997620105743408,
|
| 1756 |
+
"g_norm": 0.014928634278476238,
|
| 1757 |
+
"shape": [
|
| 1758 |
+
192,
|
| 1759 |
+
192,
|
| 1760 |
+
3,
|
| 1761 |
+
3
|
| 1762 |
+
]
|
| 1763 |
+
},
|
| 1764 |
+
"dec_ups.4.op.bias": {
|
| 1765 |
+
"w_norm": 0.19552351534366608,
|
| 1766 |
+
"g_norm": 0.00039662071503698826,
|
| 1767 |
+
"shape": [
|
| 1768 |
+
192
|
| 1769 |
+
]
|
| 1770 |
+
},
|
| 1771 |
+
"out_norm.weight": {
|
| 1772 |
+
"w_norm": 13.799749374389648,
|
| 1773 |
+
"g_norm": 0.0063019501976668835,
|
| 1774 |
+
"shape": [
|
| 1775 |
+
192
|
| 1776 |
+
]
|
| 1777 |
+
},
|
| 1778 |
+
"out_norm.bias": {
|
| 1779 |
+
"w_norm": 0.07100910693407059,
|
| 1780 |
+
"g_norm": 0.009717518463730812,
|
| 1781 |
+
"shape": [
|
| 1782 |
+
192
|
| 1783 |
+
]
|
| 1784 |
+
},
|
| 1785 |
+
"out_conv.weight": {
|
| 1786 |
+
"w_norm": 0.12911434471607208,
|
| 1787 |
+
"g_norm": 0.9639920592308044,
|
| 1788 |
+
"shape": [
|
| 1789 |
+
3,
|
| 1790 |
+
192,
|
| 1791 |
+
3,
|
| 1792 |
+
3
|
| 1793 |
+
]
|
| 1794 |
+
},
|
| 1795 |
+
"out_conv.bias": {
|
| 1796 |
+
"w_norm": 0.00815030001103878,
|
| 1797 |
+
"g_norm": 0.06712251156568527,
|
| 1798 |
+
"shape": [
|
| 1799 |
+
3
|
| 1800 |
+
]
|
| 1801 |
+
}
|
| 1802 |
+
},
|
| 1803 |
+
"activations": {
|
| 1804 |
+
"input_conv": {
|
| 1805 |
+
"mean": 0.00026234411052428186,
|
| 1806 |
+
"std": 0.3449311852455139,
|
| 1807 |
+
"abs_max": 2.703125,
|
| 1808 |
+
"frac_dead": 0.000533885438926518,
|
| 1809 |
+
"frac_sat": 0.0
|
| 1810 |
+
},
|
| 1811 |
+
"time_mlp.1": {
|
| 1812 |
+
"mean": 0.021856388077139854,
|
| 1813 |
+
"std": 0.2685975730419159,
|
| 1814 |
+
"abs_max": 3.671875,
|
| 1815 |
+
"frac_dead": 0.0,
|
| 1816 |
+
"frac_sat": 0.0
|
| 1817 |
+
},
|
| 1818 |
+
"time_mlp.3": {
|
| 1819 |
+
"mean": 0.02677418850362301,
|
| 1820 |
+
"std": 0.18945495784282684,
|
| 1821 |
+
"abs_max": 0.78515625,
|
| 1822 |
+
"frac_dead": 0.00032552084303461015,
|
| 1823 |
+
"frac_sat": 0.0
|
| 1824 |
+
},
|
| 1825 |
+
"enc_blocks.0.norm1": {
|
| 1826 |
+
"mean": -0.004714304581284523,
|
| 1827 |
+
"std": 0.9962726831436157,
|
| 1828 |
+
"abs_max": 6.979825019836426,
|
| 1829 |
+
"frac_dead": 7.980399914231384e-07,
|
| 1830 |
+
"frac_sat": 0.0
|
| 1831 |
+
},
|
| 1832 |
+
"enc_blocks.0.conv1": {
|
| 1833 |
+
"mean": -0.027971331030130386,
|
| 1834 |
+
"std": 0.5426817536354065,
|
| 1835 |
+
"abs_max": 9.0625,
|
| 1836 |
+
"frac_dead": 6.779366231057793e-05,
|
| 1837 |
+
"frac_sat": 0.0
|
| 1838 |
+
},
|
| 1839 |
+
"enc_blocks.0.t_proj": {
|
| 1840 |
+
"mean": -0.0008105086162686348,
|
| 1841 |
+
"std": 0.17592351138591766,
|
| 1842 |
+
"abs_max": 1.4296875,
|
| 1843 |
+
"frac_dead": 0.0,
|
| 1844 |
+
"frac_sat": 0.0
|
| 1845 |
+
},
|
| 1846 |
+
"enc_blocks.0.norm2": {
|
| 1847 |
+
"mean": -0.002945144195109606,
|
| 1848 |
+
"std": 0.9993420839309692,
|
| 1849 |
+
"abs_max": 10.731555938720703,
|
| 1850 |
+
"frac_dead": 6.622738624173508e-07,
|
| 1851 |
+
"frac_sat": 2.9802322387695312e-08
|
| 1852 |
+
},
|
| 1853 |
+
"enc_blocks.0.conv2": {
|
| 1854 |
+
"mean": -0.010785315185785294,
|
| 1855 |
+
"std": 0.8624541163444519,
|
| 1856 |
+
"abs_max": 8.125,
|
| 1857 |
+
"frac_dead": 3.263685357524082e-05,
|
| 1858 |
+
"frac_sat": 0.0
|
| 1859 |
+
},
|
| 1860 |
+
"enc_downs.0.op": {
|
| 1861 |
+
"mean": -0.10003193467855453,
|
| 1862 |
+
"std": 1.335770606994629,
|
| 1863 |
+
"abs_max": 12.875,
|
| 1864 |
+
"frac_dead": 1.9497341781971045e-05,
|
| 1865 |
+
"frac_sat": 2.384185791015625e-07
|
| 1866 |
+
},
|
| 1867 |
+
"enc_blocks.1.norm1": {
|
| 1868 |
+
"mean": -0.0012766718864440918,
|
| 1869 |
+
"std": 0.9998140931129456,
|
| 1870 |
+
"abs_max": 9.364228248596191,
|
| 1871 |
+
"frac_dead": 4.106097719613899e-07,
|
| 1872 |
+
"frac_sat": 0.0
|
| 1873 |
+
},
|
| 1874 |
+
"enc_blocks.1.conv1": {
|
| 1875 |
+
"mean": -0.24094407260417938,
|
| 1876 |
+
"std": 1.0454233884811401,
|
| 1877 |
+
"abs_max": 10.25,
|
| 1878 |
+
"frac_dead": 3.066327917622402e-05,
|
| 1879 |
+
"frac_sat": 1.3245476715439963e-08
|
| 1880 |
+
},
|
| 1881 |
+
"enc_blocks.1.t_proj": {
|
| 1882 |
+
"mean": 0.00418504374101758,
|
| 1883 |
+
"std": 0.1214127168059349,
|
| 1884 |
+
"abs_max": 0.86328125,
|
| 1885 |
+
"frac_dead": 0.0,
|
| 1886 |
+
"frac_sat": 0.0
|
| 1887 |
+
},
|
| 1888 |
+
"enc_blocks.1.norm2": {
|
| 1889 |
+
"mean": -0.003717670915648341,
|
| 1890 |
+
"std": 0.9974992871284485,
|
| 1891 |
+
"abs_max": 8.737704277038574,
|
| 1892 |
+
"frac_dead": 3.97364289028701e-07,
|
| 1893 |
+
"frac_sat": 0.0
|
| 1894 |
+
},
|
| 1895 |
+
"enc_blocks.1.conv2": {
|
| 1896 |
+
"mean": 0.028257178142666817,
|
| 1897 |
+
"std": 0.7950882911682129,
|
| 1898 |
+
"abs_max": 9.4375,
|
| 1899 |
+
"frac_dead": 5.786948895547539e-05,
|
| 1900 |
+
"frac_sat": 0.0
|
| 1901 |
+
},
|
| 1902 |
+
"enc_downs.1.op": {
|
| 1903 |
+
"mean": -0.054014790803194046,
|
| 1904 |
+
"std": 1.8062548637390137,
|
| 1905 |
+
"abs_max": 11.375,
|
| 1906 |
+
"frac_dead": 1.9550323486328125e-05,
|
| 1907 |
+
"frac_sat": 7.417467031700653e-07
|
| 1908 |
+
},
|
| 1909 |
+
"enc_blocks.2.norm1": {
|
| 1910 |
+
"mean": -0.001649506390094757,
|
| 1911 |
+
"std": 0.9994705319404602,
|
| 1912 |
+
"abs_max": 7.457204818725586,
|
| 1913 |
+
"frac_dead": 1.0066562481370056e-06,
|
| 1914 |
+
"frac_sat": 0.0
|
| 1915 |
+
},
|
| 1916 |
+
"enc_blocks.2.conv1": {
|
| 1917 |
+
"mean": -0.19917495548725128,
|
| 1918 |
+
"std": 1.0274417400360107,
|
| 1919 |
+
"abs_max": 11.5,
|
| 1920 |
+
"frac_dead": 4.151132452534512e-05,
|
| 1921 |
+
"frac_sat": 1.5099843722055084e-06
|
| 1922 |
+
},
|
| 1923 |
+
"enc_blocks.2.t_proj": {
|
| 1924 |
+
"mean": -0.002049730159342289,
|
| 1925 |
+
"std": 0.17591041326522827,
|
| 1926 |
+
"abs_max": 1.4296875,
|
| 1927 |
+
"frac_dead": 0.0,
|
| 1928 |
+
"frac_sat": 0.0
|
| 1929 |
+
},
|
| 1930 |
+
"enc_blocks.2.norm2": {
|
| 1931 |
+
"mean": -0.006312957499176264,
|
| 1932 |
+
"std": 0.9954407811164856,
|
| 1933 |
+
"abs_max": 13.2487211227417,
|
| 1934 |
+
"frac_dead": 3.178914482759865e-07,
|
| 1935 |
+
"frac_sat": 1.8543667010817444e-06
|
| 1936 |
+
},
|
| 1937 |
+
"enc_blocks.2.conv2": {
|
| 1938 |
+
"mean": 0.06416342407464981,
|
| 1939 |
+
"std": 0.6680892705917358,
|
| 1940 |
+
"abs_max": 6.40625,
|
| 1941 |
+
"frac_dead": 3.208054476999678e-05,
|
| 1942 |
+
"frac_sat": 0.0
|
| 1943 |
+
},
|
| 1944 |
+
"enc_blocks.2.skip": {
|
| 1945 |
+
"mean": -0.02201053313910961,
|
| 1946 |
+
"std": 1.0272172689437866,
|
| 1947 |
+
"abs_max": 8.375,
|
| 1948 |
+
"frac_dead": 0.00010784467303892598,
|
| 1949 |
+
"frac_sat": 0.0
|
| 1950 |
+
},
|
| 1951 |
+
"enc_downs.2.op": {
|
| 1952 |
+
"mean": 0.13211365044116974,
|
| 1953 |
+
"std": 1.7092348337173462,
|
| 1954 |
+
"abs_max": 9.9375,
|
| 1955 |
+
"frac_dead": 1.33514404296875e-05,
|
| 1956 |
+
"frac_sat": 0.0
|
| 1957 |
+
},
|
| 1958 |
+
"enc_blocks.3.norm1": {
|
| 1959 |
+
"mean": 6.256169581320137e-05,
|
| 1960 |
+
"std": 1.0005357265472412,
|
| 1961 |
+
"abs_max": 8.886502265930176,
|
| 1962 |
+
"frac_dead": 8.477105097881577e-07,
|
| 1963 |
+
"frac_sat": 0.0
|
| 1964 |
+
},
|
| 1965 |
+
"enc_blocks.3.conv1": {
|
| 1966 |
+
"mean": -0.1537984013557434,
|
| 1967 |
+
"std": 1.3866938352584839,
|
| 1968 |
+
"abs_max": 20.75,
|
| 1969 |
+
"frac_dead": 2.1404690414783545e-05,
|
| 1970 |
+
"frac_sat": 0.0002346038818359375
|
| 1971 |
+
},
|
| 1972 |
+
"enc_blocks.3.t_proj": {
|
| 1973 |
+
"mean": 0.0013901263009756804,
|
| 1974 |
+
"std": 0.09665051102638245,
|
| 1975 |
+
"abs_max": 0.56640625,
|
| 1976 |
+
"frac_dead": 0.0,
|
| 1977 |
+
"frac_sat": 0.0
|
| 1978 |
+
},
|
| 1979 |
+
"enc_blocks.3.norm2": {
|
| 1980 |
+
"mean": -0.005035887472331524,
|
| 1981 |
+
"std": 0.9948577284812927,
|
| 1982 |
+
"abs_max": 11.17821979522705,
|
| 1983 |
+
"frac_dead": 5.298190899338806e-07,
|
| 1984 |
+
"frac_sat": 2.013312496274011e-06
|
| 1985 |
+
},
|
| 1986 |
+
"enc_blocks.3.conv2": {
|
| 1987 |
+
"mean": -0.0170371625572443,
|
| 1988 |
+
"std": 0.7531740069389343,
|
| 1989 |
+
"abs_max": 10.875,
|
| 1990 |
+
"frac_dead": 3.4332275390625e-05,
|
| 1991 |
+
"frac_sat": 1.059638137235197e-07
|
| 1992 |
+
},
|
| 1993 |
+
"enc_attns.3.norm": {
|
| 1994 |
+
"mean": 0.00035631077480502427,
|
| 1995 |
+
"std": 0.9987042546272278,
|
| 1996 |
+
"abs_max": 11.254258155822754,
|
| 1997 |
+
"frac_dead": 1.6954210195763153e-06,
|
| 1998 |
+
"frac_sat": 3.178914482759865e-07
|
| 1999 |
+
},
|
| 2000 |
+
"enc_attns.3.qkv": {
|
| 2001 |
+
"mean": -0.0017836717888712883,
|
| 2002 |
+
"std": 0.6752532720565796,
|
| 2003 |
+
"abs_max": 10.125,
|
| 2004 |
+
"frac_dead": 9.808717004489154e-05,
|
| 2005 |
+
"frac_sat": 3.5321271241173235e-08
|
| 2006 |
+
},
|
| 2007 |
+
"enc_attns.3.proj": {
|
| 2008 |
+
"mean": -0.01955404318869114,
|
| 2009 |
+
"std": 0.49900659918785095,
|
| 2010 |
+
"abs_max": 5.0625,
|
| 2011 |
+
"frac_dead": 0.00014368693518918008,
|
| 2012 |
+
"frac_sat": 0.0
|
| 2013 |
+
},
|
| 2014 |
+
"enc_downs.3.op": {
|
| 2015 |
+
"mean": -0.12393307685852051,
|
| 2016 |
+
"std": 2.2092812061309814,
|
| 2017 |
+
"abs_max": 13.125,
|
| 2018 |
+
"frac_dead": 1.6530355424038135e-05,
|
| 2019 |
+
"frac_sat": 0.00044462416553869843
|
| 2020 |
+
},
|
| 2021 |
+
"enc_blocks.4.norm1": {
|
| 2022 |
+
"mean": -0.00021128480148036033,
|
| 2023 |
+
"std": 1.0009337663650513,
|
| 2024 |
+
"abs_max": 10.801759719848633,
|
| 2025 |
+
"frac_dead": 2.1192763597355224e-06,
|
| 2026 |
+
"frac_sat": 1.6954210195763153e-06
|
| 2027 |
+
},
|
| 2028 |
+
"enc_blocks.4.conv1": {
|
| 2029 |
+
"mean": -0.17340169847011566,
|
| 2030 |
+
"std": 1.5200656652450562,
|
| 2031 |
+
"abs_max": 26.625,
|
| 2032 |
+
"frac_dead": 2.288818359375e-05,
|
| 2033 |
+
"frac_sat": 0.000911712646484375
|
| 2034 |
+
},
|
| 2035 |
+
"enc_blocks.4.t_proj": {
|
| 2036 |
+
"mean": -0.0030492052901536226,
|
| 2037 |
+
"std": 0.08274214714765549,
|
| 2038 |
+
"abs_max": 0.5625,
|
| 2039 |
+
"frac_dead": 0.0,
|
| 2040 |
+
"frac_sat": 0.0
|
| 2041 |
+
},
|
| 2042 |
+
"enc_blocks.4.norm2": {
|
| 2043 |
+
"mean": -0.005301392171531916,
|
| 2044 |
+
"std": 0.9954476952552795,
|
| 2045 |
+
"abs_max": 16.856233596801758,
|
| 2046 |
+
"frac_dead": 6.35782896551973e-07,
|
| 2047 |
+
"frac_sat": 5.91278076171875e-05
|
| 2048 |
+
},
|
| 2049 |
+
"enc_blocks.4.conv2": {
|
| 2050 |
+
"mean": 0.03106667287647724,
|
| 2051 |
+
"std": 0.6966545581817627,
|
| 2052 |
+
"abs_max": 5.6875,
|
| 2053 |
+
"frac_dead": 2.47955322265625e-05,
|
| 2054 |
+
"frac_sat": 0.0
|
| 2055 |
+
},
|
| 2056 |
+
"enc_blocks.4.skip": {
|
| 2057 |
+
"mean": -0.03207727149128914,
|
| 2058 |
+
"std": 1.269452691078186,
|
| 2059 |
+
"abs_max": 9.3125,
|
| 2060 |
+
"frac_dead": 5.955166489002295e-05,
|
| 2061 |
+
"frac_sat": 0.0
|
| 2062 |
+
},
|
| 2063 |
+
"enc_attns.4.norm": {
|
| 2064 |
+
"mean": 2.906637018895708e-05,
|
| 2065 |
+
"std": 0.9980901479721069,
|
| 2066 |
+
"abs_max": 10.796359062194824,
|
| 2067 |
+
"frac_dead": 1.0596381798677612e-06,
|
| 2068 |
+
"frac_sat": 2.119276274470394e-07
|
| 2069 |
+
},
|
| 2070 |
+
"enc_attns.4.qkv": {
|
| 2071 |
+
"mean": -0.006540832109749317,
|
| 2072 |
+
"std": 0.6688223481178284,
|
| 2073 |
+
"abs_max": 7.96875,
|
| 2074 |
+
"frac_dead": 7.191410986706614e-05,
|
| 2075 |
+
"frac_sat": 0.0
|
| 2076 |
+
},
|
| 2077 |
+
"enc_attns.4.proj": {
|
| 2078 |
+
"mean": -0.0005021503893658519,
|
| 2079 |
+
"std": 0.3761708736419678,
|
| 2080 |
+
"abs_max": 3.6875,
|
| 2081 |
+
"frac_dead": 0.00011041429388569668,
|
| 2082 |
+
"frac_sat": 0.0
|
| 2083 |
+
},
|
| 2084 |
+
"enc_downs.4.op": {
|
| 2085 |
+
"mean": 0.05056021735072136,
|
| 2086 |
+
"std": 2.155576705932617,
|
| 2087 |
+
"abs_max": 14.1875,
|
| 2088 |
+
"frac_dead": 8.47710543894209e-06,
|
| 2089 |
+
"frac_sat": 0.0006747775478288531
|
| 2090 |
+
},
|
| 2091 |
+
"enc_blocks.5.norm1": {
|
| 2092 |
+
"mean": -5.367605263018049e-05,
|
| 2093 |
+
"std": 1.00002121925354,
|
| 2094 |
+
"abs_max": 8.307374000549316,
|
| 2095 |
+
"frac_dead": 0.0,
|
| 2096 |
+
"frac_sat": 0.0
|
| 2097 |
+
},
|
| 2098 |
+
"enc_blocks.5.conv1": {
|
| 2099 |
+
"mean": -0.35223713517189026,
|
| 2100 |
+
"std": 1.462549090385437,
|
| 2101 |
+
"abs_max": 19.5,
|
| 2102 |
+
"frac_dead": 2.1192763597355224e-05,
|
| 2103 |
+
"frac_sat": 0.0005204942426644266
|
| 2104 |
+
},
|
| 2105 |
+
"enc_blocks.5.t_proj": {
|
| 2106 |
+
"mean": -0.0026074519846588373,
|
| 2107 |
+
"std": 0.07342220097780228,
|
| 2108 |
+
"abs_max": 0.46484375,
|
| 2109 |
+
"frac_dead": 0.0,
|
| 2110 |
+
"frac_sat": 0.0
|
| 2111 |
+
},
|
| 2112 |
+
"enc_blocks.5.norm2": {
|
| 2113 |
+
"mean": -0.004012688063085079,
|
| 2114 |
+
"std": 0.9965819716453552,
|
| 2115 |
+
"abs_max": 10.07455825805664,
|
| 2116 |
+
"frac_dead": 8.477105097881577e-07,
|
| 2117 |
+
"frac_sat": 8.477105097881577e-07
|
| 2118 |
+
},
|
| 2119 |
+
"enc_blocks.5.conv2": {
|
| 2120 |
+
"mean": 0.08693259209394455,
|
| 2121 |
+
"std": 1.073614478111267,
|
| 2122 |
+
"abs_max": 6.3125,
|
| 2123 |
+
"frac_dead": 2.0345052689663135e-05,
|
| 2124 |
+
"frac_sat": 0.0
|
| 2125 |
+
},
|
| 2126 |
+
"enc_attns.5.norm": {
|
| 2127 |
+
"mean": 9.941578900907189e-05,
|
| 2128 |
+
"std": 0.998691201210022,
|
| 2129 |
+
"abs_max": 7.955250263214111,
|
| 2130 |
+
"frac_dead": 8.477105097881577e-07,
|
| 2131 |
+
"frac_sat": 0.0
|
| 2132 |
+
},
|
| 2133 |
+
"enc_attns.5.qkv": {
|
| 2134 |
+
"mean": -0.008100636303424835,
|
| 2135 |
+
"std": 0.7708551287651062,
|
| 2136 |
+
"abs_max": 5.625,
|
| 2137 |
+
"frac_dead": 7.572880713269114e-05,
|
| 2138 |
+
"frac_sat": 0.0
|
| 2139 |
+
},
|
| 2140 |
+
"enc_attns.5.proj": {
|
| 2141 |
+
"mean": -0.008143105544149876,
|
| 2142 |
+
"std": 0.2769840359687805,
|
| 2143 |
+
"abs_max": 3.234375,
|
| 2144 |
+
"frac_dead": 0.00019073486328125,
|
| 2145 |
+
"frac_sat": 0.0
|
| 2146 |
+
},
|
| 2147 |
+
"mid1.norm1": {
|
| 2148 |
+
"mean": 0.00015265148249454796,
|
| 2149 |
+
"std": 1.000354528427124,
|
| 2150 |
+
"abs_max": 7.3556437492370605,
|
| 2151 |
+
"frac_dead": 8.477105097881577e-07,
|
| 2152 |
+
"frac_sat": 0.0
|
| 2153 |
+
},
|
| 2154 |
+
"mid1.conv1": {
|
| 2155 |
+
"mean": -0.569250762462616,
|
| 2156 |
+
"std": 1.7395997047424316,
|
| 2157 |
+
"abs_max": 22.875,
|
| 2158 |
+
"frac_dead": 2.5431314497836865e-05,
|
| 2159 |
+
"frac_sat": 0.0027398003730922937
|
| 2160 |
+
},
|
| 2161 |
+
"mid1.t_proj": {
|
| 2162 |
+
"mean": 0.0030422655399888754,
|
| 2163 |
+
"std": 0.07387731224298477,
|
| 2164 |
+
"abs_max": 0.361328125,
|
| 2165 |
+
"frac_dead": 0.0,
|
| 2166 |
+
"frac_sat": 0.0
|
| 2167 |
+
},
|
| 2168 |
+
"mid1.norm2": {
|
| 2169 |
+
"mean": -0.0043882750906050205,
|
| 2170 |
+
"std": 0.9972242712974548,
|
| 2171 |
+
"abs_max": 10.796936988830566,
|
| 2172 |
+
"frac_dead": 2.543131586207892e-06,
|
| 2173 |
+
"frac_sat": 3.3908420391526306e-06
|
| 2174 |
+
},
|
| 2175 |
+
"mid1.conv2": {
|
| 2176 |
+
"mean": 0.01897788606584072,
|
| 2177 |
+
"std": 0.6111142635345459,
|
| 2178 |
+
"abs_max": 4.125,
|
| 2179 |
+
"frac_dead": 2.458360540913418e-05,
|
| 2180 |
+
"frac_sat": 0.0
|
| 2181 |
+
},
|
| 2182 |
+
"mid_attn.norm": {
|
| 2183 |
+
"mean": 4.941597580909729e-05,
|
| 2184 |
+
"std": 0.9988196492195129,
|
| 2185 |
+
"abs_max": 7.046948432922363,
|
| 2186 |
+
"frac_dead": 2.543131586207892e-06,
|
| 2187 |
+
"frac_sat": 0.0
|
| 2188 |
+
},
|
| 2189 |
+
"mid_attn.qkv": {
|
| 2190 |
+
"mean": 0.005729823373258114,
|
| 2191 |
+
"std": 0.7872936725616455,
|
| 2192 |
+
"abs_max": 6.1875,
|
| 2193 |
+
"frac_dead": 6.640399078605697e-05,
|
| 2194 |
+
"frac_sat": 0.0
|
| 2195 |
+
},
|
| 2196 |
+
"mid_attn.proj": {
|
| 2197 |
+
"mean": -0.005338292568922043,
|
| 2198 |
+
"std": 0.24077516794204712,
|
| 2199 |
+
"abs_max": 1.765625,
|
| 2200 |
+
"frac_dead": 0.00017547607421875,
|
| 2201 |
+
"frac_sat": 0.0
|
| 2202 |
+
},
|
| 2203 |
+
"mid2.norm1": {
|
| 2204 |
+
"mean": 0.00037162296939641237,
|
| 2205 |
+
"std": 1.0006673336029053,
|
| 2206 |
+
"abs_max": 7.135955333709717,
|
| 2207 |
+
"frac_dead": 8.477105097881577e-07,
|
| 2208 |
+
"frac_sat": 0.0
|
| 2209 |
+
},
|
| 2210 |
+
"mid2.conv1": {
|
| 2211 |
+
"mean": -0.6954900622367859,
|
| 2212 |
+
"std": 1.9247785806655884,
|
| 2213 |
+
"abs_max": 24.75,
|
| 2214 |
+
"frac_dead": 2.1192763597355224e-05,
|
| 2215 |
+
"frac_sat": 0.005978054367005825
|
| 2216 |
+
},
|
| 2217 |
+
"mid2.t_proj": {
|
| 2218 |
+
"mean": 0.0007212297641672194,
|
| 2219 |
+
"std": 0.07543076574802399,
|
| 2220 |
+
"abs_max": 0.4609375,
|
| 2221 |
+
"frac_dead": 0.0,
|
| 2222 |
+
"frac_sat": 0.0
|
| 2223 |
+
},
|
| 2224 |
+
"mid2.norm2": {
|
| 2225 |
+
"mean": -0.005086460150778294,
|
| 2226 |
+
"std": 0.9972400665283203,
|
| 2227 |
+
"abs_max": 9.937588691711426,
|
| 2228 |
+
"frac_dead": 8.477105097881577e-07,
|
| 2229 |
+
"frac_sat": 0.0
|
| 2230 |
+
},
|
| 2231 |
+
"mid2.conv2": {
|
| 2232 |
+
"mean": 0.004386311862617731,
|
| 2233 |
+
"std": 0.6135098338127136,
|
| 2234 |
+
"abs_max": 4.0625,
|
| 2235 |
+
"frac_dead": 2.7126736313221045e-05,
|
| 2236 |
+
"frac_sat": 0.0
|
| 2237 |
+
},
|
| 2238 |
+
"dec_blocks.0.norm1": {
|
| 2239 |
+
"mean": 0.00032785534858703613,
|
| 2240 |
+
"std": 1.0008313655853271,
|
| 2241 |
+
"abs_max": 8.297355651855469,
|
| 2242 |
+
"frac_dead": 0.0,
|
| 2243 |
+
"frac_sat": 0.0
|
| 2244 |
+
},
|
| 2245 |
+
"dec_blocks.0.conv1": {
|
| 2246 |
+
"mean": -2.0214412212371826,
|
| 2247 |
+
"std": 3.9078080654144287,
|
| 2248 |
+
"abs_max": 50.0,
|
| 2249 |
+
"frac_dead": 3.3908420391526306e-06,
|
| 2250 |
+
"frac_sat": 0.05213843658566475
|
| 2251 |
+
},
|
| 2252 |
+
"dec_blocks.0.t_proj": {
|
| 2253 |
+
"mean": -0.001085665193386376,
|
| 2254 |
+
"std": 0.077158622443676,
|
| 2255 |
+
"abs_max": 0.361328125,
|
| 2256 |
+
"frac_dead": 0.0,
|
| 2257 |
+
"frac_sat": 0.0
|
| 2258 |
+
},
|
| 2259 |
+
"dec_blocks.0.norm2": {
|
| 2260 |
+
"mean": -0.006708974484354258,
|
| 2261 |
+
"std": 0.9966805577278137,
|
| 2262 |
+
"abs_max": 10.097886085510254,
|
| 2263 |
+
"frac_dead": 1.6954210195763153e-06,
|
| 2264 |
+
"frac_sat": 1.6954210195763153e-06
|
| 2265 |
+
},
|
| 2266 |
+
"dec_blocks.0.conv2": {
|
| 2267 |
+
"mean": 0.011965113691985607,
|
| 2268 |
+
"std": 0.45058712363243103,
|
| 2269 |
+
"abs_max": 2.140625,
|
| 2270 |
+
"frac_dead": 2.7126736313221045e-05,
|
| 2271 |
+
"frac_sat": 0.0
|
| 2272 |
+
},
|
| 2273 |
+
"dec_blocks.0.skip": {
|
| 2274 |
+
"mean": -0.00819339882582426,
|
| 2275 |
+
"std": 1.3349742889404297,
|
| 2276 |
+
"abs_max": 9.375,
|
| 2277 |
+
"frac_dead": 2.8822158128605224e-05,
|
| 2278 |
+
"frac_sat": 0.0
|
| 2279 |
+
},
|
| 2280 |
+
"dec_attns.0.norm": {
|
| 2281 |
+
"mean": -7.761849701637402e-06,
|
| 2282 |
+
"std": 0.9980735778808594,
|
| 2283 |
+
"abs_max": 6.372596740722656,
|
| 2284 |
+
"frac_dead": 0.0,
|
| 2285 |
+
"frac_sat": 0.0
|
| 2286 |
+
},
|
| 2287 |
+
"dec_attns.0.qkv": {
|
| 2288 |
+
"mean": -0.0017916549695655704,
|
| 2289 |
+
"std": 0.6728198528289795,
|
| 2290 |
+
"abs_max": 5.21875,
|
| 2291 |
+
"frac_dead": 6.329571624519303e-05,
|
| 2292 |
+
"frac_sat": 0.0
|
| 2293 |
+
},
|
| 2294 |
+
"dec_attns.0.proj": {
|
| 2295 |
+
"mean": 0.007564673200249672,
|
| 2296 |
+
"std": 0.4087885618209839,
|
| 2297 |
+
"abs_max": 3.0,
|
| 2298 |
+
"frac_dead": 0.00010850694525288418,
|
| 2299 |
+
"frac_sat": 0.0
|
| 2300 |
+
},
|
| 2301 |
+
"dec_ups.0.op": {
|
| 2302 |
+
"mean": -0.17015184462070465,
|
| 2303 |
+
"std": 1.6812559366226196,
|
| 2304 |
+
"abs_max": 11.4375,
|
| 2305 |
+
"frac_dead": 9.960598617908545e-06,
|
| 2306 |
+
"frac_sat": 3.496805948088877e-05
|
| 2307 |
+
},
|
| 2308 |
+
"dec_blocks.1.norm1": {
|
| 2309 |
+
"mean": -0.00024216788006015122,
|
| 2310 |
+
"std": 1.0003820657730103,
|
| 2311 |
+
"abs_max": 9.206901550292969,
|
| 2312 |
+
"frac_dead": 3.178914482759865e-07,
|
| 2313 |
+
"frac_sat": 0.0
|
| 2314 |
+
},
|
| 2315 |
+
"dec_blocks.1.conv1": {
|
| 2316 |
+
"mean": -1.724596381187439,
|
| 2317 |
+
"std": 3.3776965141296387,
|
| 2318 |
+
"abs_max": 57.5,
|
| 2319 |
+
"frac_dead": 4.662407718569739e-06,
|
| 2320 |
+
"frac_sat": 0.03395335003733635
|
| 2321 |
+
},
|
| 2322 |
+
"dec_blocks.1.t_proj": {
|
| 2323 |
+
"mean": -0.0024088711943477392,
|
| 2324 |
+
"std": 0.07598824799060822,
|
| 2325 |
+
"abs_max": 0.49609375,
|
| 2326 |
+
"frac_dead": 0.0,
|
| 2327 |
+
"frac_sat": 0.0
|
| 2328 |
+
},
|
| 2329 |
+
"dec_blocks.1.norm2": {
|
| 2330 |
+
"mean": -0.00754527747631073,
|
| 2331 |
+
"std": 0.9961175918579102,
|
| 2332 |
+
"abs_max": 14.618008613586426,
|
| 2333 |
+
"frac_dead": 4.238552719471045e-06,
|
| 2334 |
+
"frac_sat": 5.997551852487959e-05
|
| 2335 |
+
},
|
| 2336 |
+
"dec_blocks.1.conv2": {
|
| 2337 |
+
"mean": -9.273696377931628e-06,
|
| 2338 |
+
"std": 0.6917805075645447,
|
| 2339 |
+
"abs_max": 4.25,
|
| 2340 |
+
"frac_dead": 2.47955322265625e-05,
|
| 2341 |
+
"frac_sat": 0.0
|
| 2342 |
+
},
|
| 2343 |
+
"dec_blocks.1.skip": {
|
| 2344 |
+
"mean": -0.006895574275404215,
|
| 2345 |
+
"std": 0.9343755841255188,
|
| 2346 |
+
"abs_max": 6.59375,
|
| 2347 |
+
"frac_dead": 3.687540811370127e-05,
|
| 2348 |
+
"frac_sat": 0.0
|
| 2349 |
+
},
|
| 2350 |
+
"dec_attns.1.norm": {
|
| 2351 |
+
"mean": 1.738758510327898e-05,
|
| 2352 |
+
"std": 0.9983002543449402,
|
| 2353 |
+
"abs_max": 6.872876167297363,
|
| 2354 |
+
"frac_dead": 2.119276274470394e-07,
|
| 2355 |
+
"frac_sat": 0.0
|
| 2356 |
+
},
|
| 2357 |
+
"dec_attns.1.qkv": {
|
| 2358 |
+
"mean": -0.0012654407182708383,
|
| 2359 |
+
"std": 0.6738247871398926,
|
| 2360 |
+
"abs_max": 5.1875,
|
| 2361 |
+
"frac_dead": 6.824069714639336e-05,
|
| 2362 |
+
"frac_sat": 0.0
|
| 2363 |
+
},
|
| 2364 |
+
"dec_attns.1.proj": {
|
| 2365 |
+
"mean": -0.007716475520282984,
|
| 2366 |
+
"std": 0.3779028654098511,
|
| 2367 |
+
"abs_max": 2.96875,
|
| 2368 |
+
"frac_dead": 0.00011444091796875,
|
| 2369 |
+
"frac_sat": 0.0
|
| 2370 |
+
},
|
| 2371 |
+
"dec_ups.1.op": {
|
| 2372 |
+
"mean": -0.013204764574766159,
|
| 2373 |
+
"std": 1.2440454959869385,
|
| 2374 |
+
"abs_max": 8.4375,
|
| 2375 |
+
"frac_dead": 1.2980567589693237e-05,
|
| 2376 |
+
"frac_sat": 0.0
|
| 2377 |
+
},
|
| 2378 |
+
"dec_blocks.2.norm1": {
|
| 2379 |
+
"mean": -0.00021323782857507467,
|
| 2380 |
+
"std": 1.000333309173584,
|
| 2381 |
+
"abs_max": 8.981050491333008,
|
| 2382 |
+
"frac_dead": 8.830317597130488e-07,
|
| 2383 |
+
"frac_sat": 0.0
|
| 2384 |
+
},
|
| 2385 |
+
"dec_blocks.2.conv1": {
|
| 2386 |
+
"mean": -0.8999322056770325,
|
| 2387 |
+
"std": 2.557943105697632,
|
| 2388 |
+
"abs_max": 42.0,
|
| 2389 |
+
"frac_dead": 1.3457404747896362e-05,
|
| 2390 |
+
"frac_sat": 0.018307263031601906
|
| 2391 |
+
},
|
| 2392 |
+
"dec_blocks.2.t_proj": {
|
| 2393 |
+
"mean": -0.00454047741368413,
|
| 2394 |
+
"std": 0.08159489184617996,
|
| 2395 |
+
"abs_max": 0.3828125,
|
| 2396 |
+
"frac_dead": 0.0,
|
| 2397 |
+
"frac_sat": 0.0
|
| 2398 |
+
},
|
| 2399 |
+
"dec_blocks.2.norm2": {
|
| 2400 |
+
"mean": -0.008707795292139053,
|
| 2401 |
+
"std": 0.9994327425956726,
|
| 2402 |
+
"abs_max": 13.457539558410645,
|
| 2403 |
+
"frac_dead": 8.477105097881577e-07,
|
| 2404 |
+
"frac_sat": 3.030565130757168e-05
|
| 2405 |
+
},
|
| 2406 |
+
"dec_blocks.2.conv2": {
|
| 2407 |
+
"mean": 0.012484816834330559,
|
| 2408 |
+
"std": 0.3138270080089569,
|
| 2409 |
+
"abs_max": 1.7265625,
|
| 2410 |
+
"frac_dead": 6.633334851358086e-05,
|
| 2411 |
+
"frac_sat": 0.0
|
| 2412 |
+
},
|
| 2413 |
+
"dec_blocks.2.skip": {
|
| 2414 |
+
"mean": 0.02413799613714218,
|
| 2415 |
+
"std": 1.0058594942092896,
|
| 2416 |
+
"abs_max": 6.75,
|
| 2417 |
+
"frac_dead": 3.751118856598623e-05,
|
| 2418 |
+
"frac_sat": 0.0
|
| 2419 |
+
},
|
| 2420 |
+
"dec_attns.2.norm": {
|
| 2421 |
+
"mean": -6.64558683638461e-05,
|
| 2422 |
+
"std": 0.997047483921051,
|
| 2423 |
+
"abs_max": 6.378185272216797,
|
| 2424 |
+
"frac_dead": 8.477105097881577e-07,
|
| 2425 |
+
"frac_sat": 0.0
|
| 2426 |
+
},
|
| 2427 |
+
"dec_attns.2.qkv": {
|
| 2428 |
+
"mean": 0.004041221924126148,
|
| 2429 |
+
"std": 0.6196464896202087,
|
| 2430 |
+
"abs_max": 5.3125,
|
| 2431 |
+
"frac_dead": 9.222383960150182e-05,
|
| 2432 |
+
"frac_sat": 0.0
|
| 2433 |
+
},
|
| 2434 |
+
"dec_attns.2.proj": {
|
| 2435 |
+
"mean": -0.003088516416028142,
|
| 2436 |
+
"std": 0.3288596570491791,
|
| 2437 |
+
"abs_max": 2.625,
|
| 2438 |
+
"frac_dead": 0.00018077426648233086,
|
| 2439 |
+
"frac_sat": 0.0
|
| 2440 |
+
},
|
| 2441 |
+
"dec_ups.2.op": {
|
| 2442 |
+
"mean": 0.06600978970527649,
|
| 2443 |
+
"std": 0.8171164393424988,
|
| 2444 |
+
"abs_max": 5.96875,
|
| 2445 |
+
"frac_dead": 2.7126736313221045e-05,
|
| 2446 |
+
"frac_sat": 0.0
|
| 2447 |
+
},
|
| 2448 |
+
"dec_blocks.3.norm1": {
|
| 2449 |
+
"mean": -3.3551619708305225e-05,
|
| 2450 |
+
"std": 1.0003398656845093,
|
| 2451 |
+
"abs_max": 7.884114742279053,
|
| 2452 |
+
"frac_dead": 9.271833505408722e-07,
|
| 2453 |
+
"frac_sat": 0.0
|
| 2454 |
+
},
|
| 2455 |
+
"dec_blocks.3.conv1": {
|
| 2456 |
+
"mean": -0.808504045009613,
|
| 2457 |
+
"std": 1.9701484441757202,
|
| 2458 |
+
"abs_max": 30.25,
|
| 2459 |
+
"frac_dead": 1.796086689864751e-05,
|
| 2460 |
+
"frac_sat": 0.003796603996306658
|
| 2461 |
+
},
|
| 2462 |
+
"dec_blocks.3.t_proj": {
|
| 2463 |
+
"mean": -0.000398904929170385,
|
| 2464 |
+
"std": 0.08900758624076843,
|
| 2465 |
+
"abs_max": 0.48046875,
|
| 2466 |
+
"frac_dead": 0.0,
|
| 2467 |
+
"frac_sat": 0.0
|
| 2468 |
+
},
|
| 2469 |
+
"dec_blocks.3.norm2": {
|
| 2470 |
+
"mean": -0.008217168040573597,
|
| 2471 |
+
"std": 0.9956994652748108,
|
| 2472 |
+
"abs_max": 11.516836166381836,
|
| 2473 |
+
"frac_dead": 4.238552548940788e-07,
|
| 2474 |
+
"frac_sat": 4.212061412545154e-06
|
| 2475 |
+
},
|
| 2476 |
+
"dec_blocks.3.conv2": {
|
| 2477 |
+
"mean": 0.010301654227077961,
|
| 2478 |
+
"std": 0.3566579222679138,
|
| 2479 |
+
"abs_max": 3.125,
|
| 2480 |
+
"frac_dead": 5.910131585551426e-05,
|
| 2481 |
+
"frac_sat": 0.0
|
| 2482 |
+
},
|
| 2483 |
+
"dec_blocks.3.skip": {
|
| 2484 |
+
"mean": -0.017353812232613564,
|
| 2485 |
+
"std": 0.6160323619842529,
|
| 2486 |
+
"abs_max": 4.15625,
|
| 2487 |
+
"frac_dead": 7.367134094238281e-05,
|
| 2488 |
+
"frac_sat": 0.0
|
| 2489 |
+
},
|
| 2490 |
+
"dec_ups.3.op": {
|
| 2491 |
+
"mean": -0.02672424353659153,
|
| 2492 |
+
"std": 0.6458936929702759,
|
| 2493 |
+
"abs_max": 7.09375,
|
| 2494 |
+
"frac_dead": 3.521309918141924e-05,
|
| 2495 |
+
"frac_sat": 0.0
|
| 2496 |
+
},
|
| 2497 |
+
"dec_blocks.4.norm1": {
|
| 2498 |
+
"mean": -0.00045055372174829245,
|
| 2499 |
+
"std": 0.9998191595077515,
|
| 2500 |
+
"abs_max": 10.019762992858887,
|
| 2501 |
+
"frac_dead": 7.108405952749308e-07,
|
| 2502 |
+
"frac_sat": 4.4151589051466544e-09
|
| 2503 |
+
},
|
| 2504 |
+
"dec_blocks.4.conv1": {
|
| 2505 |
+
"mean": -0.7278802990913391,
|
| 2506 |
+
"std": 1.838720440864563,
|
| 2507 |
+
"abs_max": 19.625,
|
| 2508 |
+
"frac_dead": 1.2795130714948755e-05,
|
| 2509 |
+
"frac_sat": 0.000757283647544682
|
| 2510 |
+
},
|
| 2511 |
+
"dec_blocks.4.t_proj": {
|
| 2512 |
+
"mean": -0.003106130985543132,
|
| 2513 |
+
"std": 0.09513773769140244,
|
| 2514 |
+
"abs_max": 0.453125,
|
| 2515 |
+
"frac_dead": 0.0,
|
| 2516 |
+
"frac_sat": 0.0
|
| 2517 |
+
},
|
| 2518 |
+
"dec_blocks.4.norm2": {
|
| 2519 |
+
"mean": -0.0046213483437895775,
|
| 2520 |
+
"std": 0.996771514415741,
|
| 2521 |
+
"abs_max": 10.407454490661621,
|
| 2522 |
+
"frac_dead": 4.635916752704361e-07,
|
| 2523 |
+
"frac_sat": 1.059638137235197e-07
|
| 2524 |
+
},
|
| 2525 |
+
"dec_blocks.4.conv2": {
|
| 2526 |
+
"mean": -0.05116056278347969,
|
| 2527 |
+
"std": 0.52388596534729,
|
| 2528 |
+
"abs_max": 6.03125,
|
| 2529 |
+
"frac_dead": 7.355213165283203e-05,
|
| 2530 |
+
"frac_sat": 0.0
|
| 2531 |
+
},
|
| 2532 |
+
"dec_blocks.4.skip": {
|
| 2533 |
+
"mean": -0.051753975450992584,
|
| 2534 |
+
"std": 0.7470393776893616,
|
| 2535 |
+
"abs_max": 6.25,
|
| 2536 |
+
"frac_dead": 7.97775064711459e-05,
|
| 2537 |
+
"frac_sat": 0.0
|
| 2538 |
+
},
|
| 2539 |
+
"dec_ups.4.op": {
|
| 2540 |
+
"mean": -0.22740226984024048,
|
| 2541 |
+
"std": 1.3430253267288208,
|
| 2542 |
+
"abs_max": 18.5,
|
| 2543 |
+
"frac_dead": 3.66667918569874e-05,
|
| 2544 |
+
"frac_sat": 0.00016431344556622207
|
| 2545 |
+
},
|
| 2546 |
+
"dec_blocks.5.norm1": {
|
| 2547 |
+
"mean": -0.0034408338833600283,
|
| 2548 |
+
"std": 0.9988682866096497,
|
| 2549 |
+
"abs_max": 13.417811393737793,
|
| 2550 |
+
"frac_dead": 1.0381141919424408e-06,
|
| 2551 |
+
"frac_sat": 1.564621925354004e-06
|
| 2552 |
+
},
|
| 2553 |
+
"dec_blocks.5.conv1": {
|
| 2554 |
+
"mean": -0.1277112513780594,
|
| 2555 |
+
"std": 1.29868483543396,
|
| 2556 |
+
"abs_max": 10.4375,
|
| 2557 |
+
"frac_dead": 1.688467091298662e-05,
|
| 2558 |
+
"frac_sat": 1.655684656043377e-08
|
| 2559 |
+
},
|
| 2560 |
+
"dec_blocks.5.t_proj": {
|
| 2561 |
+
"mean": 0.009061946533620358,
|
| 2562 |
+
"std": 0.16928990185260773,
|
| 2563 |
+
"abs_max": 1.0859375,
|
| 2564 |
+
"frac_dead": 0.0,
|
| 2565 |
+
"frac_sat": 0.0
|
| 2566 |
+
},
|
| 2567 |
+
"dec_blocks.5.norm2": {
|
| 2568 |
+
"mean": -0.00034307813621126115,
|
| 2569 |
+
"std": 1.0034955739974976,
|
| 2570 |
+
"abs_max": 9.536330223083496,
|
| 2571 |
+
"frac_dead": 5.46375929388887e-07,
|
| 2572 |
+
"frac_sat": 0.0
|
| 2573 |
+
},
|
| 2574 |
+
"dec_blocks.5.conv2": {
|
| 2575 |
+
"mean": -0.30506497621536255,
|
| 2576 |
+
"std": 1.493102788925171,
|
| 2577 |
+
"abs_max": 17.875,
|
| 2578 |
+
"frac_dead": 2.8067164748790674e-05,
|
| 2579 |
+
"frac_sat": 0.00032416978501714766
|
| 2580 |
+
},
|
| 2581 |
+
"dec_blocks.5.skip": {
|
| 2582 |
+
"mean": -0.15524762868881226,
|
| 2583 |
+
"std": 1.0975347757339478,
|
| 2584 |
+
"abs_max": 10.1875,
|
| 2585 |
+
"frac_dead": 6.643599772360176e-05,
|
| 2586 |
+
"frac_sat": 2.317958447406454e-08
|
| 2587 |
+
},
|
| 2588 |
+
"out_norm": {
|
| 2589 |
+
"mean": -0.00539039820432663,
|
| 2590 |
+
"std": 0.9967310428619385,
|
| 2591 |
+
"abs_max": 10.310091972351074,
|
| 2592 |
+
"frac_dead": 3.675619666410057e-07,
|
| 2593 |
+
"frac_sat": 1.3245476715439963e-08
|
| 2594 |
+
},
|
| 2595 |
+
"out_conv": {
|
| 2596 |
+
"mean": 0.1330542266368866,
|
| 2597 |
+
"std": 0.20147153735160828,
|
| 2598 |
+
"abs_max": 0.96484375,
|
| 2599 |
+
"frac_dead": 2.3312039047596045e-05,
|
| 2600 |
+
"frac_sat": 0.0
|
| 2601 |
+
}
|
| 2602 |
+
}
|
| 2603 |
+
}
|
diagnostics/per_layer_iter2000.json
ADDED
|
@@ -0,0 +1,2603 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"iter": 2000,
|
| 3 |
+
"per_layer": {
|
| 4 |
+
"time_mlp.1.weight": {
|
| 5 |
+
"w_norm": 18.912269592285156,
|
| 6 |
+
"g_norm": 0.12332244962453842,
|
| 7 |
+
"shape": [
|
| 8 |
+
1024,
|
| 9 |
+
256
|
| 10 |
+
]
|
| 11 |
+
},
|
| 12 |
+
"time_mlp.1.bias": {
|
| 13 |
+
"w_norm": 1.1593341827392578,
|
| 14 |
+
"g_norm": 0.010901880450546741,
|
| 15 |
+
"shape": [
|
| 16 |
+
1024
|
| 17 |
+
]
|
| 18 |
+
},
|
| 19 |
+
"time_mlp.3.weight": {
|
| 20 |
+
"w_norm": 9.703208923339844,
|
| 21 |
+
"g_norm": 0.01711881533265114,
|
| 22 |
+
"shape": [
|
| 23 |
+
256,
|
| 24 |
+
1024
|
| 25 |
+
]
|
| 26 |
+
},
|
| 27 |
+
"time_mlp.3.bias": {
|
| 28 |
+
"w_norm": 0.2899875342845917,
|
| 29 |
+
"g_norm": 0.006888911128044128,
|
| 30 |
+
"shape": [
|
| 31 |
+
256
|
| 32 |
+
]
|
| 33 |
+
},
|
| 34 |
+
"input_conv.weight": {
|
| 35 |
+
"w_norm": 7.9453887939453125,
|
| 36 |
+
"g_norm": 0.06517047435045242,
|
| 37 |
+
"shape": [
|
| 38 |
+
192,
|
| 39 |
+
6,
|
| 40 |
+
3,
|
| 41 |
+
3
|
| 42 |
+
]
|
| 43 |
+
},
|
| 44 |
+
"input_conv.bias": {
|
| 45 |
+
"w_norm": 1.189807653427124,
|
| 46 |
+
"g_norm": 0.02409413456916809,
|
| 47 |
+
"shape": [
|
| 48 |
+
192
|
| 49 |
+
]
|
| 50 |
+
},
|
| 51 |
+
"enc_blocks.0.norm1.weight": {
|
| 52 |
+
"w_norm": 13.80312442779541,
|
| 53 |
+
"g_norm": 0.007493571378290653,
|
| 54 |
+
"shape": [
|
| 55 |
+
192
|
| 56 |
+
]
|
| 57 |
+
},
|
| 58 |
+
"enc_blocks.0.norm1.bias": {
|
| 59 |
+
"w_norm": 0.09649404138326645,
|
| 60 |
+
"g_norm": 0.011275016702711582,
|
| 61 |
+
"shape": [
|
| 62 |
+
192
|
| 63 |
+
]
|
| 64 |
+
},
|
| 65 |
+
"enc_blocks.0.conv1.weight": {
|
| 66 |
+
"w_norm": 8.424869537353516,
|
| 67 |
+
"g_norm": 0.1024579405784607,
|
| 68 |
+
"shape": [
|
| 69 |
+
192,
|
| 70 |
+
192,
|
| 71 |
+
3,
|
| 72 |
+
3
|
| 73 |
+
]
|
| 74 |
+
},
|
| 75 |
+
"enc_blocks.0.conv1.bias": {
|
| 76 |
+
"w_norm": 0.20213939249515533,
|
| 77 |
+
"g_norm": 0.008418641053140163,
|
| 78 |
+
"shape": [
|
| 79 |
+
192
|
| 80 |
+
]
|
| 81 |
+
},
|
| 82 |
+
"enc_blocks.0.t_proj.weight": {
|
| 83 |
+
"w_norm": 8.089444160461426,
|
| 84 |
+
"g_norm": 0.012320656329393387,
|
| 85 |
+
"shape": [
|
| 86 |
+
192,
|
| 87 |
+
256
|
| 88 |
+
]
|
| 89 |
+
},
|
| 90 |
+
"enc_blocks.0.t_proj.bias": {
|
| 91 |
+
"w_norm": 0.5025810599327087,
|
| 92 |
+
"g_norm": 0.008417835459113121,
|
| 93 |
+
"shape": [
|
| 94 |
+
192
|
| 95 |
+
]
|
| 96 |
+
},
|
| 97 |
+
"enc_blocks.0.norm2.weight": {
|
| 98 |
+
"w_norm": 13.835018157958984,
|
| 99 |
+
"g_norm": 0.005644944962114096,
|
| 100 |
+
"shape": [
|
| 101 |
+
192
|
| 102 |
+
]
|
| 103 |
+
},
|
| 104 |
+
"enc_blocks.0.norm2.bias": {
|
| 105 |
+
"w_norm": 0.06193859502673149,
|
| 106 |
+
"g_norm": 0.0049712806940078735,
|
| 107 |
+
"shape": [
|
| 108 |
+
192
|
| 109 |
+
]
|
| 110 |
+
},
|
| 111 |
+
"enc_blocks.0.conv2.weight": {
|
| 112 |
+
"w_norm": 8.313619613647461,
|
| 113 |
+
"g_norm": 0.04298524931073189,
|
| 114 |
+
"shape": [
|
| 115 |
+
192,
|
| 116 |
+
192,
|
| 117 |
+
3,
|
| 118 |
+
3
|
| 119 |
+
]
|
| 120 |
+
},
|
| 121 |
+
"enc_blocks.0.conv2.bias": {
|
| 122 |
+
"w_norm": 0.18812546133995056,
|
| 123 |
+
"g_norm": 0.0026280551683157682,
|
| 124 |
+
"shape": [
|
| 125 |
+
192
|
| 126 |
+
]
|
| 127 |
+
},
|
| 128 |
+
"enc_blocks.1.norm1.weight": {
|
| 129 |
+
"w_norm": 13.851577758789062,
|
| 130 |
+
"g_norm": 6.118719466030598e-05,
|
| 131 |
+
"shape": [
|
| 132 |
+
192
|
| 133 |
+
]
|
| 134 |
+
},
|
| 135 |
+
"enc_blocks.1.norm1.bias": {
|
| 136 |
+
"w_norm": 0.032818518579006195,
|
| 137 |
+
"g_norm": 5.334233719622716e-05,
|
| 138 |
+
"shape": [
|
| 139 |
+
192
|
| 140 |
+
]
|
| 141 |
+
},
|
| 142 |
+
"enc_blocks.1.conv1.weight": {
|
| 143 |
+
"w_norm": 8.230306625366211,
|
| 144 |
+
"g_norm": 0.0018623622599989176,
|
| 145 |
+
"shape": [
|
| 146 |
+
192,
|
| 147 |
+
192,
|
| 148 |
+
3,
|
| 149 |
+
3
|
| 150 |
+
]
|
| 151 |
+
},
|
| 152 |
+
"enc_blocks.1.conv1.bias": {
|
| 153 |
+
"w_norm": 0.19205407798290253,
|
| 154 |
+
"g_norm": 8.16969622974284e-05,
|
| 155 |
+
"shape": [
|
| 156 |
+
192
|
| 157 |
+
]
|
| 158 |
+
},
|
| 159 |
+
"enc_blocks.1.t_proj.weight": {
|
| 160 |
+
"w_norm": 8.11756420135498,
|
| 161 |
+
"g_norm": 8.078471728367731e-05,
|
| 162 |
+
"shape": [
|
| 163 |
+
192,
|
| 164 |
+
256
|
| 165 |
+
]
|
| 166 |
+
},
|
| 167 |
+
"enc_blocks.1.t_proj.bias": {
|
| 168 |
+
"w_norm": 0.5217092633247375,
|
| 169 |
+
"g_norm": 8.170954242814332e-05,
|
| 170 |
+
"shape": [
|
| 171 |
+
192
|
| 172 |
+
]
|
| 173 |
+
},
|
| 174 |
+
"enc_blocks.1.norm2.weight": {
|
| 175 |
+
"w_norm": 13.818221092224121,
|
| 176 |
+
"g_norm": 0.00012724759289994836,
|
| 177 |
+
"shape": [
|
| 178 |
+
192
|
| 179 |
+
]
|
| 180 |
+
},
|
| 181 |
+
"enc_blocks.1.norm2.bias": {
|
| 182 |
+
"w_norm": 0.05621594935655594,
|
| 183 |
+
"g_norm": 7.47037265682593e-05,
|
| 184 |
+
"shape": [
|
| 185 |
+
192
|
| 186 |
+
]
|
| 187 |
+
},
|
| 188 |
+
"enc_blocks.1.conv2.weight": {
|
| 189 |
+
"w_norm": 8.19234848022461,
|
| 190 |
+
"g_norm": 0.002554898150265217,
|
| 191 |
+
"shape": [
|
| 192 |
+
192,
|
| 193 |
+
192,
|
| 194 |
+
3,
|
| 195 |
+
3
|
| 196 |
+
]
|
| 197 |
+
},
|
| 198 |
+
"enc_blocks.1.conv2.bias": {
|
| 199 |
+
"w_norm": 0.208110049366951,
|
| 200 |
+
"g_norm": 8.532928768545389e-05,
|
| 201 |
+
"shape": [
|
| 202 |
+
192
|
| 203 |
+
]
|
| 204 |
+
},
|
| 205 |
+
"enc_blocks.2.norm1.weight": {
|
| 206 |
+
"w_norm": 13.827280044555664,
|
| 207 |
+
"g_norm": 4.5593504182761535e-05,
|
| 208 |
+
"shape": [
|
| 209 |
+
192
|
| 210 |
+
]
|
| 211 |
+
},
|
| 212 |
+
"enc_blocks.2.norm1.bias": {
|
| 213 |
+
"w_norm": 0.051892783492803574,
|
| 214 |
+
"g_norm": 4.180220275884494e-05,
|
| 215 |
+
"shape": [
|
| 216 |
+
192
|
| 217 |
+
]
|
| 218 |
+
},
|
| 219 |
+
"enc_blocks.2.conv1.weight": {
|
| 220 |
+
"w_norm": 11.666332244873047,
|
| 221 |
+
"g_norm": 0.001202629879117012,
|
| 222 |
+
"shape": [
|
| 223 |
+
384,
|
| 224 |
+
192,
|
| 225 |
+
3,
|
| 226 |
+
3
|
| 227 |
+
]
|
| 228 |
+
},
|
| 229 |
+
"enc_blocks.2.conv1.bias": {
|
| 230 |
+
"w_norm": 0.2665994167327881,
|
| 231 |
+
"g_norm": 5.171901284484193e-05,
|
| 232 |
+
"shape": [
|
| 233 |
+
384
|
| 234 |
+
]
|
| 235 |
+
},
|
| 236 |
+
"enc_blocks.2.t_proj.weight": {
|
| 237 |
+
"w_norm": 11.719046592712402,
|
| 238 |
+
"g_norm": 4.928291673422791e-05,
|
| 239 |
+
"shape": [
|
| 240 |
+
384,
|
| 241 |
+
256
|
| 242 |
+
]
|
| 243 |
+
},
|
| 244 |
+
"enc_blocks.2.t_proj.bias": {
|
| 245 |
+
"w_norm": 0.7221358418464661,
|
| 246 |
+
"g_norm": 5.172061355551705e-05,
|
| 247 |
+
"shape": [
|
| 248 |
+
384
|
| 249 |
+
]
|
| 250 |
+
},
|
| 251 |
+
"enc_blocks.2.norm2.weight": {
|
| 252 |
+
"w_norm": 19.450790405273438,
|
| 253 |
+
"g_norm": 0.00012247661652509123,
|
| 254 |
+
"shape": [
|
| 255 |
+
384
|
| 256 |
+
]
|
| 257 |
+
},
|
| 258 |
+
"enc_blocks.2.norm2.bias": {
|
| 259 |
+
"w_norm": 0.16962189972400665,
|
| 260 |
+
"g_norm": 5.319947740645148e-05,
|
| 261 |
+
"shape": [
|
| 262 |
+
384
|
| 263 |
+
]
|
| 264 |
+
},
|
| 265 |
+
"enc_blocks.2.conv2.weight": {
|
| 266 |
+
"w_norm": 11.663330078125,
|
| 267 |
+
"g_norm": 0.001001948956400156,
|
| 268 |
+
"shape": [
|
| 269 |
+
384,
|
| 270 |
+
384,
|
| 271 |
+
3,
|
| 272 |
+
3
|
| 273 |
+
]
|
| 274 |
+
},
|
| 275 |
+
"enc_blocks.2.conv2.bias": {
|
| 276 |
+
"w_norm": 0.20253244042396545,
|
| 277 |
+
"g_norm": 2.679444514797069e-05,
|
| 278 |
+
"shape": [
|
| 279 |
+
384
|
| 280 |
+
]
|
| 281 |
+
},
|
| 282 |
+
"enc_blocks.2.skip.weight": {
|
| 283 |
+
"w_norm": 11.206422805786133,
|
| 284 |
+
"g_norm": 0.0009856950491666794,
|
| 285 |
+
"shape": [
|
| 286 |
+
384,
|
| 287 |
+
192,
|
| 288 |
+
1,
|
| 289 |
+
1
|
| 290 |
+
]
|
| 291 |
+
},
|
| 292 |
+
"enc_blocks.2.skip.bias": {
|
| 293 |
+
"w_norm": 0.827253520488739,
|
| 294 |
+
"g_norm": 2.679444514797069e-05,
|
| 295 |
+
"shape": [
|
| 296 |
+
384
|
| 297 |
+
]
|
| 298 |
+
},
|
| 299 |
+
"enc_blocks.3.norm1.weight": {
|
| 300 |
+
"w_norm": 19.589677810668945,
|
| 301 |
+
"g_norm": 2.0965567273378838e-06,
|
| 302 |
+
"shape": [
|
| 303 |
+
384
|
| 304 |
+
]
|
| 305 |
+
},
|
| 306 |
+
"enc_blocks.3.norm1.bias": {
|
| 307 |
+
"w_norm": 0.04693835228681564,
|
| 308 |
+
"g_norm": 2.1864386781089706e-06,
|
| 309 |
+
"shape": [
|
| 310 |
+
384
|
| 311 |
+
]
|
| 312 |
+
},
|
| 313 |
+
"enc_blocks.3.conv1.weight": {
|
| 314 |
+
"w_norm": 11.792016983032227,
|
| 315 |
+
"g_norm": 6.856802065158263e-05,
|
| 316 |
+
"shape": [
|
| 317 |
+
384,
|
| 318 |
+
384,
|
| 319 |
+
3,
|
| 320 |
+
3
|
| 321 |
+
]
|
| 322 |
+
},
|
| 323 |
+
"enc_blocks.3.conv1.bias": {
|
| 324 |
+
"w_norm": 0.1976812332868576,
|
| 325 |
+
"g_norm": 2.096104935844778e-06,
|
| 326 |
+
"shape": [
|
| 327 |
+
384
|
| 328 |
+
]
|
| 329 |
+
},
|
| 330 |
+
"enc_blocks.3.t_proj.weight": {
|
| 331 |
+
"w_norm": 11.38448429107666,
|
| 332 |
+
"g_norm": 2.964314489872777e-06,
|
| 333 |
+
"shape": [
|
| 334 |
+
384,
|
| 335 |
+
256
|
| 336 |
+
]
|
| 337 |
+
},
|
| 338 |
+
"enc_blocks.3.t_proj.bias": {
|
| 339 |
+
"w_norm": 0.6924961805343628,
|
| 340 |
+
"g_norm": 2.096133584927884e-06,
|
| 341 |
+
"shape": [
|
| 342 |
+
384
|
| 343 |
+
]
|
| 344 |
+
},
|
| 345 |
+
"enc_blocks.3.norm2.weight": {
|
| 346 |
+
"w_norm": 19.44411277770996,
|
| 347 |
+
"g_norm": 2.8147405828349292e-06,
|
| 348 |
+
"shape": [
|
| 349 |
+
384
|
| 350 |
+
]
|
| 351 |
+
},
|
| 352 |
+
"enc_blocks.3.norm2.bias": {
|
| 353 |
+
"w_norm": 0.1529274880886078,
|
| 354 |
+
"g_norm": 2.841748937498778e-06,
|
| 355 |
+
"shape": [
|
| 356 |
+
384
|
| 357 |
+
]
|
| 358 |
+
},
|
| 359 |
+
"enc_blocks.3.conv2.weight": {
|
| 360 |
+
"w_norm": 11.47637939453125,
|
| 361 |
+
"g_norm": 0.00019026761583518237,
|
| 362 |
+
"shape": [
|
| 363 |
+
384,
|
| 364 |
+
384,
|
| 365 |
+
3,
|
| 366 |
+
3
|
| 367 |
+
]
|
| 368 |
+
},
|
| 369 |
+
"enc_blocks.3.conv2.bias": {
|
| 370 |
+
"w_norm": 0.1905771642923355,
|
| 371 |
+
"g_norm": 6.41557244307478e-06,
|
| 372 |
+
"shape": [
|
| 373 |
+
384
|
| 374 |
+
]
|
| 375 |
+
},
|
| 376 |
+
"enc_blocks.4.norm1.weight": {
|
| 377 |
+
"w_norm": 19.593414306640625,
|
| 378 |
+
"g_norm": 1.7410387727068155e-06,
|
| 379 |
+
"shape": [
|
| 380 |
+
384
|
| 381 |
+
]
|
| 382 |
+
},
|
| 383 |
+
"enc_blocks.4.norm1.bias": {
|
| 384 |
+
"w_norm": 0.05097125843167305,
|
| 385 |
+
"g_norm": 1.8455435792930075e-06,
|
| 386 |
+
"shape": [
|
| 387 |
+
384
|
| 388 |
+
]
|
| 389 |
+
},
|
| 390 |
+
"enc_blocks.4.conv1.weight": {
|
| 391 |
+
"w_norm": 16.576250076293945,
|
| 392 |
+
"g_norm": 5.6877506722230464e-05,
|
| 393 |
+
"shape": [
|
| 394 |
+
768,
|
| 395 |
+
384,
|
| 396 |
+
3,
|
| 397 |
+
3
|
| 398 |
+
]
|
| 399 |
+
},
|
| 400 |
+
"enc_blocks.4.conv1.bias": {
|
| 401 |
+
"w_norm": 0.28481438755989075,
|
| 402 |
+
"g_norm": 1.893393459795334e-06,
|
| 403 |
+
"shape": [
|
| 404 |
+
768
|
| 405 |
+
]
|
| 406 |
+
},
|
| 407 |
+
"enc_blocks.4.t_proj.weight": {
|
| 408 |
+
"w_norm": 16.05063247680664,
|
| 409 |
+
"g_norm": 2.453651859468664e-06,
|
| 410 |
+
"shape": [
|
| 411 |
+
768,
|
| 412 |
+
256
|
| 413 |
+
]
|
| 414 |
+
},
|
| 415 |
+
"enc_blocks.4.t_proj.bias": {
|
| 416 |
+
"w_norm": 0.9925292730331421,
|
| 417 |
+
"g_norm": 1.8932430521090282e-06,
|
| 418 |
+
"shape": [
|
| 419 |
+
768
|
| 420 |
+
]
|
| 421 |
+
},
|
| 422 |
+
"enc_blocks.4.norm2.weight": {
|
| 423 |
+
"w_norm": 27.487300872802734,
|
| 424 |
+
"g_norm": 2.2128754153527552e-06,
|
| 425 |
+
"shape": [
|
| 426 |
+
768
|
| 427 |
+
]
|
| 428 |
+
},
|
| 429 |
+
"enc_blocks.4.norm2.bias": {
|
| 430 |
+
"w_norm": 0.23648281395435333,
|
| 431 |
+
"g_norm": 2.872084451155388e-06,
|
| 432 |
+
"shape": [
|
| 433 |
+
768
|
| 434 |
+
]
|
| 435 |
+
},
|
| 436 |
+
"enc_blocks.4.conv2.weight": {
|
| 437 |
+
"w_norm": 16.297828674316406,
|
| 438 |
+
"g_norm": 0.00014143867883831263,
|
| 439 |
+
"shape": [
|
| 440 |
+
768,
|
| 441 |
+
768,
|
| 442 |
+
3,
|
| 443 |
+
3
|
| 444 |
+
]
|
| 445 |
+
},
|
| 446 |
+
"enc_blocks.4.conv2.bias": {
|
| 447 |
+
"w_norm": 0.19452296197414398,
|
| 448 |
+
"g_norm": 4.4945381887373514e-06,
|
| 449 |
+
"shape": [
|
| 450 |
+
768
|
| 451 |
+
]
|
| 452 |
+
},
|
| 453 |
+
"enc_blocks.4.skip.weight": {
|
| 454 |
+
"w_norm": 15.839735984802246,
|
| 455 |
+
"g_norm": 0.00015103576879482716,
|
| 456 |
+
"shape": [
|
| 457 |
+
768,
|
| 458 |
+
384,
|
| 459 |
+
1,
|
| 460 |
+
1
|
| 461 |
+
]
|
| 462 |
+
},
|
| 463 |
+
"enc_blocks.4.skip.bias": {
|
| 464 |
+
"w_norm": 0.8077318072319031,
|
| 465 |
+
"g_norm": 4.4945381887373514e-06,
|
| 466 |
+
"shape": [
|
| 467 |
+
768
|
| 468 |
+
]
|
| 469 |
+
},
|
| 470 |
+
"enc_blocks.5.norm1.weight": {
|
| 471 |
+
"w_norm": 27.712413787841797,
|
| 472 |
+
"g_norm": 2.0804559426323976e-06,
|
| 473 |
+
"shape": [
|
| 474 |
+
768
|
| 475 |
+
]
|
| 476 |
+
},
|
| 477 |
+
"enc_blocks.5.norm1.bias": {
|
| 478 |
+
"w_norm": 0.05509699136018753,
|
| 479 |
+
"g_norm": 1.9732342479983345e-06,
|
| 480 |
+
"shape": [
|
| 481 |
+
768
|
| 482 |
+
]
|
| 483 |
+
},
|
| 484 |
+
"enc_blocks.5.conv1.weight": {
|
| 485 |
+
"w_norm": 16.845355987548828,
|
| 486 |
+
"g_norm": 4.273263766663149e-05,
|
| 487 |
+
"shape": [
|
| 488 |
+
768,
|
| 489 |
+
768,
|
| 490 |
+
3,
|
| 491 |
+
3
|
| 492 |
+
]
|
| 493 |
+
},
|
| 494 |
+
"enc_blocks.5.conv1.bias": {
|
| 495 |
+
"w_norm": 0.2029583752155304,
|
| 496 |
+
"g_norm": 8.974413958640071e-07,
|
| 497 |
+
"shape": [
|
| 498 |
+
768
|
| 499 |
+
]
|
| 500 |
+
},
|
| 501 |
+
"enc_blocks.5.t_proj.weight": {
|
| 502 |
+
"w_norm": 16.008506774902344,
|
| 503 |
+
"g_norm": 8.729576848054421e-07,
|
| 504 |
+
"shape": [
|
| 505 |
+
768,
|
| 506 |
+
256
|
| 507 |
+
]
|
| 508 |
+
},
|
| 509 |
+
"enc_blocks.5.t_proj.bias": {
|
| 510 |
+
"w_norm": 0.9988090395927429,
|
| 511 |
+
"g_norm": 8.977645507002308e-07,
|
| 512 |
+
"shape": [
|
| 513 |
+
768
|
| 514 |
+
]
|
| 515 |
+
},
|
| 516 |
+
"enc_blocks.5.norm2.weight": {
|
| 517 |
+
"w_norm": 27.54796600341797,
|
| 518 |
+
"g_norm": 2.552696969360113e-06,
|
| 519 |
+
"shape": [
|
| 520 |
+
768
|
| 521 |
+
]
|
| 522 |
+
},
|
| 523 |
+
"enc_blocks.5.norm2.bias": {
|
| 524 |
+
"w_norm": 0.1847270131111145,
|
| 525 |
+
"g_norm": 2.101767222484341e-06,
|
| 526 |
+
"shape": [
|
| 527 |
+
768
|
| 528 |
+
]
|
| 529 |
+
},
|
| 530 |
+
"enc_blocks.5.conv2.weight": {
|
| 531 |
+
"w_norm": 16.440540313720703,
|
| 532 |
+
"g_norm": 0.00010133799514733255,
|
| 533 |
+
"shape": [
|
| 534 |
+
768,
|
| 535 |
+
768,
|
| 536 |
+
3,
|
| 537 |
+
3
|
| 538 |
+
]
|
| 539 |
+
},
|
| 540 |
+
"enc_blocks.5.conv2.bias": {
|
| 541 |
+
"w_norm": 0.19306622445583344,
|
| 542 |
+
"g_norm": 2.4288849544973345e-06,
|
| 543 |
+
"shape": [
|
| 544 |
+
768
|
| 545 |
+
]
|
| 546 |
+
},
|
| 547 |
+
"enc_attns.3.norm.weight": {
|
| 548 |
+
"w_norm": 19.538881301879883,
|
| 549 |
+
"g_norm": 1.6037432715165778e-06,
|
| 550 |
+
"shape": [
|
| 551 |
+
384
|
| 552 |
+
]
|
| 553 |
+
},
|
| 554 |
+
"enc_attns.3.norm.bias": {
|
| 555 |
+
"w_norm": 0.04634387046098709,
|
| 556 |
+
"g_norm": 2.5036943043232895e-06,
|
| 557 |
+
"shape": [
|
| 558 |
+
384
|
| 559 |
+
]
|
| 560 |
+
},
|
| 561 |
+
"enc_attns.3.qkv.weight": {
|
| 562 |
+
"w_norm": 19.671646118164062,
|
| 563 |
+
"g_norm": 4.9293481424683705e-05,
|
| 564 |
+
"shape": [
|
| 565 |
+
1152,
|
| 566 |
+
384,
|
| 567 |
+
1,
|
| 568 |
+
1
|
| 569 |
+
]
|
| 570 |
+
},
|
| 571 |
+
"enc_attns.3.qkv.bias": {
|
| 572 |
+
"w_norm": 1.0068951845169067,
|
| 573 |
+
"g_norm": 3.873734385706484e-06,
|
| 574 |
+
"shape": [
|
| 575 |
+
1152
|
| 576 |
+
]
|
| 577 |
+
},
|
| 578 |
+
"enc_attns.3.proj.weight": {
|
| 579 |
+
"w_norm": 11.309000968933105,
|
| 580 |
+
"g_norm": 4.3482283217599615e-05,
|
| 581 |
+
"shape": [
|
| 582 |
+
384,
|
| 583 |
+
384,
|
| 584 |
+
1,
|
| 585 |
+
1
|
| 586 |
+
]
|
| 587 |
+
},
|
| 588 |
+
"enc_attns.3.proj.bias": {
|
| 589 |
+
"w_norm": 0.5629428625106812,
|
| 590 |
+
"g_norm": 6.388872861862183e-06,
|
| 591 |
+
"shape": [
|
| 592 |
+
384
|
| 593 |
+
]
|
| 594 |
+
},
|
| 595 |
+
"enc_attns.4.norm.weight": {
|
| 596 |
+
"w_norm": 27.623943328857422,
|
| 597 |
+
"g_norm": 1.3003150343138259e-06,
|
| 598 |
+
"shape": [
|
| 599 |
+
768
|
| 600 |
+
]
|
| 601 |
+
},
|
| 602 |
+
"enc_attns.4.norm.bias": {
|
| 603 |
+
"w_norm": 0.048902470618486404,
|
| 604 |
+
"g_norm": 1.8792874243445112e-06,
|
| 605 |
+
"shape": [
|
| 606 |
+
768
|
| 607 |
+
]
|
| 608 |
+
},
|
| 609 |
+
"enc_attns.4.qkv.weight": {
|
| 610 |
+
"w_norm": 27.94339942932129,
|
| 611 |
+
"g_norm": 5.037298615206964e-05,
|
| 612 |
+
"shape": [
|
| 613 |
+
2304,
|
| 614 |
+
768,
|
| 615 |
+
1,
|
| 616 |
+
1
|
| 617 |
+
]
|
| 618 |
+
},
|
| 619 |
+
"enc_attns.4.qkv.bias": {
|
| 620 |
+
"w_norm": 1.0119142532348633,
|
| 621 |
+
"g_norm": 2.70444297711947e-06,
|
| 622 |
+
"shape": [
|
| 623 |
+
2304
|
| 624 |
+
]
|
| 625 |
+
},
|
| 626 |
+
"enc_attns.4.proj.weight": {
|
| 627 |
+
"w_norm": 16.00703239440918,
|
| 628 |
+
"g_norm": 5.209520168136805e-05,
|
| 629 |
+
"shape": [
|
| 630 |
+
768,
|
| 631 |
+
768,
|
| 632 |
+
1,
|
| 633 |
+
1
|
| 634 |
+
]
|
| 635 |
+
},
|
| 636 |
+
"enc_attns.4.proj.bias": {
|
| 637 |
+
"w_norm": 0.5728441476821899,
|
| 638 |
+
"g_norm": 4.516856279224157e-06,
|
| 639 |
+
"shape": [
|
| 640 |
+
768
|
| 641 |
+
]
|
| 642 |
+
},
|
| 643 |
+
"enc_attns.5.norm.weight": {
|
| 644 |
+
"w_norm": 27.662988662719727,
|
| 645 |
+
"g_norm": 3.406920257020829e-07,
|
| 646 |
+
"shape": [
|
| 647 |
+
768
|
| 648 |
+
]
|
| 649 |
+
},
|
| 650 |
+
"enc_attns.5.norm.bias": {
|
| 651 |
+
"w_norm": 0.04228585958480835,
|
| 652 |
+
"g_norm": 8.539649343219935e-07,
|
| 653 |
+
"shape": [
|
| 654 |
+
768
|
| 655 |
+
]
|
| 656 |
+
},
|
| 657 |
+
"enc_attns.5.qkv.weight": {
|
| 658 |
+
"w_norm": 27.92650604248047,
|
| 659 |
+
"g_norm": 1.5280100342351943e-05,
|
| 660 |
+
"shape": [
|
| 661 |
+
2304,
|
| 662 |
+
768,
|
| 663 |
+
1,
|
| 664 |
+
1
|
| 665 |
+
]
|
| 666 |
+
},
|
| 667 |
+
"enc_attns.5.qkv.bias": {
|
| 668 |
+
"w_norm": 0.9987878799438477,
|
| 669 |
+
"g_norm": 1.4080259234106052e-06,
|
| 670 |
+
"shape": [
|
| 671 |
+
2304
|
| 672 |
+
]
|
| 673 |
+
},
|
| 674 |
+
"enc_attns.5.proj.weight": {
|
| 675 |
+
"w_norm": 16.03470802307129,
|
| 676 |
+
"g_norm": 1.587315637152642e-05,
|
| 677 |
+
"shape": [
|
| 678 |
+
768,
|
| 679 |
+
768,
|
| 680 |
+
1,
|
| 681 |
+
1
|
| 682 |
+
]
|
| 683 |
+
},
|
| 684 |
+
"enc_attns.5.proj.bias": {
|
| 685 |
+
"w_norm": 0.5973420143127441,
|
| 686 |
+
"g_norm": 2.459404413457378e-06,
|
| 687 |
+
"shape": [
|
| 688 |
+
768
|
| 689 |
+
]
|
| 690 |
+
},
|
| 691 |
+
"enc_downs.0.op.weight": {
|
| 692 |
+
"w_norm": 8.154297828674316,
|
| 693 |
+
"g_norm": 0.004857305437326431,
|
| 694 |
+
"shape": [
|
| 695 |
+
192,
|
| 696 |
+
192,
|
| 697 |
+
3,
|
| 698 |
+
3
|
| 699 |
+
]
|
| 700 |
+
},
|
| 701 |
+
"enc_downs.0.op.bias": {
|
| 702 |
+
"w_norm": 0.18503881990909576,
|
| 703 |
+
"g_norm": 9.78953976300545e-05,
|
| 704 |
+
"shape": [
|
| 705 |
+
192
|
| 706 |
+
]
|
| 707 |
+
},
|
| 708 |
+
"enc_downs.1.op.weight": {
|
| 709 |
+
"w_norm": 8.09019947052002,
|
| 710 |
+
"g_norm": 0.0018013118533417583,
|
| 711 |
+
"shape": [
|
| 712 |
+
192,
|
| 713 |
+
192,
|
| 714 |
+
3,
|
| 715 |
+
3
|
| 716 |
+
]
|
| 717 |
+
},
|
| 718 |
+
"enc_downs.1.op.bias": {
|
| 719 |
+
"w_norm": 0.19495517015457153,
|
| 720 |
+
"g_norm": 2.7523738026502542e-05,
|
| 721 |
+
"shape": [
|
| 722 |
+
192
|
| 723 |
+
]
|
| 724 |
+
},
|
| 725 |
+
"enc_downs.2.op.weight": {
|
| 726 |
+
"w_norm": 11.496183395385742,
|
| 727 |
+
"g_norm": 0.0004138555086683482,
|
| 728 |
+
"shape": [
|
| 729 |
+
384,
|
| 730 |
+
384,
|
| 731 |
+
3,
|
| 732 |
+
3
|
| 733 |
+
]
|
| 734 |
+
},
|
| 735 |
+
"enc_downs.2.op.bias": {
|
| 736 |
+
"w_norm": 0.18931472301483154,
|
| 737 |
+
"g_norm": 6.417965778382495e-06,
|
| 738 |
+
"shape": [
|
| 739 |
+
384
|
| 740 |
+
]
|
| 741 |
+
},
|
| 742 |
+
"enc_downs.3.op.weight": {
|
| 743 |
+
"w_norm": 11.428959846496582,
|
| 744 |
+
"g_norm": 0.00020830711582675576,
|
| 745 |
+
"shape": [
|
| 746 |
+
384,
|
| 747 |
+
384,
|
| 748 |
+
3,
|
| 749 |
+
3
|
| 750 |
+
]
|
| 751 |
+
},
|
| 752 |
+
"enc_downs.3.op.bias": {
|
| 753 |
+
"w_norm": 0.1918579638004303,
|
| 754 |
+
"g_norm": 2.7779085485235555e-06,
|
| 755 |
+
"shape": [
|
| 756 |
+
384
|
| 757 |
+
]
|
| 758 |
+
},
|
| 759 |
+
"enc_downs.4.op.weight": {
|
| 760 |
+
"w_norm": 16.336414337158203,
|
| 761 |
+
"g_norm": 0.0002319672203157097,
|
| 762 |
+
"shape": [
|
| 763 |
+
768,
|
| 764 |
+
768,
|
| 765 |
+
3,
|
| 766 |
+
3
|
| 767 |
+
]
|
| 768 |
+
},
|
| 769 |
+
"enc_downs.4.op.bias": {
|
| 770 |
+
"w_norm": 0.1980322301387787,
|
| 771 |
+
"g_norm": 2.4985531581478426e-06,
|
| 772 |
+
"shape": [
|
| 773 |
+
768
|
| 774 |
+
]
|
| 775 |
+
},
|
| 776 |
+
"mid1.norm1.weight": {
|
| 777 |
+
"w_norm": 27.713109970092773,
|
| 778 |
+
"g_norm": 5.832573606312508e-07,
|
| 779 |
+
"shape": [
|
| 780 |
+
768
|
| 781 |
+
]
|
| 782 |
+
},
|
| 783 |
+
"mid1.norm1.bias": {
|
| 784 |
+
"w_norm": 0.04988333582878113,
|
| 785 |
+
"g_norm": 4.770976715917641e-07,
|
| 786 |
+
"shape": [
|
| 787 |
+
768
|
| 788 |
+
]
|
| 789 |
+
},
|
| 790 |
+
"mid1.conv1.weight": {
|
| 791 |
+
"w_norm": 16.79037094116211,
|
| 792 |
+
"g_norm": 2.5980823920690455e-05,
|
| 793 |
+
"shape": [
|
| 794 |
+
768,
|
| 795 |
+
768,
|
| 796 |
+
3,
|
| 797 |
+
3
|
| 798 |
+
]
|
| 799 |
+
},
|
| 800 |
+
"mid1.conv1.bias": {
|
| 801 |
+
"w_norm": 0.20654506981372833,
|
| 802 |
+
"g_norm": 5.159295710654987e-07,
|
| 803 |
+
"shape": [
|
| 804 |
+
768
|
| 805 |
+
]
|
| 806 |
+
},
|
| 807 |
+
"mid1.t_proj.weight": {
|
| 808 |
+
"w_norm": 16.01586151123047,
|
| 809 |
+
"g_norm": 5.154277573637955e-07,
|
| 810 |
+
"shape": [
|
| 811 |
+
768,
|
| 812 |
+
256
|
| 813 |
+
]
|
| 814 |
+
},
|
| 815 |
+
"mid1.t_proj.bias": {
|
| 816 |
+
"w_norm": 1.031219482421875,
|
| 817 |
+
"g_norm": 5.159647002983547e-07,
|
| 818 |
+
"shape": [
|
| 819 |
+
768
|
| 820 |
+
]
|
| 821 |
+
},
|
| 822 |
+
"mid1.norm2.weight": {
|
| 823 |
+
"w_norm": 27.551307678222656,
|
| 824 |
+
"g_norm": 7.436812552441552e-07,
|
| 825 |
+
"shape": [
|
| 826 |
+
768
|
| 827 |
+
]
|
| 828 |
+
},
|
| 829 |
+
"mid1.norm2.bias": {
|
| 830 |
+
"w_norm": 0.20324093103408813,
|
| 831 |
+
"g_norm": 1.0805827059812145e-06,
|
| 832 |
+
"shape": [
|
| 833 |
+
768
|
| 834 |
+
]
|
| 835 |
+
},
|
| 836 |
+
"mid1.conv2.weight": {
|
| 837 |
+
"w_norm": 16.460031509399414,
|
| 838 |
+
"g_norm": 5.092180072097108e-05,
|
| 839 |
+
"shape": [
|
| 840 |
+
768,
|
| 841 |
+
768,
|
| 842 |
+
3,
|
| 843 |
+
3
|
| 844 |
+
]
|
| 845 |
+
},
|
| 846 |
+
"mid1.conv2.bias": {
|
| 847 |
+
"w_norm": 0.1937713921070099,
|
| 848 |
+
"g_norm": 1.829196662583854e-06,
|
| 849 |
+
"shape": [
|
| 850 |
+
768
|
| 851 |
+
]
|
| 852 |
+
},
|
| 853 |
+
"mid_attn.norm.weight": {
|
| 854 |
+
"w_norm": 27.66978645324707,
|
| 855 |
+
"g_norm": 2.9024468517491187e-07,
|
| 856 |
+
"shape": [
|
| 857 |
+
768
|
| 858 |
+
]
|
| 859 |
+
},
|
| 860 |
+
"mid_attn.norm.bias": {
|
| 861 |
+
"w_norm": 0.044777318835258484,
|
| 862 |
+
"g_norm": 6.936780891919625e-07,
|
| 863 |
+
"shape": [
|
| 864 |
+
768
|
| 865 |
+
]
|
| 866 |
+
},
|
| 867 |
+
"mid_attn.qkv.weight": {
|
| 868 |
+
"w_norm": 27.8992919921875,
|
| 869 |
+
"g_norm": 1.2187340871605556e-05,
|
| 870 |
+
"shape": [
|
| 871 |
+
2304,
|
| 872 |
+
768,
|
| 873 |
+
1,
|
| 874 |
+
1
|
| 875 |
+
]
|
| 876 |
+
},
|
| 877 |
+
"mid_attn.qkv.bias": {
|
| 878 |
+
"w_norm": 0.9861307144165039,
|
| 879 |
+
"g_norm": 1.0868640174521715e-06,
|
| 880 |
+
"shape": [
|
| 881 |
+
2304
|
| 882 |
+
]
|
| 883 |
+
},
|
| 884 |
+
"mid_attn.proj.weight": {
|
| 885 |
+
"w_norm": 16.032732009887695,
|
| 886 |
+
"g_norm": 1.2030658581352327e-05,
|
| 887 |
+
"shape": [
|
| 888 |
+
768,
|
| 889 |
+
768,
|
| 890 |
+
1,
|
| 891 |
+
1
|
| 892 |
+
]
|
| 893 |
+
},
|
| 894 |
+
"mid_attn.proj.bias": {
|
| 895 |
+
"w_norm": 0.576869010925293,
|
| 896 |
+
"g_norm": 1.8453860093359253e-06,
|
| 897 |
+
"shape": [
|
| 898 |
+
768
|
| 899 |
+
]
|
| 900 |
+
},
|
| 901 |
+
"mid2.norm1.weight": {
|
| 902 |
+
"w_norm": 27.717756271362305,
|
| 903 |
+
"g_norm": 1.05101287317666e-06,
|
| 904 |
+
"shape": [
|
| 905 |
+
768
|
| 906 |
+
]
|
| 907 |
+
},
|
| 908 |
+
"mid2.norm1.bias": {
|
| 909 |
+
"w_norm": 0.06068818271160126,
|
| 910 |
+
"g_norm": 1.0774595011753263e-06,
|
| 911 |
+
"shape": [
|
| 912 |
+
768
|
| 913 |
+
]
|
| 914 |
+
},
|
| 915 |
+
"mid2.conv1.weight": {
|
| 916 |
+
"w_norm": 16.839242935180664,
|
| 917 |
+
"g_norm": 2.6506331778364256e-05,
|
| 918 |
+
"shape": [
|
| 919 |
+
768,
|
| 920 |
+
768,
|
| 921 |
+
3,
|
| 922 |
+
3
|
| 923 |
+
]
|
| 924 |
+
},
|
| 925 |
+
"mid2.conv1.bias": {
|
| 926 |
+
"w_norm": 0.19942057132720947,
|
| 927 |
+
"g_norm": 5.999360723762948e-07,
|
| 928 |
+
"shape": [
|
| 929 |
+
768
|
| 930 |
+
]
|
| 931 |
+
},
|
| 932 |
+
"mid2.t_proj.weight": {
|
| 933 |
+
"w_norm": 16.003231048583984,
|
| 934 |
+
"g_norm": 6.345590009004809e-07,
|
| 935 |
+
"shape": [
|
| 936 |
+
768,
|
| 937 |
+
256
|
| 938 |
+
]
|
| 939 |
+
},
|
| 940 |
+
"mid2.t_proj.bias": {
|
| 941 |
+
"w_norm": 0.9975320100784302,
|
| 942 |
+
"g_norm": 5.998435312903894e-07,
|
| 943 |
+
"shape": [
|
| 944 |
+
768
|
| 945 |
+
]
|
| 946 |
+
},
|
| 947 |
+
"mid2.norm2.weight": {
|
| 948 |
+
"w_norm": 27.539823532104492,
|
| 949 |
+
"g_norm": 1.1429430060161394e-06,
|
| 950 |
+
"shape": [
|
| 951 |
+
768
|
| 952 |
+
]
|
| 953 |
+
},
|
| 954 |
+
"mid2.norm2.bias": {
|
| 955 |
+
"w_norm": 0.2095635086297989,
|
| 956 |
+
"g_norm": 1.4760953490622342e-06,
|
| 957 |
+
"shape": [
|
| 958 |
+
768
|
| 959 |
+
]
|
| 960 |
+
},
|
| 961 |
+
"mid2.conv2.weight": {
|
| 962 |
+
"w_norm": 16.427698135375977,
|
| 963 |
+
"g_norm": 4.7607252781745046e-05,
|
| 964 |
+
"shape": [
|
| 965 |
+
768,
|
| 966 |
+
768,
|
| 967 |
+
3,
|
| 968 |
+
3
|
| 969 |
+
]
|
| 970 |
+
},
|
| 971 |
+
"mid2.conv2.bias": {
|
| 972 |
+
"w_norm": 0.19000190496444702,
|
| 973 |
+
"g_norm": 1.873412543318409e-06,
|
| 974 |
+
"shape": [
|
| 975 |
+
768
|
| 976 |
+
]
|
| 977 |
+
},
|
| 978 |
+
"dec_blocks.0.norm1.weight": {
|
| 979 |
+
"w_norm": 39.21333312988281,
|
| 980 |
+
"g_norm": 1.0148718274649582e-06,
|
| 981 |
+
"shape": [
|
| 982 |
+
1536
|
| 983 |
+
]
|
| 984 |
+
},
|
| 985 |
+
"dec_blocks.0.norm1.bias": {
|
| 986 |
+
"w_norm": 0.09431219100952148,
|
| 987 |
+
"g_norm": 9.196940595757042e-07,
|
| 988 |
+
"shape": [
|
| 989 |
+
1536
|
| 990 |
+
]
|
| 991 |
+
},
|
| 992 |
+
"dec_blocks.0.conv1.weight": {
|
| 993 |
+
"w_norm": 17.71900177001953,
|
| 994 |
+
"g_norm": 3.547394589986652e-05,
|
| 995 |
+
"shape": [
|
| 996 |
+
768,
|
| 997 |
+
1536,
|
| 998 |
+
3,
|
| 999 |
+
3
|
| 1000 |
+
]
|
| 1001 |
+
},
|
| 1002 |
+
"dec_blocks.0.conv1.bias": {
|
| 1003 |
+
"w_norm": 0.1529998481273651,
|
| 1004 |
+
"g_norm": 5.319739102560561e-07,
|
| 1005 |
+
"shape": [
|
| 1006 |
+
768
|
| 1007 |
+
]
|
| 1008 |
+
},
|
| 1009 |
+
"dec_blocks.0.t_proj.weight": {
|
| 1010 |
+
"w_norm": 16.024818420410156,
|
| 1011 |
+
"g_norm": 6.486535539806937e-07,
|
| 1012 |
+
"shape": [
|
| 1013 |
+
768,
|
| 1014 |
+
256
|
| 1015 |
+
]
|
| 1016 |
+
},
|
| 1017 |
+
"dec_blocks.0.t_proj.bias": {
|
| 1018 |
+
"w_norm": 1.02411949634552,
|
| 1019 |
+
"g_norm": 5.319541855897114e-07,
|
| 1020 |
+
"shape": [
|
| 1021 |
+
768
|
| 1022 |
+
]
|
| 1023 |
+
},
|
| 1024 |
+
"dec_blocks.0.norm2.weight": {
|
| 1025 |
+
"w_norm": 27.482181549072266,
|
| 1026 |
+
"g_norm": 1.8672670876185293e-06,
|
| 1027 |
+
"shape": [
|
| 1028 |
+
768
|
| 1029 |
+
]
|
| 1030 |
+
},
|
| 1031 |
+
"dec_blocks.0.norm2.bias": {
|
| 1032 |
+
"w_norm": 0.2610810697078705,
|
| 1033 |
+
"g_norm": 3.00572719424963e-06,
|
| 1034 |
+
"shape": [
|
| 1035 |
+
768
|
| 1036 |
+
]
|
| 1037 |
+
},
|
| 1038 |
+
"dec_blocks.0.conv2.weight": {
|
| 1039 |
+
"w_norm": 16.24523162841797,
|
| 1040 |
+
"g_norm": 9.572529961587861e-05,
|
| 1041 |
+
"shape": [
|
| 1042 |
+
768,
|
| 1043 |
+
768,
|
| 1044 |
+
3,
|
| 1045 |
+
3
|
| 1046 |
+
]
|
| 1047 |
+
},
|
| 1048 |
+
"dec_blocks.0.conv2.bias": {
|
| 1049 |
+
"w_norm": 0.19820284843444824,
|
| 1050 |
+
"g_norm": 4.452592747838935e-06,
|
| 1051 |
+
"shape": [
|
| 1052 |
+
768
|
| 1053 |
+
]
|
| 1054 |
+
},
|
| 1055 |
+
"dec_blocks.0.skip.weight": {
|
| 1056 |
+
"w_norm": 15.814351081848145,
|
| 1057 |
+
"g_norm": 0.00045108215999789536,
|
| 1058 |
+
"shape": [
|
| 1059 |
+
768,
|
| 1060 |
+
1536,
|
| 1061 |
+
1,
|
| 1062 |
+
1
|
| 1063 |
+
]
|
| 1064 |
+
},
|
| 1065 |
+
"dec_blocks.0.skip.bias": {
|
| 1066 |
+
"w_norm": 0.4101272523403168,
|
| 1067 |
+
"g_norm": 4.452592747838935e-06,
|
| 1068 |
+
"shape": [
|
| 1069 |
+
768
|
| 1070 |
+
]
|
| 1071 |
+
},
|
| 1072 |
+
"dec_blocks.1.norm1.weight": {
|
| 1073 |
+
"w_norm": 39.209964752197266,
|
| 1074 |
+
"g_norm": 3.3631847600190667e-06,
|
| 1075 |
+
"shape": [
|
| 1076 |
+
1536
|
| 1077 |
+
]
|
| 1078 |
+
},
|
| 1079 |
+
"dec_blocks.1.norm1.bias": {
|
| 1080 |
+
"w_norm": 0.08060162514448166,
|
| 1081 |
+
"g_norm": 2.9157504286558833e-06,
|
| 1082 |
+
"shape": [
|
| 1083 |
+
1536
|
| 1084 |
+
]
|
| 1085 |
+
},
|
| 1086 |
+
"dec_blocks.1.conv1.weight": {
|
| 1087 |
+
"w_norm": 17.587966918945312,
|
| 1088 |
+
"g_norm": 0.0001042117946781218,
|
| 1089 |
+
"shape": [
|
| 1090 |
+
768,
|
| 1091 |
+
1536,
|
| 1092 |
+
3,
|
| 1093 |
+
3
|
| 1094 |
+
]
|
| 1095 |
+
},
|
| 1096 |
+
"dec_blocks.1.conv1.bias": {
|
| 1097 |
+
"w_norm": 0.14630834758281708,
|
| 1098 |
+
"g_norm": 1.5931634607113665e-06,
|
| 1099 |
+
"shape": [
|
| 1100 |
+
768
|
| 1101 |
+
]
|
| 1102 |
+
},
|
| 1103 |
+
"dec_blocks.1.t_proj.weight": {
|
| 1104 |
+
"w_norm": 16.054262161254883,
|
| 1105 |
+
"g_norm": 1.2944892660016194e-06,
|
| 1106 |
+
"shape": [
|
| 1107 |
+
768,
|
| 1108 |
+
256
|
| 1109 |
+
]
|
| 1110 |
+
},
|
| 1111 |
+
"dec_blocks.1.t_proj.bias": {
|
| 1112 |
+
"w_norm": 0.9880488514900208,
|
| 1113 |
+
"g_norm": 1.5929435903672129e-06,
|
| 1114 |
+
"shape": [
|
| 1115 |
+
768
|
| 1116 |
+
]
|
| 1117 |
+
},
|
| 1118 |
+
"dec_blocks.1.norm2.weight": {
|
| 1119 |
+
"w_norm": 27.479366302490234,
|
| 1120 |
+
"g_norm": 3.6664212075265823e-06,
|
| 1121 |
+
"shape": [
|
| 1122 |
+
768
|
| 1123 |
+
]
|
| 1124 |
+
},
|
| 1125 |
+
"dec_blocks.1.norm2.bias": {
|
| 1126 |
+
"w_norm": 0.2891330122947693,
|
| 1127 |
+
"g_norm": 5.97324697082513e-06,
|
| 1128 |
+
"shape": [
|
| 1129 |
+
768
|
| 1130 |
+
]
|
| 1131 |
+
},
|
| 1132 |
+
"dec_blocks.1.conv2.weight": {
|
| 1133 |
+
"w_norm": 16.209659576416016,
|
| 1134 |
+
"g_norm": 0.00018095829000230879,
|
| 1135 |
+
"shape": [
|
| 1136 |
+
768,
|
| 1137 |
+
768,
|
| 1138 |
+
3,
|
| 1139 |
+
3
|
| 1140 |
+
]
|
| 1141 |
+
},
|
| 1142 |
+
"dec_blocks.1.conv2.bias": {
|
| 1143 |
+
"w_norm": 0.19671788811683655,
|
| 1144 |
+
"g_norm": 9.126384611590765e-06,
|
| 1145 |
+
"shape": [
|
| 1146 |
+
768
|
| 1147 |
+
]
|
| 1148 |
+
},
|
| 1149 |
+
"dec_blocks.1.skip.weight": {
|
| 1150 |
+
"w_norm": 15.800586700439453,
|
| 1151 |
+
"g_norm": 0.0006704161060042679,
|
| 1152 |
+
"shape": [
|
| 1153 |
+
768,
|
| 1154 |
+
1536,
|
| 1155 |
+
1,
|
| 1156 |
+
1
|
| 1157 |
+
]
|
| 1158 |
+
},
|
| 1159 |
+
"dec_blocks.1.skip.bias": {
|
| 1160 |
+
"w_norm": 0.4059737026691437,
|
| 1161 |
+
"g_norm": 9.126384611590765e-06,
|
| 1162 |
+
"shape": [
|
| 1163 |
+
768
|
| 1164 |
+
]
|
| 1165 |
+
},
|
| 1166 |
+
"dec_blocks.2.norm1.weight": {
|
| 1167 |
+
"w_norm": 33.95747375488281,
|
| 1168 |
+
"g_norm": 3.9997848944040015e-06,
|
| 1169 |
+
"shape": [
|
| 1170 |
+
1152
|
| 1171 |
+
]
|
| 1172 |
+
},
|
| 1173 |
+
"dec_blocks.2.norm1.bias": {
|
| 1174 |
+
"w_norm": 0.08351629227399826,
|
| 1175 |
+
"g_norm": 3.245766492909752e-06,
|
| 1176 |
+
"shape": [
|
| 1177 |
+
1152
|
| 1178 |
+
]
|
| 1179 |
+
},
|
| 1180 |
+
"dec_blocks.2.conv1.weight": {
|
| 1181 |
+
"w_norm": 12.456512451171875,
|
| 1182 |
+
"g_norm": 0.00016187748406082392,
|
| 1183 |
+
"shape": [
|
| 1184 |
+
384,
|
| 1185 |
+
1152,
|
| 1186 |
+
3,
|
| 1187 |
+
3
|
| 1188 |
+
]
|
| 1189 |
+
},
|
| 1190 |
+
"dec_blocks.2.conv1.bias": {
|
| 1191 |
+
"w_norm": 0.12158936262130737,
|
| 1192 |
+
"g_norm": 2.730666437855689e-06,
|
| 1193 |
+
"shape": [
|
| 1194 |
+
384
|
| 1195 |
+
]
|
| 1196 |
+
},
|
| 1197 |
+
"dec_blocks.2.t_proj.weight": {
|
| 1198 |
+
"w_norm": 11.345858573913574,
|
| 1199 |
+
"g_norm": 2.7153444079885958e-06,
|
| 1200 |
+
"shape": [
|
| 1201 |
+
384,
|
| 1202 |
+
256
|
| 1203 |
+
]
|
| 1204 |
+
},
|
| 1205 |
+
"dec_blocks.2.t_proj.bias": {
|
| 1206 |
+
"w_norm": 0.7102914452552795,
|
| 1207 |
+
"g_norm": 2.730762389546726e-06,
|
| 1208 |
+
"shape": [
|
| 1209 |
+
384
|
| 1210 |
+
]
|
| 1211 |
+
},
|
| 1212 |
+
"dec_blocks.2.norm2.weight": {
|
| 1213 |
+
"w_norm": 19.454023361206055,
|
| 1214 |
+
"g_norm": 7.419084795401432e-06,
|
| 1215 |
+
"shape": [
|
| 1216 |
+
384
|
| 1217 |
+
]
|
| 1218 |
+
},
|
| 1219 |
+
"dec_blocks.2.norm2.bias": {
|
| 1220 |
+
"w_norm": 0.18189598619937897,
|
| 1221 |
+
"g_norm": 9.322840924141929e-06,
|
| 1222 |
+
"shape": [
|
| 1223 |
+
384
|
| 1224 |
+
]
|
| 1225 |
+
},
|
| 1226 |
+
"dec_blocks.2.conv2.weight": {
|
| 1227 |
+
"w_norm": 11.435891151428223,
|
| 1228 |
+
"g_norm": 0.00031997592304833233,
|
| 1229 |
+
"shape": [
|
| 1230 |
+
384,
|
| 1231 |
+
384,
|
| 1232 |
+
3,
|
| 1233 |
+
3
|
| 1234 |
+
]
|
| 1235 |
+
},
|
| 1236 |
+
"dec_blocks.2.conv2.bias": {
|
| 1237 |
+
"w_norm": 0.18925972282886505,
|
| 1238 |
+
"g_norm": 1.760806844686158e-05,
|
| 1239 |
+
"shape": [
|
| 1240 |
+
384
|
| 1241 |
+
]
|
| 1242 |
+
},
|
| 1243 |
+
"dec_blocks.2.skip.weight": {
|
| 1244 |
+
"w_norm": 11.040159225463867,
|
| 1245 |
+
"g_norm": 0.0008981034043245018,
|
| 1246 |
+
"shape": [
|
| 1247 |
+
384,
|
| 1248 |
+
1152,
|
| 1249 |
+
1,
|
| 1250 |
+
1
|
| 1251 |
+
]
|
| 1252 |
+
},
|
| 1253 |
+
"dec_blocks.2.skip.bias": {
|
| 1254 |
+
"w_norm": 0.3314064145088196,
|
| 1255 |
+
"g_norm": 1.760806844686158e-05,
|
| 1256 |
+
"shape": [
|
| 1257 |
+
384
|
| 1258 |
+
]
|
| 1259 |
+
},
|
| 1260 |
+
"dec_blocks.3.norm1.weight": {
|
| 1261 |
+
"w_norm": 27.715042114257812,
|
| 1262 |
+
"g_norm": 2.8067692255717702e-05,
|
| 1263 |
+
"shape": [
|
| 1264 |
+
768
|
| 1265 |
+
]
|
| 1266 |
+
},
|
| 1267 |
+
"dec_blocks.3.norm1.bias": {
|
| 1268 |
+
"w_norm": 0.07817665487527847,
|
| 1269 |
+
"g_norm": 2.4464015950798057e-05,
|
| 1270 |
+
"shape": [
|
| 1271 |
+
768
|
| 1272 |
+
]
|
| 1273 |
+
},
|
| 1274 |
+
"dec_blocks.3.conv1.weight": {
|
| 1275 |
+
"w_norm": 12.271158218383789,
|
| 1276 |
+
"g_norm": 0.0007932300213724375,
|
| 1277 |
+
"shape": [
|
| 1278 |
+
384,
|
| 1279 |
+
768,
|
| 1280 |
+
3,
|
| 1281 |
+
3
|
| 1282 |
+
]
|
| 1283 |
+
},
|
| 1284 |
+
"dec_blocks.3.conv1.bias": {
|
| 1285 |
+
"w_norm": 0.14353975653648376,
|
| 1286 |
+
"g_norm": 1.6304338714689948e-05,
|
| 1287 |
+
"shape": [
|
| 1288 |
+
384
|
| 1289 |
+
]
|
| 1290 |
+
},
|
| 1291 |
+
"dec_blocks.3.t_proj.weight": {
|
| 1292 |
+
"w_norm": 11.377923011779785,
|
| 1293 |
+
"g_norm": 1.7768579709809273e-05,
|
| 1294 |
+
"shape": [
|
| 1295 |
+
384,
|
| 1296 |
+
256
|
| 1297 |
+
]
|
| 1298 |
+
},
|
| 1299 |
+
"dec_blocks.3.t_proj.bias": {
|
| 1300 |
+
"w_norm": 0.7092573642730713,
|
| 1301 |
+
"g_norm": 1.6306374163832515e-05,
|
| 1302 |
+
"shape": [
|
| 1303 |
+
384
|
| 1304 |
+
]
|
| 1305 |
+
},
|
| 1306 |
+
"dec_blocks.3.norm2.weight": {
|
| 1307 |
+
"w_norm": 19.434457778930664,
|
| 1308 |
+
"g_norm": 3.273187030572444e-05,
|
| 1309 |
+
"shape": [
|
| 1310 |
+
384
|
| 1311 |
+
]
|
| 1312 |
+
},
|
| 1313 |
+
"dec_blocks.3.norm2.bias": {
|
| 1314 |
+
"w_norm": 0.18299466371536255,
|
| 1315 |
+
"g_norm": 3.3433821954531595e-05,
|
| 1316 |
+
"shape": [
|
| 1317 |
+
384
|
| 1318 |
+
]
|
| 1319 |
+
},
|
| 1320 |
+
"dec_blocks.3.conv2.weight": {
|
| 1321 |
+
"w_norm": 11.477055549621582,
|
| 1322 |
+
"g_norm": 0.001047117984853685,
|
| 1323 |
+
"shape": [
|
| 1324 |
+
384,
|
| 1325 |
+
384,
|
| 1326 |
+
3,
|
| 1327 |
+
3
|
| 1328 |
+
]
|
| 1329 |
+
},
|
| 1330 |
+
"dec_blocks.3.conv2.bias": {
|
| 1331 |
+
"w_norm": 0.19348949193954468,
|
| 1332 |
+
"g_norm": 4.914805685984902e-05,
|
| 1333 |
+
"shape": [
|
| 1334 |
+
384
|
| 1335 |
+
]
|
| 1336 |
+
},
|
| 1337 |
+
"dec_blocks.3.skip.weight": {
|
| 1338 |
+
"w_norm": 11.101613998413086,
|
| 1339 |
+
"g_norm": 0.0015457079280167818,
|
| 1340 |
+
"shape": [
|
| 1341 |
+
384,
|
| 1342 |
+
768,
|
| 1343 |
+
1,
|
| 1344 |
+
1
|
| 1345 |
+
]
|
| 1346 |
+
},
|
| 1347 |
+
"dec_blocks.3.skip.bias": {
|
| 1348 |
+
"w_norm": 0.40536269545555115,
|
| 1349 |
+
"g_norm": 4.914805685984902e-05,
|
| 1350 |
+
"shape": [
|
| 1351 |
+
384
|
| 1352 |
+
]
|
| 1353 |
+
},
|
| 1354 |
+
"dec_blocks.4.norm1.weight": {
|
| 1355 |
+
"w_norm": 23.982229232788086,
|
| 1356 |
+
"g_norm": 7.216406083898619e-05,
|
| 1357 |
+
"shape": [
|
| 1358 |
+
576
|
| 1359 |
+
]
|
| 1360 |
+
},
|
| 1361 |
+
"dec_blocks.4.norm1.bias": {
|
| 1362 |
+
"w_norm": 0.07750255614519119,
|
| 1363 |
+
"g_norm": 4.411262125358917e-05,
|
| 1364 |
+
"shape": [
|
| 1365 |
+
576
|
| 1366 |
+
]
|
| 1367 |
+
},
|
| 1368 |
+
"dec_blocks.4.conv1.weight": {
|
| 1369 |
+
"w_norm": 8.488591194152832,
|
| 1370 |
+
"g_norm": 0.0018581050680950284,
|
| 1371 |
+
"shape": [
|
| 1372 |
+
192,
|
| 1373 |
+
576,
|
| 1374 |
+
3,
|
| 1375 |
+
3
|
| 1376 |
+
]
|
| 1377 |
+
},
|
| 1378 |
+
"dec_blocks.4.conv1.bias": {
|
| 1379 |
+
"w_norm": 0.11655846983194351,
|
| 1380 |
+
"g_norm": 4.4020147470291704e-05,
|
| 1381 |
+
"shape": [
|
| 1382 |
+
192
|
| 1383 |
+
]
|
| 1384 |
+
},
|
| 1385 |
+
"dec_blocks.4.t_proj.weight": {
|
| 1386 |
+
"w_norm": 8.032074928283691,
|
| 1387 |
+
"g_norm": 4.8955851525533944e-05,
|
| 1388 |
+
"shape": [
|
| 1389 |
+
192,
|
| 1390 |
+
256
|
| 1391 |
+
]
|
| 1392 |
+
},
|
| 1393 |
+
"dec_blocks.4.t_proj.bias": {
|
| 1394 |
+
"w_norm": 0.48134493827819824,
|
| 1395 |
+
"g_norm": 4.4041335058864206e-05,
|
| 1396 |
+
"shape": [
|
| 1397 |
+
192
|
| 1398 |
+
]
|
| 1399 |
+
},
|
| 1400 |
+
"dec_blocks.4.norm2.weight": {
|
| 1401 |
+
"w_norm": 13.797520637512207,
|
| 1402 |
+
"g_norm": 0.00010544883843977004,
|
| 1403 |
+
"shape": [
|
| 1404 |
+
192
|
| 1405 |
+
]
|
| 1406 |
+
},
|
| 1407 |
+
"dec_blocks.4.norm2.bias": {
|
| 1408 |
+
"w_norm": 0.06829909980297089,
|
| 1409 |
+
"g_norm": 8.821025403449312e-05,
|
| 1410 |
+
"shape": [
|
| 1411 |
+
192
|
| 1412 |
+
]
|
| 1413 |
+
},
|
| 1414 |
+
"dec_blocks.4.conv2.weight": {
|
| 1415 |
+
"w_norm": 8.082486152648926,
|
| 1416 |
+
"g_norm": 0.0036709108389914036,
|
| 1417 |
+
"shape": [
|
| 1418 |
+
192,
|
| 1419 |
+
192,
|
| 1420 |
+
3,
|
| 1421 |
+
3
|
| 1422 |
+
]
|
| 1423 |
+
},
|
| 1424 |
+
"dec_blocks.4.conv2.bias": {
|
| 1425 |
+
"w_norm": 0.19469857215881348,
|
| 1426 |
+
"g_norm": 0.00015675797476433218,
|
| 1427 |
+
"shape": [
|
| 1428 |
+
192
|
| 1429 |
+
]
|
| 1430 |
+
},
|
| 1431 |
+
"dec_blocks.4.skip.weight": {
|
| 1432 |
+
"w_norm": 7.862843990325928,
|
| 1433 |
+
"g_norm": 0.006234226748347282,
|
| 1434 |
+
"shape": [
|
| 1435 |
+
192,
|
| 1436 |
+
576,
|
| 1437 |
+
1,
|
| 1438 |
+
1
|
| 1439 |
+
]
|
| 1440 |
+
},
|
| 1441 |
+
"dec_blocks.4.skip.bias": {
|
| 1442 |
+
"w_norm": 0.3469906747341156,
|
| 1443 |
+
"g_norm": 0.00015675797476433218,
|
| 1444 |
+
"shape": [
|
| 1445 |
+
192
|
| 1446 |
+
]
|
| 1447 |
+
},
|
| 1448 |
+
"dec_blocks.5.norm1.weight": {
|
| 1449 |
+
"w_norm": 19.546398162841797,
|
| 1450 |
+
"g_norm": 0.0015075217233970761,
|
| 1451 |
+
"shape": [
|
| 1452 |
+
384
|
| 1453 |
+
]
|
| 1454 |
+
},
|
| 1455 |
+
"dec_blocks.5.norm1.bias": {
|
| 1456 |
+
"w_norm": 0.10585547238588333,
|
| 1457 |
+
"g_norm": 0.001897464506328106,
|
| 1458 |
+
"shape": [
|
| 1459 |
+
384
|
| 1460 |
+
]
|
| 1461 |
+
},
|
| 1462 |
+
"dec_blocks.5.conv1.weight": {
|
| 1463 |
+
"w_norm": 8.54923152923584,
|
| 1464 |
+
"g_norm": 0.02115660533308983,
|
| 1465 |
+
"shape": [
|
| 1466 |
+
192,
|
| 1467 |
+
384,
|
| 1468 |
+
3,
|
| 1469 |
+
3
|
| 1470 |
+
]
|
| 1471 |
+
},
|
| 1472 |
+
"dec_blocks.5.conv1.bias": {
|
| 1473 |
+
"w_norm": 0.13828550279140472,
|
| 1474 |
+
"g_norm": 0.000887448841240257,
|
| 1475 |
+
"shape": [
|
| 1476 |
+
192
|
| 1477 |
+
]
|
| 1478 |
+
},
|
| 1479 |
+
"dec_blocks.5.t_proj.weight": {
|
| 1480 |
+
"w_norm": 8.180492401123047,
|
| 1481 |
+
"g_norm": 0.0015367768937721848,
|
| 1482 |
+
"shape": [
|
| 1483 |
+
192,
|
| 1484 |
+
256
|
| 1485 |
+
]
|
| 1486 |
+
},
|
| 1487 |
+
"dec_blocks.5.t_proj.bias": {
|
| 1488 |
+
"w_norm": 0.5051036477088928,
|
| 1489 |
+
"g_norm": 0.0008874782361090183,
|
| 1490 |
+
"shape": [
|
| 1491 |
+
192
|
| 1492 |
+
]
|
| 1493 |
+
},
|
| 1494 |
+
"dec_blocks.5.norm2.weight": {
|
| 1495 |
+
"w_norm": 13.89153003692627,
|
| 1496 |
+
"g_norm": 0.001414273981936276,
|
| 1497 |
+
"shape": [
|
| 1498 |
+
192
|
| 1499 |
+
]
|
| 1500 |
+
},
|
| 1501 |
+
"dec_blocks.5.norm2.bias": {
|
| 1502 |
+
"w_norm": 0.04068835452198982,
|
| 1503 |
+
"g_norm": 0.0013583751861006021,
|
| 1504 |
+
"shape": [
|
| 1505 |
+
192
|
| 1506 |
+
]
|
| 1507 |
+
},
|
| 1508 |
+
"dec_blocks.5.conv2.weight": {
|
| 1509 |
+
"w_norm": 8.35645580291748,
|
| 1510 |
+
"g_norm": 0.018619170412421227,
|
| 1511 |
+
"shape": [
|
| 1512 |
+
192,
|
| 1513 |
+
192,
|
| 1514 |
+
3,
|
| 1515 |
+
3
|
| 1516 |
+
]
|
| 1517 |
+
},
|
| 1518 |
+
"dec_blocks.5.conv2.bias": {
|
| 1519 |
+
"w_norm": 0.20079921185970306,
|
| 1520 |
+
"g_norm": 0.0008557940018363297,
|
| 1521 |
+
"shape": [
|
| 1522 |
+
192
|
| 1523 |
+
]
|
| 1524 |
+
},
|
| 1525 |
+
"dec_blocks.5.skip.weight": {
|
| 1526 |
+
"w_norm": 8.094463348388672,
|
| 1527 |
+
"g_norm": 0.024322308599948883,
|
| 1528 |
+
"shape": [
|
| 1529 |
+
192,
|
| 1530 |
+
384,
|
| 1531 |
+
1,
|
| 1532 |
+
1
|
| 1533 |
+
]
|
| 1534 |
+
},
|
| 1535 |
+
"dec_blocks.5.skip.bias": {
|
| 1536 |
+
"w_norm": 0.42241600155830383,
|
| 1537 |
+
"g_norm": 0.0008557940018363297,
|
| 1538 |
+
"shape": [
|
| 1539 |
+
192
|
| 1540 |
+
]
|
| 1541 |
+
},
|
| 1542 |
+
"dec_attns.0.norm.weight": {
|
| 1543 |
+
"w_norm": 27.633703231811523,
|
| 1544 |
+
"g_norm": 2.9732186703768093e-06,
|
| 1545 |
+
"shape": [
|
| 1546 |
+
768
|
| 1547 |
+
]
|
| 1548 |
+
},
|
| 1549 |
+
"dec_attns.0.norm.bias": {
|
| 1550 |
+
"w_norm": 0.03998072072863579,
|
| 1551 |
+
"g_norm": 3.1545473575533833e-06,
|
| 1552 |
+
"shape": [
|
| 1553 |
+
768
|
| 1554 |
+
]
|
| 1555 |
+
},
|
| 1556 |
+
"dec_attns.0.qkv.weight": {
|
| 1557 |
+
"w_norm": 27.978946685791016,
|
| 1558 |
+
"g_norm": 8.574360981583595e-05,
|
| 1559 |
+
"shape": [
|
| 1560 |
+
2304,
|
| 1561 |
+
768,
|
| 1562 |
+
1,
|
| 1563 |
+
1
|
| 1564 |
+
]
|
| 1565 |
+
},
|
| 1566 |
+
"dec_attns.0.qkv.bias": {
|
| 1567 |
+
"w_norm": 1.0076439380645752,
|
| 1568 |
+
"g_norm": 3.600303898565471e-06,
|
| 1569 |
+
"shape": [
|
| 1570 |
+
2304
|
| 1571 |
+
]
|
| 1572 |
+
},
|
| 1573 |
+
"dec_attns.0.proj.weight": {
|
| 1574 |
+
"w_norm": 16.07609748840332,
|
| 1575 |
+
"g_norm": 8.897064253687859e-05,
|
| 1576 |
+
"shape": [
|
| 1577 |
+
768,
|
| 1578 |
+
768,
|
| 1579 |
+
1,
|
| 1580 |
+
1
|
| 1581 |
+
]
|
| 1582 |
+
},
|
| 1583 |
+
"dec_attns.0.proj.bias": {
|
| 1584 |
+
"w_norm": 0.5759825706481934,
|
| 1585 |
+
"g_norm": 4.993305083189625e-06,
|
| 1586 |
+
"shape": [
|
| 1587 |
+
768
|
| 1588 |
+
]
|
| 1589 |
+
},
|
| 1590 |
+
"dec_attns.1.norm.weight": {
|
| 1591 |
+
"w_norm": 27.627914428710938,
|
| 1592 |
+
"g_norm": 5.754376161348773e-06,
|
| 1593 |
+
"shape": [
|
| 1594 |
+
768
|
| 1595 |
+
]
|
| 1596 |
+
},
|
| 1597 |
+
"dec_attns.1.norm.bias": {
|
| 1598 |
+
"w_norm": 0.04270288348197937,
|
| 1599 |
+
"g_norm": 6.391747774614487e-06,
|
| 1600 |
+
"shape": [
|
| 1601 |
+
768
|
| 1602 |
+
]
|
| 1603 |
+
},
|
| 1604 |
+
"dec_attns.1.qkv.weight": {
|
| 1605 |
+
"w_norm": 27.995990753173828,
|
| 1606 |
+
"g_norm": 0.00017453204782214016,
|
| 1607 |
+
"shape": [
|
| 1608 |
+
2304,
|
| 1609 |
+
768,
|
| 1610 |
+
1,
|
| 1611 |
+
1
|
| 1612 |
+
]
|
| 1613 |
+
},
|
| 1614 |
+
"dec_attns.1.qkv.bias": {
|
| 1615 |
+
"w_norm": 1.0046143531799316,
|
| 1616 |
+
"g_norm": 7.337386250583222e-06,
|
| 1617 |
+
"shape": [
|
| 1618 |
+
2304
|
| 1619 |
+
]
|
| 1620 |
+
},
|
| 1621 |
+
"dec_attns.1.proj.weight": {
|
| 1622 |
+
"w_norm": 16.055126190185547,
|
| 1623 |
+
"g_norm": 0.00015127500228118151,
|
| 1624 |
+
"shape": [
|
| 1625 |
+
768,
|
| 1626 |
+
768,
|
| 1627 |
+
1,
|
| 1628 |
+
1
|
| 1629 |
+
]
|
| 1630 |
+
},
|
| 1631 |
+
"dec_attns.1.proj.bias": {
|
| 1632 |
+
"w_norm": 0.5793938636779785,
|
| 1633 |
+
"g_norm": 1.0101544830831699e-05,
|
| 1634 |
+
"shape": [
|
| 1635 |
+
768
|
| 1636 |
+
]
|
| 1637 |
+
},
|
| 1638 |
+
"dec_attns.2.norm.weight": {
|
| 1639 |
+
"w_norm": 19.51289939880371,
|
| 1640 |
+
"g_norm": 7.795668352628127e-06,
|
| 1641 |
+
"shape": [
|
| 1642 |
+
384
|
| 1643 |
+
]
|
| 1644 |
+
},
|
| 1645 |
+
"dec_attns.2.norm.bias": {
|
| 1646 |
+
"w_norm": 0.025595588609576225,
|
| 1647 |
+
"g_norm": 1.0193104571953882e-05,
|
| 1648 |
+
"shape": [
|
| 1649 |
+
384
|
| 1650 |
+
]
|
| 1651 |
+
},
|
| 1652 |
+
"dec_attns.2.qkv.weight": {
|
| 1653 |
+
"w_norm": 19.6967830657959,
|
| 1654 |
+
"g_norm": 0.00019536222680471838,
|
| 1655 |
+
"shape": [
|
| 1656 |
+
1152,
|
| 1657 |
+
384,
|
| 1658 |
+
1,
|
| 1659 |
+
1
|
| 1660 |
+
]
|
| 1661 |
+
},
|
| 1662 |
+
"dec_attns.2.qkv.bias": {
|
| 1663 |
+
"w_norm": 0.9850842356681824,
|
| 1664 |
+
"g_norm": 1.2773794878739864e-05,
|
| 1665 |
+
"shape": [
|
| 1666 |
+
1152
|
| 1667 |
+
]
|
| 1668 |
+
},
|
| 1669 |
+
"dec_attns.2.proj.weight": {
|
| 1670 |
+
"w_norm": 11.304216384887695,
|
| 1671 |
+
"g_norm": 0.00019062748469877988,
|
| 1672 |
+
"shape": [
|
| 1673 |
+
384,
|
| 1674 |
+
384,
|
| 1675 |
+
1,
|
| 1676 |
+
1
|
| 1677 |
+
]
|
| 1678 |
+
},
|
| 1679 |
+
"dec_attns.2.proj.bias": {
|
| 1680 |
+
"w_norm": 0.5871286988258362,
|
| 1681 |
+
"g_norm": 1.9087274267803878e-05,
|
| 1682 |
+
"shape": [
|
| 1683 |
+
384
|
| 1684 |
+
]
|
| 1685 |
+
},
|
| 1686 |
+
"dec_ups.0.op.weight": {
|
| 1687 |
+
"w_norm": 16.16544532775879,
|
| 1688 |
+
"g_norm": 0.000586340669542551,
|
| 1689 |
+
"shape": [
|
| 1690 |
+
768,
|
| 1691 |
+
768,
|
| 1692 |
+
3,
|
| 1693 |
+
3
|
| 1694 |
+
]
|
| 1695 |
+
},
|
| 1696 |
+
"dec_ups.0.op.bias": {
|
| 1697 |
+
"w_norm": 0.2004891335964203,
|
| 1698 |
+
"g_norm": 4.288574018573854e-06,
|
| 1699 |
+
"shape": [
|
| 1700 |
+
768
|
| 1701 |
+
]
|
| 1702 |
+
},
|
| 1703 |
+
"dec_ups.1.op.weight": {
|
| 1704 |
+
"w_norm": 16.172229766845703,
|
| 1705 |
+
"g_norm": 0.0009024080936796963,
|
| 1706 |
+
"shape": [
|
| 1707 |
+
768,
|
| 1708 |
+
768,
|
| 1709 |
+
3,
|
| 1710 |
+
3
|
| 1711 |
+
]
|
| 1712 |
+
},
|
| 1713 |
+
"dec_ups.1.op.bias": {
|
| 1714 |
+
"w_norm": 0.19070285558700562,
|
| 1715 |
+
"g_norm": 8.297952263092156e-06,
|
| 1716 |
+
"shape": [
|
| 1717 |
+
768
|
| 1718 |
+
]
|
| 1719 |
+
},
|
| 1720 |
+
"dec_ups.2.op.weight": {
|
| 1721 |
+
"w_norm": 11.33031177520752,
|
| 1722 |
+
"g_norm": 0.0016408424125984311,
|
| 1723 |
+
"shape": [
|
| 1724 |
+
384,
|
| 1725 |
+
384,
|
| 1726 |
+
3,
|
| 1727 |
+
3
|
| 1728 |
+
]
|
| 1729 |
+
},
|
| 1730 |
+
"dec_ups.2.op.bias": {
|
| 1731 |
+
"w_norm": 0.1964583396911621,
|
| 1732 |
+
"g_norm": 2.5792473024921492e-05,
|
| 1733 |
+
"shape": [
|
| 1734 |
+
384
|
| 1735 |
+
]
|
| 1736 |
+
},
|
| 1737 |
+
"dec_ups.3.op.weight": {
|
| 1738 |
+
"w_norm": 11.363648414611816,
|
| 1739 |
+
"g_norm": 0.0027039572596549988,
|
| 1740 |
+
"shape": [
|
| 1741 |
+
384,
|
| 1742 |
+
384,
|
| 1743 |
+
3,
|
| 1744 |
+
3
|
| 1745 |
+
]
|
| 1746 |
+
},
|
| 1747 |
+
"dec_ups.3.op.bias": {
|
| 1748 |
+
"w_norm": 0.19799400866031647,
|
| 1749 |
+
"g_norm": 6.683699757559225e-05,
|
| 1750 |
+
"shape": [
|
| 1751 |
+
384
|
| 1752 |
+
]
|
| 1753 |
+
},
|
| 1754 |
+
"dec_ups.4.op.weight": {
|
| 1755 |
+
"w_norm": 8.007488250732422,
|
| 1756 |
+
"g_norm": 0.011867591179907322,
|
| 1757 |
+
"shape": [
|
| 1758 |
+
192,
|
| 1759 |
+
192,
|
| 1760 |
+
3,
|
| 1761 |
+
3
|
| 1762 |
+
]
|
| 1763 |
+
},
|
| 1764 |
+
"dec_ups.4.op.bias": {
|
| 1765 |
+
"w_norm": 0.19662714004516602,
|
| 1766 |
+
"g_norm": 0.00022351545339915901,
|
| 1767 |
+
"shape": [
|
| 1768 |
+
192
|
| 1769 |
+
]
|
| 1770 |
+
},
|
| 1771 |
+
"out_norm.weight": {
|
| 1772 |
+
"w_norm": 13.781282424926758,
|
| 1773 |
+
"g_norm": 0.005015091970562935,
|
| 1774 |
+
"shape": [
|
| 1775 |
+
192
|
| 1776 |
+
]
|
| 1777 |
+
},
|
| 1778 |
+
"out_norm.bias": {
|
| 1779 |
+
"w_norm": 0.09246864169836044,
|
| 1780 |
+
"g_norm": 0.005548957269638777,
|
| 1781 |
+
"shape": [
|
| 1782 |
+
192
|
| 1783 |
+
]
|
| 1784 |
+
},
|
| 1785 |
+
"out_conv.weight": {
|
| 1786 |
+
"w_norm": 0.1450980007648468,
|
| 1787 |
+
"g_norm": 0.3885398805141449,
|
| 1788 |
+
"shape": [
|
| 1789 |
+
3,
|
| 1790 |
+
192,
|
| 1791 |
+
3,
|
| 1792 |
+
3
|
| 1793 |
+
]
|
| 1794 |
+
},
|
| 1795 |
+
"out_conv.bias": {
|
| 1796 |
+
"w_norm": 0.008651298470795155,
|
| 1797 |
+
"g_norm": 0.023214390501379967,
|
| 1798 |
+
"shape": [
|
| 1799 |
+
3
|
| 1800 |
+
]
|
| 1801 |
+
}
|
| 1802 |
+
},
|
| 1803 |
+
"activations": {
|
| 1804 |
+
"input_conv": {
|
| 1805 |
+
"mean": 0.007205421570688486,
|
| 1806 |
+
"std": 0.4401821494102478,
|
| 1807 |
+
"abs_max": 3.21875,
|
| 1808 |
+
"frac_dead": 0.0004186829028185457,
|
| 1809 |
+
"frac_sat": 0.0
|
| 1810 |
+
},
|
| 1811 |
+
"time_mlp.1": {
|
| 1812 |
+
"mean": 0.017198296263813972,
|
| 1813 |
+
"std": 0.2722778022289276,
|
| 1814 |
+
"abs_max": 3.90625,
|
| 1815 |
+
"frac_dead": 0.0,
|
| 1816 |
+
"frac_sat": 0.0
|
| 1817 |
+
},
|
| 1818 |
+
"time_mlp.3": {
|
| 1819 |
+
"mean": 0.025112560018897057,
|
| 1820 |
+
"std": 0.20933035016059875,
|
| 1821 |
+
"abs_max": 0.98828125,
|
| 1822 |
+
"frac_dead": 0.00032552084303461015,
|
| 1823 |
+
"frac_sat": 0.0
|
| 1824 |
+
},
|
| 1825 |
+
"enc_blocks.0.norm1": {
|
| 1826 |
+
"mean": -0.005679808557033539,
|
| 1827 |
+
"std": 0.9958928823471069,
|
| 1828 |
+
"abs_max": 6.680795669555664,
|
| 1829 |
+
"frac_dead": 6.159146437312302e-07,
|
| 1830 |
+
"frac_sat": 0.0
|
| 1831 |
+
},
|
| 1832 |
+
"enc_blocks.0.conv1": {
|
| 1833 |
+
"mean": 0.08074763417243958,
|
| 1834 |
+
"std": 0.9407960772514343,
|
| 1835 |
+
"abs_max": 10.6875,
|
| 1836 |
+
"frac_dead": 4.441208511707373e-05,
|
| 1837 |
+
"frac_sat": 2.6490953430879927e-08
|
| 1838 |
+
},
|
| 1839 |
+
"enc_blocks.0.t_proj": {
|
| 1840 |
+
"mean": -0.0006273089675232768,
|
| 1841 |
+
"std": 0.21317224204540253,
|
| 1842 |
+
"abs_max": 2.125,
|
| 1843 |
+
"frac_dead": 0.0,
|
| 1844 |
+
"frac_sat": 0.0
|
| 1845 |
+
},
|
| 1846 |
+
"enc_blocks.0.norm2": {
|
| 1847 |
+
"mean": -0.004103214479982853,
|
| 1848 |
+
"std": 0.9992560148239136,
|
| 1849 |
+
"abs_max": 9.351706504821777,
|
| 1850 |
+
"frac_dead": 8.675787057654816e-07,
|
| 1851 |
+
"frac_sat": 0.0
|
| 1852 |
+
},
|
| 1853 |
+
"enc_blocks.0.conv2": {
|
| 1854 |
+
"mean": -0.06078268960118294,
|
| 1855 |
+
"std": 0.9260948896408081,
|
| 1856 |
+
"abs_max": 11.4375,
|
| 1857 |
+
"frac_dead": 3.555748116923496e-05,
|
| 1858 |
+
"frac_sat": 8.278423280216884e-08
|
| 1859 |
+
},
|
| 1860 |
+
"enc_downs.0.op": {
|
| 1861 |
+
"mean": 0.02630145289003849,
|
| 1862 |
+
"std": 1.916042447090149,
|
| 1863 |
+
"abs_max": 22.75,
|
| 1864 |
+
"frac_dead": 2.0609961211448535e-05,
|
| 1865 |
+
"frac_sat": 0.0005958874826319516
|
| 1866 |
+
},
|
| 1867 |
+
"enc_blocks.1.norm1": {
|
| 1868 |
+
"mean": -0.0018810041947290301,
|
| 1869 |
+
"std": 1.0003385543823242,
|
| 1870 |
+
"abs_max": 10.622716903686523,
|
| 1871 |
+
"frac_dead": 1.0596381798677612e-06,
|
| 1872 |
+
"frac_sat": 2.6490953430879927e-08
|
| 1873 |
+
},
|
| 1874 |
+
"enc_blocks.1.conv1": {
|
| 1875 |
+
"mean": -0.0316561721265316,
|
| 1876 |
+
"std": 0.93098384141922,
|
| 1877 |
+
"abs_max": 11.8125,
|
| 1878 |
+
"frac_dead": 3.63720791938249e-05,
|
| 1879 |
+
"frac_sat": 2.649095449669403e-07
|
| 1880 |
+
},
|
| 1881 |
+
"enc_blocks.1.t_proj": {
|
| 1882 |
+
"mean": 0.004489074926823378,
|
| 1883 |
+
"std": 0.19143736362457275,
|
| 1884 |
+
"abs_max": 1.515625,
|
| 1885 |
+
"frac_dead": 0.0,
|
| 1886 |
+
"frac_sat": 0.0
|
| 1887 |
+
},
|
| 1888 |
+
"enc_blocks.1.norm2": {
|
| 1889 |
+
"mean": -0.004230222199112177,
|
| 1890 |
+
"std": 0.9972831010818481,
|
| 1891 |
+
"abs_max": 12.254587173461914,
|
| 1892 |
+
"frac_dead": 1.0066562481370056e-06,
|
| 1893 |
+
"frac_sat": 1.1920928955078125e-07
|
| 1894 |
+
},
|
| 1895 |
+
"enc_blocks.1.conv2": {
|
| 1896 |
+
"mean": -0.01853681169450283,
|
| 1897 |
+
"std": 0.78462815284729,
|
| 1898 |
+
"abs_max": 9.9375,
|
| 1899 |
+
"frac_dead": 4.1166942537529394e-05,
|
| 1900 |
+
"frac_sat": 0.0
|
| 1901 |
+
},
|
| 1902 |
+
"enc_downs.1.op": {
|
| 1903 |
+
"mean": -0.061043038964271545,
|
| 1904 |
+
"std": 2.2881019115448,
|
| 1905 |
+
"abs_max": 19.125,
|
| 1906 |
+
"frac_dead": 1.5523699403274804e-05,
|
| 1907 |
+
"frac_sat": 0.0006492932443507016
|
| 1908 |
+
},
|
| 1909 |
+
"enc_blocks.2.norm1": {
|
| 1910 |
+
"mean": -0.0027286489494144917,
|
| 1911 |
+
"std": 0.998390793800354,
|
| 1912 |
+
"abs_max": 8.235820770263672,
|
| 1913 |
+
"frac_dead": 6.887647714393097e-07,
|
| 1914 |
+
"frac_sat": 0.0
|
| 1915 |
+
},
|
| 1916 |
+
"enc_blocks.2.conv1": {
|
| 1917 |
+
"mean": -0.08958557993173599,
|
| 1918 |
+
"std": 1.2522926330566406,
|
| 1919 |
+
"abs_max": 18.125,
|
| 1920 |
+
"frac_dead": 3.594822555896826e-05,
|
| 1921 |
+
"frac_sat": 0.00014895862841513008
|
| 1922 |
+
},
|
| 1923 |
+
"enc_blocks.2.t_proj": {
|
| 1924 |
+
"mean": -0.004121434409171343,
|
| 1925 |
+
"std": 0.2775437831878662,
|
| 1926 |
+
"abs_max": 2.515625,
|
| 1927 |
+
"frac_dead": 0.0,
|
| 1928 |
+
"frac_sat": 0.0
|
| 1929 |
+
},
|
| 1930 |
+
"enc_blocks.2.norm2": {
|
| 1931 |
+
"mean": -0.009121330454945564,
|
| 1932 |
+
"std": 0.9945322275161743,
|
| 1933 |
+
"abs_max": 13.316609382629395,
|
| 1934 |
+
"frac_dead": 1.1656019296424347e-06,
|
| 1935 |
+
"frac_sat": 3.708733402163489e-06
|
| 1936 |
+
},
|
| 1937 |
+
"enc_blocks.2.conv2": {
|
| 1938 |
+
"mean": -0.0012582726776599884,
|
| 1939 |
+
"std": 0.8416869640350342,
|
| 1940 |
+
"abs_max": 8.625,
|
| 1941 |
+
"frac_dead": 2.6332007109886035e-05,
|
| 1942 |
+
"frac_sat": 0.0
|
| 1943 |
+
},
|
| 1944 |
+
"enc_blocks.2.skip": {
|
| 1945 |
+
"mean": 0.03872901201248169,
|
| 1946 |
+
"std": 1.1367685794830322,
|
| 1947 |
+
"abs_max": 14.8125,
|
| 1948 |
+
"frac_dead": 9.706285345600918e-05,
|
| 1949 |
+
"frac_sat": 7.70886708778562e-06
|
| 1950 |
+
},
|
| 1951 |
+
"enc_downs.2.op": {
|
| 1952 |
+
"mean": 0.11237926036119461,
|
| 1953 |
+
"std": 1.6770859956741333,
|
| 1954 |
+
"abs_max": 18.0,
|
| 1955 |
+
"frac_dead": 1.229180270456709e-05,
|
| 1956 |
+
"frac_sat": 8.911556506063789e-05
|
| 1957 |
+
},
|
| 1958 |
+
"enc_blocks.3.norm1": {
|
| 1959 |
+
"mean": -0.00042686983942985535,
|
| 1960 |
+
"std": 1.0004702806472778,
|
| 1961 |
+
"abs_max": 7.873823165893555,
|
| 1962 |
+
"frac_dead": 5.298190899338806e-07,
|
| 1963 |
+
"frac_sat": 0.0
|
| 1964 |
+
},
|
| 1965 |
+
"enc_blocks.3.conv1": {
|
| 1966 |
+
"mean": -0.19867272675037384,
|
| 1967 |
+
"std": 1.476852297782898,
|
| 1968 |
+
"abs_max": 18.0,
|
| 1969 |
+
"frac_dead": 2.076890814350918e-05,
|
| 1970 |
+
"frac_sat": 0.00023947822046466172
|
| 1971 |
+
},
|
| 1972 |
+
"enc_blocks.3.t_proj": {
|
| 1973 |
+
"mean": 0.002406445099040866,
|
| 1974 |
+
"std": 0.13080888986587524,
|
| 1975 |
+
"abs_max": 1.0390625,
|
| 1976 |
+
"frac_dead": 0.0,
|
| 1977 |
+
"frac_sat": 0.0
|
| 1978 |
+
},
|
| 1979 |
+
"enc_blocks.3.norm2": {
|
| 1980 |
+
"mean": -0.008266602642834187,
|
| 1981 |
+
"std": 0.9931374788284302,
|
| 1982 |
+
"abs_max": 11.833065032958984,
|
| 1983 |
+
"frac_dead": 1.271565793103946e-06,
|
| 1984 |
+
"frac_sat": 3.814697265625e-06
|
| 1985 |
+
},
|
| 1986 |
+
"enc_blocks.3.conv2": {
|
| 1987 |
+
"mean": 0.04491338133811951,
|
| 1988 |
+
"std": 0.7213199138641357,
|
| 1989 |
+
"abs_max": 6.40625,
|
| 1990 |
+
"frac_dead": 3.041161471628584e-05,
|
| 1991 |
+
"frac_sat": 0.0
|
| 1992 |
+
},
|
| 1993 |
+
"enc_attns.3.norm": {
|
| 1994 |
+
"mean": 0.0003345790901221335,
|
| 1995 |
+
"std": 0.997136116027832,
|
| 1996 |
+
"abs_max": 8.509593963623047,
|
| 1997 |
+
"frac_dead": 1.6954210195763153e-06,
|
| 1998 |
+
"frac_sat": 0.0
|
| 1999 |
+
},
|
| 2000 |
+
"enc_attns.3.qkv": {
|
| 2001 |
+
"mean": 0.0020568061154335737,
|
| 2002 |
+
"std": 0.606049656867981,
|
| 2003 |
+
"abs_max": 5.75,
|
| 2004 |
+
"frac_dead": 0.00010850694525288418,
|
| 2005 |
+
"frac_sat": 0.0
|
| 2006 |
+
},
|
| 2007 |
+
"enc_attns.3.proj": {
|
| 2008 |
+
"mean": -0.004439083393663168,
|
| 2009 |
+
"std": 0.24444590508937836,
|
| 2010 |
+
"abs_max": 1.0546875,
|
| 2011 |
+
"frac_dead": 0.0002475314831826836,
|
| 2012 |
+
"frac_sat": 0.0
|
| 2013 |
+
},
|
| 2014 |
+
"enc_downs.3.op": {
|
| 2015 |
+
"mean": -0.11371082812547684,
|
| 2016 |
+
"std": 2.028658151626587,
|
| 2017 |
+
"abs_max": 20.875,
|
| 2018 |
+
"frac_dead": 1.2715657248918433e-05,
|
| 2019 |
+
"frac_sat": 0.0005972120561636984
|
| 2020 |
+
},
|
| 2021 |
+
"enc_blocks.4.norm1": {
|
| 2022 |
+
"mean": 0.00025804672623053193,
|
| 2023 |
+
"std": 1.001221776008606,
|
| 2024 |
+
"abs_max": 7.8459954261779785,
|
| 2025 |
+
"frac_dead": 2.1192763597355224e-06,
|
| 2026 |
+
"frac_sat": 0.0
|
| 2027 |
+
},
|
| 2028 |
+
"enc_blocks.4.conv1": {
|
| 2029 |
+
"mean": -0.3934686481952667,
|
| 2030 |
+
"std": 1.384687900543213,
|
| 2031 |
+
"abs_max": 17.25,
|
| 2032 |
+
"frac_dead": 2.3100112230167724e-05,
|
| 2033 |
+
"frac_sat": 0.00015407138562295586
|
| 2034 |
+
},
|
| 2035 |
+
"enc_blocks.4.t_proj": {
|
| 2036 |
+
"mean": -0.002096842974424362,
|
| 2037 |
+
"std": 0.09800918400287628,
|
| 2038 |
+
"abs_max": 0.6796875,
|
| 2039 |
+
"frac_dead": 0.0,
|
| 2040 |
+
"frac_sat": 0.0
|
| 2041 |
+
},
|
| 2042 |
+
"enc_blocks.4.norm2": {
|
| 2043 |
+
"mean": -0.008577298372983932,
|
| 2044 |
+
"std": 0.9937466382980347,
|
| 2045 |
+
"abs_max": 10.411909103393555,
|
| 2046 |
+
"frac_dead": 2.119276274470394e-07,
|
| 2047 |
+
"frac_sat": 8.477105097881577e-07
|
| 2048 |
+
},
|
| 2049 |
+
"enc_blocks.4.conv2": {
|
| 2050 |
+
"mean": -0.012939635664224625,
|
| 2051 |
+
"std": 0.9054562449455261,
|
| 2052 |
+
"abs_max": 5.40625,
|
| 2053 |
+
"frac_dead": 1.5894571333774365e-05,
|
| 2054 |
+
"frac_sat": 0.0
|
| 2055 |
+
},
|
| 2056 |
+
"enc_blocks.4.skip": {
|
| 2057 |
+
"mean": -0.05520473048090935,
|
| 2058 |
+
"std": 1.2102307081222534,
|
| 2059 |
+
"abs_max": 12.3125,
|
| 2060 |
+
"frac_dead": 6.37902194284834e-05,
|
| 2061 |
+
"frac_sat": 6.781684078305261e-06
|
| 2062 |
+
},
|
| 2063 |
+
"enc_attns.4.norm": {
|
| 2064 |
+
"mean": -1.8503931642044336e-05,
|
| 2065 |
+
"std": 0.9971495866775513,
|
| 2066 |
+
"abs_max": 7.199902057647705,
|
| 2067 |
+
"frac_dead": 1.0596381798677612e-06,
|
| 2068 |
+
"frac_sat": 0.0
|
| 2069 |
+
},
|
| 2070 |
+
"enc_attns.4.qkv": {
|
| 2071 |
+
"mean": -0.004289580043405294,
|
| 2072 |
+
"std": 0.7320247292518616,
|
| 2073 |
+
"abs_max": 8.125,
|
| 2074 |
+
"frac_dead": 6.859390850877389e-05,
|
| 2075 |
+
"frac_sat": 0.0
|
| 2076 |
+
},
|
| 2077 |
+
"enc_attns.4.proj": {
|
| 2078 |
+
"mean": 0.002922792686149478,
|
| 2079 |
+
"std": 0.3093913197517395,
|
| 2080 |
+
"abs_max": 2.65625,
|
| 2081 |
+
"frac_dead": 0.0001275804388569668,
|
| 2082 |
+
"frac_sat": 0.0
|
| 2083 |
+
},
|
| 2084 |
+
"enc_downs.4.op": {
|
| 2085 |
+
"mean": 0.02217293158173561,
|
| 2086 |
+
"std": 3.0861616134643555,
|
| 2087 |
+
"abs_max": 25.875,
|
| 2088 |
+
"frac_dead": 2.543131586207892e-06,
|
| 2089 |
+
"frac_sat": 0.007915073074400425
|
| 2090 |
+
},
|
| 2091 |
+
"enc_blocks.5.norm1": {
|
| 2092 |
+
"mean": -8.908328163670376e-05,
|
| 2093 |
+
"std": 1.000154733657837,
|
| 2094 |
+
"abs_max": 7.034839153289795,
|
| 2095 |
+
"frac_dead": 1.6954210195763153e-06,
|
| 2096 |
+
"frac_sat": 0.0
|
| 2097 |
+
},
|
| 2098 |
+
"enc_blocks.5.conv1": {
|
| 2099 |
+
"mean": -0.2771743834018707,
|
| 2100 |
+
"std": 2.046868324279785,
|
| 2101 |
+
"abs_max": 19.25,
|
| 2102 |
+
"frac_dead": 1.3563368156610522e-05,
|
| 2103 |
+
"frac_sat": 0.0013817681465297937
|
| 2104 |
+
},
|
| 2105 |
+
"enc_blocks.5.t_proj": {
|
| 2106 |
+
"mean": -0.0009553485433571041,
|
| 2107 |
+
"std": 0.08868180215358734,
|
| 2108 |
+
"abs_max": 0.62890625,
|
| 2109 |
+
"frac_dead": 0.0,
|
| 2110 |
+
"frac_sat": 0.0
|
| 2111 |
+
},
|
| 2112 |
+
"enc_blocks.5.norm2": {
|
| 2113 |
+
"mean": -0.005851236637681723,
|
| 2114 |
+
"std": 0.9946741461753845,
|
| 2115 |
+
"abs_max": 8.874484062194824,
|
| 2116 |
+
"frac_dead": 8.477105097881577e-07,
|
| 2117 |
+
"frac_sat": 0.0
|
| 2118 |
+
},
|
| 2119 |
+
"enc_blocks.5.conv2": {
|
| 2120 |
+
"mean": 0.0862385705113411,
|
| 2121 |
+
"std": 1.4481452703475952,
|
| 2122 |
+
"abs_max": 8.5625,
|
| 2123 |
+
"frac_dead": 1.3563368156610522e-05,
|
| 2124 |
+
"frac_sat": 0.0
|
| 2125 |
+
},
|
| 2126 |
+
"enc_attns.5.norm": {
|
| 2127 |
+
"mean": 5.221697938395664e-05,
|
| 2128 |
+
"std": 0.9983348846435547,
|
| 2129 |
+
"abs_max": 7.151754379272461,
|
| 2130 |
+
"frac_dead": 2.543131586207892e-06,
|
| 2131 |
+
"frac_sat": 0.0
|
| 2132 |
+
},
|
| 2133 |
+
"enc_attns.5.qkv": {
|
| 2134 |
+
"mean": -0.004897332284599543,
|
| 2135 |
+
"std": 0.8810105323791504,
|
| 2136 |
+
"abs_max": 8.375,
|
| 2137 |
+
"frac_dead": 6.357828533509746e-05,
|
| 2138 |
+
"frac_sat": 0.0
|
| 2139 |
+
},
|
| 2140 |
+
"enc_attns.5.proj": {
|
| 2141 |
+
"mean": -0.006027805618941784,
|
| 2142 |
+
"std": 0.19361162185668945,
|
| 2143 |
+
"abs_max": 1.53125,
|
| 2144 |
+
"frac_dead": 0.0002526177268009633,
|
| 2145 |
+
"frac_sat": 0.0
|
| 2146 |
+
},
|
| 2147 |
+
"mid1.norm1": {
|
| 2148 |
+
"mean": 0.00011206004273844883,
|
| 2149 |
+
"std": 1.0003716945648193,
|
| 2150 |
+
"abs_max": 7.116409778594971,
|
| 2151 |
+
"frac_dead": 8.477105097881577e-07,
|
| 2152 |
+
"frac_sat": 0.0
|
| 2153 |
+
},
|
| 2154 |
+
"mid1.conv1": {
|
| 2155 |
+
"mean": -0.7086418867111206,
|
| 2156 |
+
"std": 1.8661444187164307,
|
| 2157 |
+
"abs_max": 21.0,
|
| 2158 |
+
"frac_dead": 1.52587890625e-05,
|
| 2159 |
+
"frac_sat": 0.003971523605287075
|
| 2160 |
+
},
|
| 2161 |
+
"mid1.t_proj": {
|
| 2162 |
+
"mean": 0.002929791109636426,
|
| 2163 |
+
"std": 0.08738984912633896,
|
| 2164 |
+
"abs_max": 0.5078125,
|
| 2165 |
+
"frac_dead": 5.425347262644209e-05,
|
| 2166 |
+
"frac_sat": 0.0
|
| 2167 |
+
},
|
| 2168 |
+
"mid1.norm2": {
|
| 2169 |
+
"mean": -0.007256262004375458,
|
| 2170 |
+
"std": 0.9950929880142212,
|
| 2171 |
+
"abs_max": 10.638596534729004,
|
| 2172 |
+
"frac_dead": 8.477105097881577e-07,
|
| 2173 |
+
"frac_sat": 1.6954210195763153e-06
|
| 2174 |
+
},
|
| 2175 |
+
"mid1.conv2": {
|
| 2176 |
+
"mean": 0.03888840973377228,
|
| 2177 |
+
"std": 0.7718620300292969,
|
| 2178 |
+
"abs_max": 4.5,
|
| 2179 |
+
"frac_dead": 2.204047268605791e-05,
|
| 2180 |
+
"frac_sat": 0.0
|
| 2181 |
+
},
|
| 2182 |
+
"mid_attn.norm": {
|
| 2183 |
+
"mean": 0.00013342582678887993,
|
| 2184 |
+
"std": 0.9982437491416931,
|
| 2185 |
+
"abs_max": 6.748208045959473,
|
| 2186 |
+
"frac_dead": 8.477105097881577e-07,
|
| 2187 |
+
"frac_sat": 0.0
|
| 2188 |
+
},
|
| 2189 |
+
"mid_attn.qkv": {
|
| 2190 |
+
"mean": -0.010101129300892353,
|
| 2191 |
+
"std": 0.871902585029602,
|
| 2192 |
+
"abs_max": 7.5,
|
| 2193 |
+
"frac_dead": 5.8774596254806966e-05,
|
| 2194 |
+
"frac_sat": 0.0
|
| 2195 |
+
},
|
| 2196 |
+
"mid_attn.proj": {
|
| 2197 |
+
"mean": -0.004268778953701258,
|
| 2198 |
+
"std": 0.17432516813278198,
|
| 2199 |
+
"abs_max": 1.2734375,
|
| 2200 |
+
"frac_dead": 0.00027974447584711015,
|
| 2201 |
+
"frac_sat": 0.0
|
| 2202 |
+
},
|
| 2203 |
+
"mid2.norm1": {
|
| 2204 |
+
"mean": 0.0004497667250689119,
|
| 2205 |
+
"std": 1.0008673667907715,
|
| 2206 |
+
"abs_max": 6.641637325286865,
|
| 2207 |
+
"frac_dead": 8.477105097881577e-07,
|
| 2208 |
+
"frac_sat": 0.0
|
| 2209 |
+
},
|
| 2210 |
+
"mid2.conv1": {
|
| 2211 |
+
"mean": -0.9663907289505005,
|
| 2212 |
+
"std": 2.2991340160369873,
|
| 2213 |
+
"abs_max": 24.25,
|
| 2214 |
+
"frac_dead": 2.373589450144209e-05,
|
| 2215 |
+
"frac_sat": 0.01361083984375
|
| 2216 |
+
},
|
| 2217 |
+
"mid2.t_proj": {
|
| 2218 |
+
"mean": 0.0009166261297650635,
|
| 2219 |
+
"std": 0.0890422984957695,
|
| 2220 |
+
"abs_max": 0.6640625,
|
| 2221 |
+
"frac_dead": 0.0,
|
| 2222 |
+
"frac_sat": 0.0
|
| 2223 |
+
},
|
| 2224 |
+
"mid2.norm2": {
|
| 2225 |
+
"mean": -0.00821609515696764,
|
| 2226 |
+
"std": 0.9965092539787292,
|
| 2227 |
+
"abs_max": 10.094039916992188,
|
| 2228 |
+
"frac_dead": 8.477105097881577e-07,
|
| 2229 |
+
"frac_sat": 8.477105097881577e-07
|
| 2230 |
+
},
|
| 2231 |
+
"mid2.conv2": {
|
| 2232 |
+
"mean": 0.08653509616851807,
|
| 2233 |
+
"std": 1.1487525701522827,
|
| 2234 |
+
"abs_max": 5.75,
|
| 2235 |
+
"frac_dead": 1.9497341781971045e-05,
|
| 2236 |
+
"frac_sat": 0.0
|
| 2237 |
+
},
|
| 2238 |
+
"dec_blocks.0.norm1": {
|
| 2239 |
+
"mean": 0.0010238753166049719,
|
| 2240 |
+
"std": 1.0017040967941284,
|
| 2241 |
+
"abs_max": 7.182895660400391,
|
| 2242 |
+
"frac_dead": 8.477105097881577e-07,
|
| 2243 |
+
"frac_sat": 0.0
|
| 2244 |
+
},
|
| 2245 |
+
"dec_blocks.0.conv1": {
|
| 2246 |
+
"mean": -2.5499513149261475,
|
| 2247 |
+
"std": 5.234790802001953,
|
| 2248 |
+
"abs_max": 55.25,
|
| 2249 |
+
"frac_dead": 5.086263172415784e-06,
|
| 2250 |
+
"frac_sat": 0.0761905238032341
|
| 2251 |
+
},
|
| 2252 |
+
"dec_blocks.0.t_proj": {
|
| 2253 |
+
"mean": -0.0014294469729065895,
|
| 2254 |
+
"std": 0.09479585289955139,
|
| 2255 |
+
"abs_max": 0.60546875,
|
| 2256 |
+
"frac_dead": 0.0,
|
| 2257 |
+
"frac_sat": 0.0
|
| 2258 |
+
},
|
| 2259 |
+
"dec_blocks.0.norm2": {
|
| 2260 |
+
"mean": -0.010944444686174393,
|
| 2261 |
+
"std": 0.9966337084770203,
|
| 2262 |
+
"abs_max": 10.068451881408691,
|
| 2263 |
+
"frac_dead": 2.543131586207892e-06,
|
| 2264 |
+
"frac_sat": 1.6954210195763153e-06
|
| 2265 |
+
},
|
| 2266 |
+
"dec_blocks.0.conv2": {
|
| 2267 |
+
"mean": 0.011901033110916615,
|
| 2268 |
+
"std": 0.5798095464706421,
|
| 2269 |
+
"abs_max": 3.125,
|
| 2270 |
+
"frac_dead": 2.7126736313221045e-05,
|
| 2271 |
+
"frac_sat": 0.0
|
| 2272 |
+
},
|
| 2273 |
+
"dec_blocks.0.skip": {
|
| 2274 |
+
"mean": -0.044557470828294754,
|
| 2275 |
+
"std": 1.870935082435608,
|
| 2276 |
+
"abs_max": 14.6875,
|
| 2277 |
+
"frac_dead": 1.610649997019209e-05,
|
| 2278 |
+
"frac_sat": 0.00020429823780432343
|
| 2279 |
+
},
|
| 2280 |
+
"dec_attns.0.norm": {
|
| 2281 |
+
"mean": -3.814531737589277e-05,
|
| 2282 |
+
"std": 0.9969810247421265,
|
| 2283 |
+
"abs_max": 6.740102767944336,
|
| 2284 |
+
"frac_dead": 0.0,
|
| 2285 |
+
"frac_sat": 0.0
|
| 2286 |
+
},
|
| 2287 |
+
"dec_attns.0.qkv": {
|
| 2288 |
+
"mean": -0.007594096474349499,
|
| 2289 |
+
"std": 0.7095833420753479,
|
| 2290 |
+
"abs_max": 4.78125,
|
| 2291 |
+
"frac_dead": 7.262053259182721e-05,
|
| 2292 |
+
"frac_sat": 0.0
|
| 2293 |
+
},
|
| 2294 |
+
"dec_attns.0.proj": {
|
| 2295 |
+
"mean": 0.02804269827902317,
|
| 2296 |
+
"std": 0.650712788105011,
|
| 2297 |
+
"abs_max": 2.953125,
|
| 2298 |
+
"frac_dead": 6.442599988076836e-05,
|
| 2299 |
+
"frac_sat": 0.0
|
| 2300 |
+
},
|
| 2301 |
+
"dec_ups.0.op": {
|
| 2302 |
+
"mean": 0.01264216285198927,
|
| 2303 |
+
"std": 2.6158359050750732,
|
| 2304 |
+
"abs_max": 23.0,
|
| 2305 |
+
"frac_dead": 6.357828624459216e-06,
|
| 2306 |
+
"frac_sat": 0.001959271030500531
|
| 2307 |
+
},
|
| 2308 |
+
"dec_blocks.1.norm1": {
|
| 2309 |
+
"mean": 0.0003692789177875966,
|
| 2310 |
+
"std": 1.0006011724472046,
|
| 2311 |
+
"abs_max": 7.659249305725098,
|
| 2312 |
+
"frac_dead": 5.298190899338806e-07,
|
| 2313 |
+
"frac_sat": 0.0
|
| 2314 |
+
},
|
| 2315 |
+
"dec_blocks.1.conv1": {
|
| 2316 |
+
"mean": -2.1808557510375977,
|
| 2317 |
+
"std": 3.315422534942627,
|
| 2318 |
+
"abs_max": 48.0,
|
| 2319 |
+
"frac_dead": 5.510118171514478e-06,
|
| 2320 |
+
"frac_sat": 0.03862402215600014
|
| 2321 |
+
},
|
| 2322 |
+
"dec_blocks.1.t_proj": {
|
| 2323 |
+
"mean": -0.0016897112363949418,
|
| 2324 |
+
"std": 0.09164174646139145,
|
| 2325 |
+
"abs_max": 0.6328125,
|
| 2326 |
+
"frac_dead": 0.0,
|
| 2327 |
+
"frac_sat": 0.0
|
| 2328 |
+
},
|
| 2329 |
+
"dec_blocks.1.norm2": {
|
| 2330 |
+
"mean": -0.012301405891776085,
|
| 2331 |
+
"std": 0.9972518086433411,
|
| 2332 |
+
"abs_max": 11.372942924499512,
|
| 2333 |
+
"frac_dead": 6.35782896551973e-07,
|
| 2334 |
+
"frac_sat": 7.841322258173022e-06
|
| 2335 |
+
},
|
| 2336 |
+
"dec_blocks.1.conv2": {
|
| 2337 |
+
"mean": -0.0036889174953103065,
|
| 2338 |
+
"std": 0.41822341084480286,
|
| 2339 |
+
"abs_max": 2.515625,
|
| 2340 |
+
"frac_dead": 3.708733493112959e-05,
|
| 2341 |
+
"frac_sat": 0.0
|
| 2342 |
+
},
|
| 2343 |
+
"dec_blocks.1.skip": {
|
| 2344 |
+
"mean": -0.013397175818681717,
|
| 2345 |
+
"std": 1.399378776550293,
|
| 2346 |
+
"abs_max": 10.625,
|
| 2347 |
+
"frac_dead": 2.7974447220913135e-05,
|
| 2348 |
+
"frac_sat": 6.35782896551973e-07
|
| 2349 |
+
},
|
| 2350 |
+
"dec_attns.1.norm": {
|
| 2351 |
+
"mean": -2.5563769668224268e-05,
|
| 2352 |
+
"std": 0.996963381767273,
|
| 2353 |
+
"abs_max": 6.62148904800415,
|
| 2354 |
+
"frac_dead": 6.35782896551973e-07,
|
| 2355 |
+
"frac_sat": 0.0
|
| 2356 |
+
},
|
| 2357 |
+
"dec_attns.1.qkv": {
|
| 2358 |
+
"mean": 0.0024721473455429077,
|
| 2359 |
+
"std": 0.6335548758506775,
|
| 2360 |
+
"abs_max": 4.03125,
|
| 2361 |
+
"frac_dead": 7.466916576959193e-05,
|
| 2362 |
+
"frac_sat": 0.0
|
| 2363 |
+
},
|
| 2364 |
+
"dec_attns.1.proj": {
|
| 2365 |
+
"mean": -0.008538656868040562,
|
| 2366 |
+
"std": 0.36578020453453064,
|
| 2367 |
+
"abs_max": 1.3671875,
|
| 2368 |
+
"frac_dead": 0.00012969970703125,
|
| 2369 |
+
"frac_sat": 0.0
|
| 2370 |
+
},
|
| 2371 |
+
"dec_ups.1.op": {
|
| 2372 |
+
"mean": 0.045211486518383026,
|
| 2373 |
+
"std": 1.5917174816131592,
|
| 2374 |
+
"abs_max": 12.5,
|
| 2375 |
+
"frac_dead": 9.324815437139478e-06,
|
| 2376 |
+
"frac_sat": 1.88085778063396e-05
|
| 2377 |
+
},
|
| 2378 |
+
"dec_blocks.2.norm1": {
|
| 2379 |
+
"mean": -3.3550793887116015e-05,
|
| 2380 |
+
"std": 1.0003987550735474,
|
| 2381 |
+
"abs_max": 8.471176147460938,
|
| 2382 |
+
"frac_dead": 8.477105097881577e-07,
|
| 2383 |
+
"frac_sat": 0.0
|
| 2384 |
+
},
|
| 2385 |
+
"dec_blocks.2.conv1": {
|
| 2386 |
+
"mean": -1.0419565439224243,
|
| 2387 |
+
"std": 3.1819097995758057,
|
| 2388 |
+
"abs_max": 34.75,
|
| 2389 |
+
"frac_dead": 9.960598617908545e-06,
|
| 2390 |
+
"frac_sat": 0.03538068011403084
|
| 2391 |
+
},
|
| 2392 |
+
"dec_blocks.2.t_proj": {
|
| 2393 |
+
"mean": -0.0044543445110321045,
|
| 2394 |
+
"std": 0.10459078103303909,
|
| 2395 |
+
"abs_max": 0.62109375,
|
| 2396 |
+
"frac_dead": 0.0,
|
| 2397 |
+
"frac_sat": 0.0
|
| 2398 |
+
},
|
| 2399 |
+
"dec_blocks.2.norm2": {
|
| 2400 |
+
"mean": -0.011841188184916973,
|
| 2401 |
+
"std": 1.0002442598342896,
|
| 2402 |
+
"abs_max": 9.476700782775879,
|
| 2403 |
+
"frac_dead": 2.4371677227463806e-06,
|
| 2404 |
+
"frac_sat": 0.0
|
| 2405 |
+
},
|
| 2406 |
+
"dec_blocks.2.conv2": {
|
| 2407 |
+
"mean": -0.003868026891723275,
|
| 2408 |
+
"std": 0.3635369539260864,
|
| 2409 |
+
"abs_max": 2.078125,
|
| 2410 |
+
"frac_dead": 5.647871148539707e-05,
|
| 2411 |
+
"frac_sat": 0.0
|
| 2412 |
+
},
|
| 2413 |
+
"dec_blocks.2.skip": {
|
| 2414 |
+
"mean": 0.0852733850479126,
|
| 2415 |
+
"std": 1.2003114223480225,
|
| 2416 |
+
"abs_max": 9.1875,
|
| 2417 |
+
"frac_dead": 3.178914266754873e-05,
|
| 2418 |
+
"frac_sat": 0.0
|
| 2419 |
+
},
|
| 2420 |
+
"dec_attns.2.norm": {
|
| 2421 |
+
"mean": 7.900595664978027e-05,
|
| 2422 |
+
"std": 0.9957061409950256,
|
| 2423 |
+
"abs_max": 6.682980060577393,
|
| 2424 |
+
"frac_dead": 2.119276274470394e-07,
|
| 2425 |
+
"frac_sat": 0.0
|
| 2426 |
+
},
|
| 2427 |
+
"dec_attns.2.qkv": {
|
| 2428 |
+
"mean": 0.013404250144958496,
|
| 2429 |
+
"std": 0.6031450629234314,
|
| 2430 |
+
"abs_max": 4.625,
|
| 2431 |
+
"frac_dead": 9.716882050270215e-05,
|
| 2432 |
+
"frac_sat": 0.0
|
| 2433 |
+
},
|
| 2434 |
+
"dec_attns.2.proj": {
|
| 2435 |
+
"mean": -0.02518179453909397,
|
| 2436 |
+
"std": 0.3698141872882843,
|
| 2437 |
+
"abs_max": 1.265625,
|
| 2438 |
+
"frac_dead": 0.00016360812878701836,
|
| 2439 |
+
"frac_sat": 0.0
|
| 2440 |
+
},
|
| 2441 |
+
"dec_ups.2.op": {
|
| 2442 |
+
"mean": 0.03491879999637604,
|
| 2443 |
+
"std": 1.0012180805206299,
|
| 2444 |
+
"abs_max": 7.09375,
|
| 2445 |
+
"frac_dead": 2.278222018503584e-05,
|
| 2446 |
+
"frac_sat": 0.0
|
| 2447 |
+
},
|
| 2448 |
+
"dec_blocks.3.norm1": {
|
| 2449 |
+
"mean": -0.0005132730002515018,
|
| 2450 |
+
"std": 1.0001797676086426,
|
| 2451 |
+
"abs_max": 10.262784957885742,
|
| 2452 |
+
"frac_dead": 7.682376690354431e-07,
|
| 2453 |
+
"frac_sat": 1.3245476715439963e-08
|
| 2454 |
+
},
|
| 2455 |
+
"dec_blocks.3.conv1": {
|
| 2456 |
+
"mean": -0.8485596776008606,
|
| 2457 |
+
"std": 2.129643440246582,
|
| 2458 |
+
"abs_max": 34.5,
|
| 2459 |
+
"frac_dead": 1.5709136278019287e-05,
|
| 2460 |
+
"frac_sat": 0.007686244323849678
|
| 2461 |
+
},
|
| 2462 |
+
"dec_blocks.3.t_proj": {
|
| 2463 |
+
"mean": -0.00039477564860135317,
|
| 2464 |
+
"std": 0.12362557649612427,
|
| 2465 |
+
"abs_max": 0.7734375,
|
| 2466 |
+
"frac_dead": 0.0,
|
| 2467 |
+
"frac_sat": 0.0
|
| 2468 |
+
},
|
| 2469 |
+
"dec_blocks.3.norm2": {
|
| 2470 |
+
"mean": -0.011190549470484257,
|
| 2471 |
+
"std": 0.995294988155365,
|
| 2472 |
+
"abs_max": 12.03353500366211,
|
| 2473 |
+
"frac_dead": 2.278221927554114e-06,
|
| 2474 |
+
"frac_sat": 1.141760094469646e-05
|
| 2475 |
+
},
|
| 2476 |
+
"dec_blocks.3.conv2": {
|
| 2477 |
+
"mean": -0.0035135315265506506,
|
| 2478 |
+
"std": 0.3895057141780853,
|
| 2479 |
+
"abs_max": 5.3125,
|
| 2480 |
+
"frac_dead": 5.745887756347656e-05,
|
| 2481 |
+
"frac_sat": 0.0
|
| 2482 |
+
},
|
| 2483 |
+
"dec_blocks.3.skip": {
|
| 2484 |
+
"mean": 0.03638695552945137,
|
| 2485 |
+
"std": 0.7669461369514465,
|
| 2486 |
+
"abs_max": 7.1875,
|
| 2487 |
+
"frac_dead": 5.66906419408042e-05,
|
| 2488 |
+
"frac_sat": 0.0
|
| 2489 |
+
},
|
| 2490 |
+
"dec_ups.3.op": {
|
| 2491 |
+
"mean": -0.04504766687750816,
|
| 2492 |
+
"std": 0.8403801321983337,
|
| 2493 |
+
"abs_max": 8.75,
|
| 2494 |
+
"frac_dead": 2.6305517167202197e-05,
|
| 2495 |
+
"frac_sat": 0.0
|
| 2496 |
+
},
|
| 2497 |
+
"dec_blocks.4.norm1": {
|
| 2498 |
+
"mean": -0.0013438367750495672,
|
| 2499 |
+
"std": 1.000179409980774,
|
| 2500 |
+
"abs_max": 11.959559440612793,
|
| 2501 |
+
"frac_dead": 9.757501402418711e-07,
|
| 2502 |
+
"frac_sat": 1.3686992872408155e-07
|
| 2503 |
+
},
|
| 2504 |
+
"dec_blocks.4.conv1": {
|
| 2505 |
+
"mean": -0.5712231397628784,
|
| 2506 |
+
"std": 1.8479676246643066,
|
| 2507 |
+
"abs_max": 22.875,
|
| 2508 |
+
"frac_dead": 1.3576614037447143e-05,
|
| 2509 |
+
"frac_sat": 0.0011172691592946649
|
| 2510 |
+
},
|
| 2511 |
+
"dec_blocks.4.t_proj": {
|
| 2512 |
+
"mean": -0.004008995834738016,
|
| 2513 |
+
"std": 0.12154782563447952,
|
| 2514 |
+
"abs_max": 0.6875,
|
| 2515 |
+
"frac_dead": 0.0,
|
| 2516 |
+
"frac_sat": 0.0
|
| 2517 |
+
},
|
| 2518 |
+
"dec_blocks.4.norm2": {
|
| 2519 |
+
"mean": -0.005222403444349766,
|
| 2520 |
+
"std": 0.996995210647583,
|
| 2521 |
+
"abs_max": 11.499641418457031,
|
| 2522 |
+
"frac_dead": 1.4437570143854828e-06,
|
| 2523 |
+
"frac_sat": 7.947286206899662e-08
|
| 2524 |
+
},
|
| 2525 |
+
"dec_blocks.4.conv2": {
|
| 2526 |
+
"mean": -0.008350415155291557,
|
| 2527 |
+
"std": 0.5789016485214233,
|
| 2528 |
+
"abs_max": 5.625,
|
| 2529 |
+
"frac_dead": 5.3591200412483886e-05,
|
| 2530 |
+
"frac_sat": 0.0
|
| 2531 |
+
},
|
| 2532 |
+
"dec_blocks.4.skip": {
|
| 2533 |
+
"mean": 0.009770325385034084,
|
| 2534 |
+
"std": 1.1018366813659668,
|
| 2535 |
+
"abs_max": 8.875,
|
| 2536 |
+
"frac_dead": 5.510118353413418e-05,
|
| 2537 |
+
"frac_sat": 0.0
|
| 2538 |
+
},
|
| 2539 |
+
"dec_ups.4.op": {
|
| 2540 |
+
"mean": -0.18501953780651093,
|
| 2541 |
+
"std": 2.0624237060546875,
|
| 2542 |
+
"abs_max": 22.75,
|
| 2543 |
+
"frac_dead": 2.249413046229165e-05,
|
| 2544 |
+
"frac_sat": 0.002946823835372925
|
| 2545 |
+
},
|
| 2546 |
+
"dec_blocks.5.norm1": {
|
| 2547 |
+
"mean": -0.003984882961958647,
|
| 2548 |
+
"std": 0.9979493021965027,
|
| 2549 |
+
"abs_max": 12.960043907165527,
|
| 2550 |
+
"frac_dead": 1.1656019296424347e-06,
|
| 2551 |
+
"frac_sat": 4.106097719613899e-07
|
| 2552 |
+
},
|
| 2553 |
+
"dec_blocks.5.conv1": {
|
| 2554 |
+
"mean": -0.0586412288248539,
|
| 2555 |
+
"std": 1.5150682926177979,
|
| 2556 |
+
"abs_max": 22.75,
|
| 2557 |
+
"frac_dead": 1.675883868301753e-05,
|
| 2558 |
+
"frac_sat": 0.0001257459371117875
|
| 2559 |
+
},
|
| 2560 |
+
"dec_blocks.5.t_proj": {
|
| 2561 |
+
"mean": 0.011308521963655949,
|
| 2562 |
+
"std": 0.215815931558609,
|
| 2563 |
+
"abs_max": 1.6015625,
|
| 2564 |
+
"frac_dead": 0.0,
|
| 2565 |
+
"frac_sat": 0.0
|
| 2566 |
+
},
|
| 2567 |
+
"dec_blocks.5.norm2": {
|
| 2568 |
+
"mean": 0.0006456120754592121,
|
| 2569 |
+
"std": 1.0051116943359375,
|
| 2570 |
+
"abs_max": 17.839805603027344,
|
| 2571 |
+
"frac_dead": 1.1026859283447266e-06,
|
| 2572 |
+
"frac_sat": 4.159079708188074e-06
|
| 2573 |
+
},
|
| 2574 |
+
"dec_blocks.5.conv2": {
|
| 2575 |
+
"mean": -0.9917439222335815,
|
| 2576 |
+
"std": 1.8552374839782715,
|
| 2577 |
+
"abs_max": 27.125,
|
| 2578 |
+
"frac_dead": 2.7272437364445068e-05,
|
| 2579 |
+
"frac_sat": 0.0017922355327755213
|
| 2580 |
+
},
|
| 2581 |
+
"dec_blocks.5.skip": {
|
| 2582 |
+
"mean": -0.26965466141700745,
|
| 2583 |
+
"std": 1.6131691932678223,
|
| 2584 |
+
"abs_max": 19.375,
|
| 2585 |
+
"frac_dead": 5.5223703384399414e-05,
|
| 2586 |
+
"frac_sat": 0.00018515189003665
|
| 2587 |
+
},
|
| 2588 |
+
"out_norm": {
|
| 2589 |
+
"mean": -0.004453430417925119,
|
| 2590 |
+
"std": 0.9956813454627991,
|
| 2591 |
+
"abs_max": 13.229631423950195,
|
| 2592 |
+
"frac_dead": 1.8510553445594269e-06,
|
| 2593 |
+
"frac_sat": 3.7418470810735016e-07
|
| 2594 |
+
},
|
| 2595 |
+
"out_conv": {
|
| 2596 |
+
"mean": 0.49139833450317383,
|
| 2597 |
+
"std": 0.23420317471027374,
|
| 2598 |
+
"abs_max": 1.046875,
|
| 2599 |
+
"frac_dead": 3.602769766075653e-06,
|
| 2600 |
+
"frac_sat": 0.0
|
| 2601 |
+
}
|
| 2602 |
+
}
|
| 2603 |
+
}
|
diagnostics/per_layer_iter2500.json
ADDED
|
@@ -0,0 +1,2603 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"iter": 2500,
|
| 3 |
+
"per_layer": {
|
| 4 |
+
"time_mlp.1.weight": {
|
| 5 |
+
"w_norm": 18.984146118164062,
|
| 6 |
+
"g_norm": 0.06113049387931824,
|
| 7 |
+
"shape": [
|
| 8 |
+
1024,
|
| 9 |
+
256
|
| 10 |
+
]
|
| 11 |
+
},
|
| 12 |
+
"time_mlp.1.bias": {
|
| 13 |
+
"w_norm": 1.159756064414978,
|
| 14 |
+
"g_norm": 0.005403383634984493,
|
| 15 |
+
"shape": [
|
| 16 |
+
1024
|
| 17 |
+
]
|
| 18 |
+
},
|
| 19 |
+
"time_mlp.3.weight": {
|
| 20 |
+
"w_norm": 9.721464157104492,
|
| 21 |
+
"g_norm": 0.00959283672273159,
|
| 22 |
+
"shape": [
|
| 23 |
+
256,
|
| 24 |
+
1024
|
| 25 |
+
]
|
| 26 |
+
},
|
| 27 |
+
"time_mlp.3.bias": {
|
| 28 |
+
"w_norm": 0.2891911566257477,
|
| 29 |
+
"g_norm": 0.0048805121332407,
|
| 30 |
+
"shape": [
|
| 31 |
+
256
|
| 32 |
+
]
|
| 33 |
+
},
|
| 34 |
+
"input_conv.weight": {
|
| 35 |
+
"w_norm": 7.95150089263916,
|
| 36 |
+
"g_norm": 0.06437147408723831,
|
| 37 |
+
"shape": [
|
| 38 |
+
192,
|
| 39 |
+
6,
|
| 40 |
+
3,
|
| 41 |
+
3
|
| 42 |
+
]
|
| 43 |
+
},
|
| 44 |
+
"input_conv.bias": {
|
| 45 |
+
"w_norm": 1.1900471448898315,
|
| 46 |
+
"g_norm": 0.0870174691081047,
|
| 47 |
+
"shape": [
|
| 48 |
+
192
|
| 49 |
+
]
|
| 50 |
+
},
|
| 51 |
+
"enc_blocks.0.norm1.weight": {
|
| 52 |
+
"w_norm": 13.795973777770996,
|
| 53 |
+
"g_norm": 0.022462671622633934,
|
| 54 |
+
"shape": [
|
| 55 |
+
192
|
| 56 |
+
]
|
| 57 |
+
},
|
| 58 |
+
"enc_blocks.0.norm1.bias": {
|
| 59 |
+
"w_norm": 0.10943365097045898,
|
| 60 |
+
"g_norm": 0.03537271171808243,
|
| 61 |
+
"shape": [
|
| 62 |
+
192
|
| 63 |
+
]
|
| 64 |
+
},
|
| 65 |
+
"enc_blocks.0.conv1.weight": {
|
| 66 |
+
"w_norm": 8.521758079528809,
|
| 67 |
+
"g_norm": 0.18474127352237701,
|
| 68 |
+
"shape": [
|
| 69 |
+
192,
|
| 70 |
+
192,
|
| 71 |
+
3,
|
| 72 |
+
3
|
| 73 |
+
]
|
| 74 |
+
},
|
| 75 |
+
"enc_blocks.0.conv1.bias": {
|
| 76 |
+
"w_norm": 0.20209793746471405,
|
| 77 |
+
"g_norm": 0.019622452557086945,
|
| 78 |
+
"shape": [
|
| 79 |
+
192
|
| 80 |
+
]
|
| 81 |
+
},
|
| 82 |
+
"enc_blocks.0.t_proj.weight": {
|
| 83 |
+
"w_norm": 8.095259666442871,
|
| 84 |
+
"g_norm": 0.009041828103363514,
|
| 85 |
+
"shape": [
|
| 86 |
+
192,
|
| 87 |
+
256
|
| 88 |
+
]
|
| 89 |
+
},
|
| 90 |
+
"enc_blocks.0.t_proj.bias": {
|
| 91 |
+
"w_norm": 0.5026264190673828,
|
| 92 |
+
"g_norm": 0.019634367898106575,
|
| 93 |
+
"shape": [
|
| 94 |
+
192
|
| 95 |
+
]
|
| 96 |
+
},
|
| 97 |
+
"enc_blocks.0.norm2.weight": {
|
| 98 |
+
"w_norm": 13.82917308807373,
|
| 99 |
+
"g_norm": 0.007581069599837065,
|
| 100 |
+
"shape": [
|
| 101 |
+
192
|
| 102 |
+
]
|
| 103 |
+
},
|
| 104 |
+
"enc_blocks.0.norm2.bias": {
|
| 105 |
+
"w_norm": 0.07328630983829498,
|
| 106 |
+
"g_norm": 0.008389967493712902,
|
| 107 |
+
"shape": [
|
| 108 |
+
192
|
| 109 |
+
]
|
| 110 |
+
},
|
| 111 |
+
"enc_blocks.0.conv2.weight": {
|
| 112 |
+
"w_norm": 8.377623558044434,
|
| 113 |
+
"g_norm": 0.05995709449052811,
|
| 114 |
+
"shape": [
|
| 115 |
+
192,
|
| 116 |
+
192,
|
| 117 |
+
3,
|
| 118 |
+
3
|
| 119 |
+
]
|
| 120 |
+
},
|
| 121 |
+
"enc_blocks.0.conv2.bias": {
|
| 122 |
+
"w_norm": 0.18875400722026825,
|
| 123 |
+
"g_norm": 0.004590434953570366,
|
| 124 |
+
"shape": [
|
| 125 |
+
192
|
| 126 |
+
]
|
| 127 |
+
},
|
| 128 |
+
"enc_blocks.1.norm1.weight": {
|
| 129 |
+
"w_norm": 13.848093032836914,
|
| 130 |
+
"g_norm": 0.00014158007979858667,
|
| 131 |
+
"shape": [
|
| 132 |
+
192
|
| 133 |
+
]
|
| 134 |
+
},
|
| 135 |
+
"enc_blocks.1.norm1.bias": {
|
| 136 |
+
"w_norm": 0.042593494057655334,
|
| 137 |
+
"g_norm": 0.0001738002902129665,
|
| 138 |
+
"shape": [
|
| 139 |
+
192
|
| 140 |
+
]
|
| 141 |
+
},
|
| 142 |
+
"enc_blocks.1.conv1.weight": {
|
| 143 |
+
"w_norm": 8.286897659301758,
|
| 144 |
+
"g_norm": 0.0024213152937591076,
|
| 145 |
+
"shape": [
|
| 146 |
+
192,
|
| 147 |
+
192,
|
| 148 |
+
3,
|
| 149 |
+
3
|
| 150 |
+
]
|
| 151 |
+
},
|
| 152 |
+
"enc_blocks.1.conv1.bias": {
|
| 153 |
+
"w_norm": 0.19198547303676605,
|
| 154 |
+
"g_norm": 0.0001342037576250732,
|
| 155 |
+
"shape": [
|
| 156 |
+
192
|
| 157 |
+
]
|
| 158 |
+
},
|
| 159 |
+
"enc_blocks.1.t_proj.weight": {
|
| 160 |
+
"w_norm": 8.197196960449219,
|
| 161 |
+
"g_norm": 0.00010108872811542824,
|
| 162 |
+
"shape": [
|
| 163 |
+
192,
|
| 164 |
+
256
|
| 165 |
+
]
|
| 166 |
+
},
|
| 167 |
+
"enc_blocks.1.t_proj.bias": {
|
| 168 |
+
"w_norm": 0.5223928093910217,
|
| 169 |
+
"g_norm": 0.00013416219735518098,
|
| 170 |
+
"shape": [
|
| 171 |
+
192
|
| 172 |
+
]
|
| 173 |
+
},
|
| 174 |
+
"enc_blocks.1.norm2.weight": {
|
| 175 |
+
"w_norm": 13.820600509643555,
|
| 176 |
+
"g_norm": 0.00015041811275295913,
|
| 177 |
+
"shape": [
|
| 178 |
+
192
|
| 179 |
+
]
|
| 180 |
+
},
|
| 181 |
+
"enc_blocks.1.norm2.bias": {
|
| 182 |
+
"w_norm": 0.06845534592866898,
|
| 183 |
+
"g_norm": 0.0001458735205233097,
|
| 184 |
+
"shape": [
|
| 185 |
+
192
|
| 186 |
+
]
|
| 187 |
+
},
|
| 188 |
+
"enc_blocks.1.conv2.weight": {
|
| 189 |
+
"w_norm": 8.267409324645996,
|
| 190 |
+
"g_norm": 0.002303261077031493,
|
| 191 |
+
"shape": [
|
| 192 |
+
192,
|
| 193 |
+
192,
|
| 194 |
+
3,
|
| 195 |
+
3
|
| 196 |
+
]
|
| 197 |
+
},
|
| 198 |
+
"enc_blocks.1.conv2.bias": {
|
| 199 |
+
"w_norm": 0.2089850902557373,
|
| 200 |
+
"g_norm": 0.00011512059427332133,
|
| 201 |
+
"shape": [
|
| 202 |
+
192
|
| 203 |
+
]
|
| 204 |
+
},
|
| 205 |
+
"enc_blocks.2.norm1.weight": {
|
| 206 |
+
"w_norm": 13.81585693359375,
|
| 207 |
+
"g_norm": 2.5233930500689894e-05,
|
| 208 |
+
"shape": [
|
| 209 |
+
192
|
| 210 |
+
]
|
| 211 |
+
},
|
| 212 |
+
"enc_blocks.2.norm1.bias": {
|
| 213 |
+
"w_norm": 0.06356502324342728,
|
| 214 |
+
"g_norm": 2.1722678866353817e-05,
|
| 215 |
+
"shape": [
|
| 216 |
+
192
|
| 217 |
+
]
|
| 218 |
+
},
|
| 219 |
+
"enc_blocks.2.conv1.weight": {
|
| 220 |
+
"w_norm": 11.720547676086426,
|
| 221 |
+
"g_norm": 0.0006722903344780207,
|
| 222 |
+
"shape": [
|
| 223 |
+
384,
|
| 224 |
+
192,
|
| 225 |
+
3,
|
| 226 |
+
3
|
| 227 |
+
]
|
| 228 |
+
},
|
| 229 |
+
"enc_blocks.2.conv1.bias": {
|
| 230 |
+
"w_norm": 0.26672983169555664,
|
| 231 |
+
"g_norm": 2.4164550268324092e-05,
|
| 232 |
+
"shape": [
|
| 233 |
+
384
|
| 234 |
+
]
|
| 235 |
+
},
|
| 236 |
+
"enc_blocks.2.t_proj.weight": {
|
| 237 |
+
"w_norm": 11.856393814086914,
|
| 238 |
+
"g_norm": 2.154646244889591e-05,
|
| 239 |
+
"shape": [
|
| 240 |
+
384,
|
| 241 |
+
256
|
| 242 |
+
]
|
| 243 |
+
},
|
| 244 |
+
"enc_blocks.2.t_proj.bias": {
|
| 245 |
+
"w_norm": 0.7216393947601318,
|
| 246 |
+
"g_norm": 2.4155662686098367e-05,
|
| 247 |
+
"shape": [
|
| 248 |
+
384
|
| 249 |
+
]
|
| 250 |
+
},
|
| 251 |
+
"enc_blocks.2.norm2.weight": {
|
| 252 |
+
"w_norm": 19.427146911621094,
|
| 253 |
+
"g_norm": 6.073198892408982e-05,
|
| 254 |
+
"shape": [
|
| 255 |
+
384
|
| 256 |
+
]
|
| 257 |
+
},
|
| 258 |
+
"enc_blocks.2.norm2.bias": {
|
| 259 |
+
"w_norm": 0.21375973522663116,
|
| 260 |
+
"g_norm": 3.191018186043948e-05,
|
| 261 |
+
"shape": [
|
| 262 |
+
384
|
| 263 |
+
]
|
| 264 |
+
},
|
| 265 |
+
"enc_blocks.2.conv2.weight": {
|
| 266 |
+
"w_norm": 11.735209465026855,
|
| 267 |
+
"g_norm": 0.0007884344086050987,
|
| 268 |
+
"shape": [
|
| 269 |
+
384,
|
| 270 |
+
384,
|
| 271 |
+
3,
|
| 272 |
+
3
|
| 273 |
+
]
|
| 274 |
+
},
|
| 275 |
+
"enc_blocks.2.conv2.bias": {
|
| 276 |
+
"w_norm": 0.20197713375091553,
|
| 277 |
+
"g_norm": 2.198413130827248e-05,
|
| 278 |
+
"shape": [
|
| 279 |
+
384
|
| 280 |
+
]
|
| 281 |
+
},
|
| 282 |
+
"enc_blocks.2.skip.weight": {
|
| 283 |
+
"w_norm": 11.192889213562012,
|
| 284 |
+
"g_norm": 0.0007470609270967543,
|
| 285 |
+
"shape": [
|
| 286 |
+
384,
|
| 287 |
+
192,
|
| 288 |
+
1,
|
| 289 |
+
1
|
| 290 |
+
]
|
| 291 |
+
},
|
| 292 |
+
"enc_blocks.2.skip.bias": {
|
| 293 |
+
"w_norm": 0.8284077644348145,
|
| 294 |
+
"g_norm": 2.198413130827248e-05,
|
| 295 |
+
"shape": [
|
| 296 |
+
384
|
| 297 |
+
]
|
| 298 |
+
},
|
| 299 |
+
"enc_blocks.3.norm1.weight": {
|
| 300 |
+
"w_norm": 19.583465576171875,
|
| 301 |
+
"g_norm": 4.372146577225067e-06,
|
| 302 |
+
"shape": [
|
| 303 |
+
384
|
| 304 |
+
]
|
| 305 |
+
},
|
| 306 |
+
"enc_blocks.3.norm1.bias": {
|
| 307 |
+
"w_norm": 0.05683818459510803,
|
| 308 |
+
"g_norm": 3.3718665690685157e-06,
|
| 309 |
+
"shape": [
|
| 310 |
+
384
|
| 311 |
+
]
|
| 312 |
+
},
|
| 313 |
+
"enc_blocks.3.conv1.weight": {
|
| 314 |
+
"w_norm": 11.894614219665527,
|
| 315 |
+
"g_norm": 7.07660146872513e-05,
|
| 316 |
+
"shape": [
|
| 317 |
+
384,
|
| 318 |
+
384,
|
| 319 |
+
3,
|
| 320 |
+
3
|
| 321 |
+
]
|
| 322 |
+
},
|
| 323 |
+
"enc_blocks.3.conv1.bias": {
|
| 324 |
+
"w_norm": 0.1975417584180832,
|
| 325 |
+
"g_norm": 2.109597971866606e-06,
|
| 326 |
+
"shape": [
|
| 327 |
+
384
|
| 328 |
+
]
|
| 329 |
+
},
|
| 330 |
+
"enc_blocks.3.t_proj.weight": {
|
| 331 |
+
"w_norm": 11.404732704162598,
|
| 332 |
+
"g_norm": 1.5137313766899751e-06,
|
| 333 |
+
"shape": [
|
| 334 |
+
384,
|
| 335 |
+
256
|
| 336 |
+
]
|
| 337 |
+
},
|
| 338 |
+
"enc_blocks.3.t_proj.bias": {
|
| 339 |
+
"w_norm": 0.6922511458396912,
|
| 340 |
+
"g_norm": 2.109284650941845e-06,
|
| 341 |
+
"shape": [
|
| 342 |
+
384
|
| 343 |
+
]
|
| 344 |
+
},
|
| 345 |
+
"enc_blocks.3.norm2.weight": {
|
| 346 |
+
"w_norm": 19.406980514526367,
|
| 347 |
+
"g_norm": 3.6609660583053483e-06,
|
| 348 |
+
"shape": [
|
| 349 |
+
384
|
| 350 |
+
]
|
| 351 |
+
},
|
| 352 |
+
"enc_blocks.3.norm2.bias": {
|
| 353 |
+
"w_norm": 0.19815607368946075,
|
| 354 |
+
"g_norm": 3.5790540096058976e-06,
|
| 355 |
+
"shape": [
|
| 356 |
+
384
|
| 357 |
+
]
|
| 358 |
+
},
|
| 359 |
+
"enc_blocks.3.conv2.weight": {
|
| 360 |
+
"w_norm": 11.528669357299805,
|
| 361 |
+
"g_norm": 0.00015909784997347742,
|
| 362 |
+
"shape": [
|
| 363 |
+
384,
|
| 364 |
+
384,
|
| 365 |
+
3,
|
| 366 |
+
3
|
| 367 |
+
]
|
| 368 |
+
},
|
| 369 |
+
"enc_blocks.3.conv2.bias": {
|
| 370 |
+
"w_norm": 0.19179506599903107,
|
| 371 |
+
"g_norm": 5.194372533878777e-06,
|
| 372 |
+
"shape": [
|
| 373 |
+
384
|
| 374 |
+
]
|
| 375 |
+
},
|
| 376 |
+
"enc_blocks.4.norm1.weight": {
|
| 377 |
+
"w_norm": 19.590572357177734,
|
| 378 |
+
"g_norm": 1.2455842579583987e-06,
|
| 379 |
+
"shape": [
|
| 380 |
+
384
|
| 381 |
+
]
|
| 382 |
+
},
|
| 383 |
+
"enc_blocks.4.norm1.bias": {
|
| 384 |
+
"w_norm": 0.06201266124844551,
|
| 385 |
+
"g_norm": 1.6669519027345814e-06,
|
| 386 |
+
"shape": [
|
| 387 |
+
384
|
| 388 |
+
]
|
| 389 |
+
},
|
| 390 |
+
"enc_blocks.4.conv1.weight": {
|
| 391 |
+
"w_norm": 16.711240768432617,
|
| 392 |
+
"g_norm": 2.725605918385554e-05,
|
| 393 |
+
"shape": [
|
| 394 |
+
768,
|
| 395 |
+
384,
|
| 396 |
+
3,
|
| 397 |
+
3
|
| 398 |
+
]
|
| 399 |
+
},
|
| 400 |
+
"enc_blocks.4.conv1.bias": {
|
| 401 |
+
"w_norm": 0.2846062183380127,
|
| 402 |
+
"g_norm": 1.3469352779793553e-06,
|
| 403 |
+
"shape": [
|
| 404 |
+
768
|
| 405 |
+
]
|
| 406 |
+
},
|
| 407 |
+
"enc_blocks.4.t_proj.weight": {
|
| 408 |
+
"w_norm": 16.05572509765625,
|
| 409 |
+
"g_norm": 7.266082775458926e-07,
|
| 410 |
+
"shape": [
|
| 411 |
+
768,
|
| 412 |
+
256
|
| 413 |
+
]
|
| 414 |
+
},
|
| 415 |
+
"enc_blocks.4.t_proj.bias": {
|
| 416 |
+
"w_norm": 0.9931569695472717,
|
| 417 |
+
"g_norm": 1.3468946917782887e-06,
|
| 418 |
+
"shape": [
|
| 419 |
+
768
|
| 420 |
+
]
|
| 421 |
+
},
|
| 422 |
+
"enc_blocks.4.norm2.weight": {
|
| 423 |
+
"w_norm": 27.43099021911621,
|
| 424 |
+
"g_norm": 1.147222178587981e-06,
|
| 425 |
+
"shape": [
|
| 426 |
+
768
|
| 427 |
+
]
|
| 428 |
+
},
|
| 429 |
+
"enc_blocks.4.norm2.bias": {
|
| 430 |
+
"w_norm": 0.3143310844898224,
|
| 431 |
+
"g_norm": 2.341718982279417e-06,
|
| 432 |
+
"shape": [
|
| 433 |
+
768
|
| 434 |
+
]
|
| 435 |
+
},
|
| 436 |
+
"enc_blocks.4.conv2.weight": {
|
| 437 |
+
"w_norm": 16.32482147216797,
|
| 438 |
+
"g_norm": 6.44704487058334e-05,
|
| 439 |
+
"shape": [
|
| 440 |
+
768,
|
| 441 |
+
768,
|
| 442 |
+
3,
|
| 443 |
+
3
|
| 444 |
+
]
|
| 445 |
+
},
|
| 446 |
+
"enc_blocks.4.conv2.bias": {
|
| 447 |
+
"w_norm": 0.1955537497997284,
|
| 448 |
+
"g_norm": 3.4157737900386564e-06,
|
| 449 |
+
"shape": [
|
| 450 |
+
768
|
| 451 |
+
]
|
| 452 |
+
},
|
| 453 |
+
"enc_blocks.4.skip.weight": {
|
| 454 |
+
"w_norm": 15.805265426635742,
|
| 455 |
+
"g_norm": 7.485902460757643e-05,
|
| 456 |
+
"shape": [
|
| 457 |
+
768,
|
| 458 |
+
384,
|
| 459 |
+
1,
|
| 460 |
+
1
|
| 461 |
+
]
|
| 462 |
+
},
|
| 463 |
+
"enc_blocks.4.skip.bias": {
|
| 464 |
+
"w_norm": 0.8076781630516052,
|
| 465 |
+
"g_norm": 3.4157737900386564e-06,
|
| 466 |
+
"shape": [
|
| 467 |
+
768
|
| 468 |
+
]
|
| 469 |
+
},
|
| 470 |
+
"enc_blocks.5.norm1.weight": {
|
| 471 |
+
"w_norm": 27.713409423828125,
|
| 472 |
+
"g_norm": 3.229412470773241e-07,
|
| 473 |
+
"shape": [
|
| 474 |
+
768
|
| 475 |
+
]
|
| 476 |
+
},
|
| 477 |
+
"enc_blocks.5.norm1.bias": {
|
| 478 |
+
"w_norm": 0.05912580341100693,
|
| 479 |
+
"g_norm": 2.589934808838734e-07,
|
| 480 |
+
"shape": [
|
| 481 |
+
768
|
| 482 |
+
]
|
| 483 |
+
},
|
| 484 |
+
"enc_blocks.5.conv1.weight": {
|
| 485 |
+
"w_norm": 16.959115982055664,
|
| 486 |
+
"g_norm": 1.2284299373277463e-05,
|
| 487 |
+
"shape": [
|
| 488 |
+
768,
|
| 489 |
+
768,
|
| 490 |
+
3,
|
| 491 |
+
3
|
| 492 |
+
]
|
| 493 |
+
},
|
| 494 |
+
"enc_blocks.5.conv1.bias": {
|
| 495 |
+
"w_norm": 0.20442481338977814,
|
| 496 |
+
"g_norm": 2.93640937343298e-07,
|
| 497 |
+
"shape": [
|
| 498 |
+
768
|
| 499 |
+
]
|
| 500 |
+
},
|
| 501 |
+
"enc_blocks.5.t_proj.weight": {
|
| 502 |
+
"w_norm": 16.0108642578125,
|
| 503 |
+
"g_norm": 1.805583877967365e-07,
|
| 504 |
+
"shape": [
|
| 505 |
+
768,
|
| 506 |
+
256
|
| 507 |
+
]
|
| 508 |
+
},
|
| 509 |
+
"enc_blocks.5.t_proj.bias": {
|
| 510 |
+
"w_norm": 0.9978851079940796,
|
| 511 |
+
"g_norm": 2.9374672294579796e-07,
|
| 512 |
+
"shape": [
|
| 513 |
+
768
|
| 514 |
+
]
|
| 515 |
+
},
|
| 516 |
+
"enc_blocks.5.norm2.weight": {
|
| 517 |
+
"w_norm": 27.502161026000977,
|
| 518 |
+
"g_norm": 4.7526637558803486e-07,
|
| 519 |
+
"shape": [
|
| 520 |
+
768
|
| 521 |
+
]
|
| 522 |
+
},
|
| 523 |
+
"enc_blocks.5.norm2.bias": {
|
| 524 |
+
"w_norm": 0.24743936955928802,
|
| 525 |
+
"g_norm": 5.183755433790793e-07,
|
| 526 |
+
"shape": [
|
| 527 |
+
768
|
| 528 |
+
]
|
| 529 |
+
},
|
| 530 |
+
"enc_blocks.5.conv2.weight": {
|
| 531 |
+
"w_norm": 16.4858341217041,
|
| 532 |
+
"g_norm": 3.6588418879546225e-05,
|
| 533 |
+
"shape": [
|
| 534 |
+
768,
|
| 535 |
+
768,
|
| 536 |
+
3,
|
| 537 |
+
3
|
| 538 |
+
]
|
| 539 |
+
},
|
| 540 |
+
"enc_blocks.5.conv2.bias": {
|
| 541 |
+
"w_norm": 0.1915251463651657,
|
| 542 |
+
"g_norm": 1.2130102504670504e-06,
|
| 543 |
+
"shape": [
|
| 544 |
+
768
|
| 545 |
+
]
|
| 546 |
+
},
|
| 547 |
+
"enc_attns.3.norm.weight": {
|
| 548 |
+
"w_norm": 19.52772331237793,
|
| 549 |
+
"g_norm": 1.2919706477987347e-06,
|
| 550 |
+
"shape": [
|
| 551 |
+
384
|
| 552 |
+
]
|
| 553 |
+
},
|
| 554 |
+
"enc_attns.3.norm.bias": {
|
| 555 |
+
"w_norm": 0.04880025237798691,
|
| 556 |
+
"g_norm": 2.042158257609117e-06,
|
| 557 |
+
"shape": [
|
| 558 |
+
384
|
| 559 |
+
]
|
| 560 |
+
},
|
| 561 |
+
"enc_attns.3.qkv.weight": {
|
| 562 |
+
"w_norm": 19.67283821105957,
|
| 563 |
+
"g_norm": 3.667322744149715e-05,
|
| 564 |
+
"shape": [
|
| 565 |
+
1152,
|
| 566 |
+
384,
|
| 567 |
+
1,
|
| 568 |
+
1
|
| 569 |
+
]
|
| 570 |
+
},
|
| 571 |
+
"enc_attns.3.qkv.bias": {
|
| 572 |
+
"w_norm": 1.007038950920105,
|
| 573 |
+
"g_norm": 3.1300469345296733e-06,
|
| 574 |
+
"shape": [
|
| 575 |
+
1152
|
| 576 |
+
]
|
| 577 |
+
},
|
| 578 |
+
"enc_attns.3.proj.weight": {
|
| 579 |
+
"w_norm": 11.307280540466309,
|
| 580 |
+
"g_norm": 3.458759965724312e-05,
|
| 581 |
+
"shape": [
|
| 582 |
+
384,
|
| 583 |
+
384,
|
| 584 |
+
1,
|
| 585 |
+
1
|
| 586 |
+
]
|
| 587 |
+
},
|
| 588 |
+
"enc_attns.3.proj.bias": {
|
| 589 |
+
"w_norm": 0.562714159488678,
|
| 590 |
+
"g_norm": 5.2210789363016374e-06,
|
| 591 |
+
"shape": [
|
| 592 |
+
384
|
| 593 |
+
]
|
| 594 |
+
},
|
| 595 |
+
"enc_attns.4.norm.weight": {
|
| 596 |
+
"w_norm": 27.598068237304688,
|
| 597 |
+
"g_norm": 1.305578621213499e-06,
|
| 598 |
+
"shape": [
|
| 599 |
+
768
|
| 600 |
+
]
|
| 601 |
+
},
|
| 602 |
+
"enc_attns.4.norm.bias": {
|
| 603 |
+
"w_norm": 0.051544707268476486,
|
| 604 |
+
"g_norm": 1.4329129953694064e-06,
|
| 605 |
+
"shape": [
|
| 606 |
+
768
|
| 607 |
+
]
|
| 608 |
+
},
|
| 609 |
+
"enc_attns.4.qkv.weight": {
|
| 610 |
+
"w_norm": 27.992589950561523,
|
| 611 |
+
"g_norm": 4.070655631949194e-05,
|
| 612 |
+
"shape": [
|
| 613 |
+
2304,
|
| 614 |
+
768,
|
| 615 |
+
1,
|
| 616 |
+
1
|
| 617 |
+
]
|
| 618 |
+
},
|
| 619 |
+
"enc_attns.4.qkv.bias": {
|
| 620 |
+
"w_norm": 1.0139776468276978,
|
| 621 |
+
"g_norm": 1.8191481103713159e-06,
|
| 622 |
+
"shape": [
|
| 623 |
+
2304
|
| 624 |
+
]
|
| 625 |
+
},
|
| 626 |
+
"enc_attns.4.proj.weight": {
|
| 627 |
+
"w_norm": 16.01902961730957,
|
| 628 |
+
"g_norm": 4.0771319618215784e-05,
|
| 629 |
+
"shape": [
|
| 630 |
+
768,
|
| 631 |
+
768,
|
| 632 |
+
1,
|
| 633 |
+
1
|
| 634 |
+
]
|
| 635 |
+
},
|
| 636 |
+
"enc_attns.4.proj.bias": {
|
| 637 |
+
"w_norm": 0.5725747346878052,
|
| 638 |
+
"g_norm": 3.0066312319831923e-06,
|
| 639 |
+
"shape": [
|
| 640 |
+
768
|
| 641 |
+
]
|
| 642 |
+
},
|
| 643 |
+
"enc_attns.5.norm.weight": {
|
| 644 |
+
"w_norm": 27.660709381103516,
|
| 645 |
+
"g_norm": 3.203698213383177e-07,
|
| 646 |
+
"shape": [
|
| 647 |
+
768
|
| 648 |
+
]
|
| 649 |
+
},
|
| 650 |
+
"enc_attns.5.norm.bias": {
|
| 651 |
+
"w_norm": 0.048515934497117996,
|
| 652 |
+
"g_norm": 4.2762215457514685e-07,
|
| 653 |
+
"shape": [
|
| 654 |
+
768
|
| 655 |
+
]
|
| 656 |
+
},
|
| 657 |
+
"enc_attns.5.qkv.weight": {
|
| 658 |
+
"w_norm": 28.006317138671875,
|
| 659 |
+
"g_norm": 1.4664356058347039e-05,
|
| 660 |
+
"shape": [
|
| 661 |
+
2304,
|
| 662 |
+
768,
|
| 663 |
+
1,
|
| 664 |
+
1
|
| 665 |
+
]
|
| 666 |
+
},
|
| 667 |
+
"enc_attns.5.qkv.bias": {
|
| 668 |
+
"w_norm": 0.9999328255653381,
|
| 669 |
+
"g_norm": 6.966732257751573e-07,
|
| 670 |
+
"shape": [
|
| 671 |
+
2304
|
| 672 |
+
]
|
| 673 |
+
},
|
| 674 |
+
"enc_attns.5.proj.weight": {
|
| 675 |
+
"w_norm": 16.06597137451172,
|
| 676 |
+
"g_norm": 1.613639142306056e-05,
|
| 677 |
+
"shape": [
|
| 678 |
+
768,
|
| 679 |
+
768,
|
| 680 |
+
1,
|
| 681 |
+
1
|
| 682 |
+
]
|
| 683 |
+
},
|
| 684 |
+
"enc_attns.5.proj.bias": {
|
| 685 |
+
"w_norm": 0.5973749160766602,
|
| 686 |
+
"g_norm": 1.1837975080197793e-06,
|
| 687 |
+
"shape": [
|
| 688 |
+
768
|
| 689 |
+
]
|
| 690 |
+
},
|
| 691 |
+
"enc_downs.0.op.weight": {
|
| 692 |
+
"w_norm": 8.197033882141113,
|
| 693 |
+
"g_norm": 0.00462199654430151,
|
| 694 |
+
"shape": [
|
| 695 |
+
192,
|
| 696 |
+
192,
|
| 697 |
+
3,
|
| 698 |
+
3
|
| 699 |
+
]
|
| 700 |
+
},
|
| 701 |
+
"enc_downs.0.op.bias": {
|
| 702 |
+
"w_norm": 0.1845771074295044,
|
| 703 |
+
"g_norm": 0.00018332099716644734,
|
| 704 |
+
"shape": [
|
| 705 |
+
192
|
| 706 |
+
]
|
| 707 |
+
},
|
| 708 |
+
"enc_downs.1.op.weight": {
|
| 709 |
+
"w_norm": 8.111349105834961,
|
| 710 |
+
"g_norm": 0.001202525570988655,
|
| 711 |
+
"shape": [
|
| 712 |
+
192,
|
| 713 |
+
192,
|
| 714 |
+
3,
|
| 715 |
+
3
|
| 716 |
+
]
|
| 717 |
+
},
|
| 718 |
+
"enc_downs.1.op.bias": {
|
| 719 |
+
"w_norm": 0.19379548728466034,
|
| 720 |
+
"g_norm": 1.966803210962098e-05,
|
| 721 |
+
"shape": [
|
| 722 |
+
192
|
| 723 |
+
]
|
| 724 |
+
},
|
| 725 |
+
"enc_downs.2.op.weight": {
|
| 726 |
+
"w_norm": 11.510343551635742,
|
| 727 |
+
"g_norm": 0.00029774088761769235,
|
| 728 |
+
"shape": [
|
| 729 |
+
384,
|
| 730 |
+
384,
|
| 731 |
+
3,
|
| 732 |
+
3
|
| 733 |
+
]
|
| 734 |
+
},
|
| 735 |
+
"enc_downs.2.op.bias": {
|
| 736 |
+
"w_norm": 0.1884956955909729,
|
| 737 |
+
"g_norm": 4.963231731380802e-06,
|
| 738 |
+
"shape": [
|
| 739 |
+
384
|
| 740 |
+
]
|
| 741 |
+
},
|
| 742 |
+
"enc_downs.3.op.weight": {
|
| 743 |
+
"w_norm": 11.436445236206055,
|
| 744 |
+
"g_norm": 0.00012070700904587284,
|
| 745 |
+
"shape": [
|
| 746 |
+
384,
|
| 747 |
+
384,
|
| 748 |
+
3,
|
| 749 |
+
3
|
| 750 |
+
]
|
| 751 |
+
},
|
| 752 |
+
"enc_downs.3.op.bias": {
|
| 753 |
+
"w_norm": 0.1916549801826477,
|
| 754 |
+
"g_norm": 2.680138095456641e-06,
|
| 755 |
+
"shape": [
|
| 756 |
+
384
|
| 757 |
+
]
|
| 758 |
+
},
|
| 759 |
+
"enc_downs.4.op.weight": {
|
| 760 |
+
"w_norm": 16.354839324951172,
|
| 761 |
+
"g_norm": 6.465298793045804e-05,
|
| 762 |
+
"shape": [
|
| 763 |
+
768,
|
| 764 |
+
768,
|
| 765 |
+
3,
|
| 766 |
+
3
|
| 767 |
+
]
|
| 768 |
+
},
|
| 769 |
+
"enc_downs.4.op.bias": {
|
| 770 |
+
"w_norm": 0.19893445074558258,
|
| 771 |
+
"g_norm": 1.2337447969912319e-06,
|
| 772 |
+
"shape": [
|
| 773 |
+
768
|
| 774 |
+
]
|
| 775 |
+
},
|
| 776 |
+
"mid1.norm1.weight": {
|
| 777 |
+
"w_norm": 27.714229583740234,
|
| 778 |
+
"g_norm": 2.476989209299063e-07,
|
| 779 |
+
"shape": [
|
| 780 |
+
768
|
| 781 |
+
]
|
| 782 |
+
},
|
| 783 |
+
"mid1.norm1.bias": {
|
| 784 |
+
"w_norm": 0.05086077004671097,
|
| 785 |
+
"g_norm": 2.4525201069991454e-07,
|
| 786 |
+
"shape": [
|
| 787 |
+
768
|
| 788 |
+
]
|
| 789 |
+
},
|
| 790 |
+
"mid1.conv1.weight": {
|
| 791 |
+
"w_norm": 16.912973403930664,
|
| 792 |
+
"g_norm": 8.582318514527287e-06,
|
| 793 |
+
"shape": [
|
| 794 |
+
768,
|
| 795 |
+
768,
|
| 796 |
+
3,
|
| 797 |
+
3
|
| 798 |
+
]
|
| 799 |
+
},
|
| 800 |
+
"mid1.conv1.bias": {
|
| 801 |
+
"w_norm": 0.2083008736371994,
|
| 802 |
+
"g_norm": 2.025992245080488e-07,
|
| 803 |
+
"shape": [
|
| 804 |
+
768
|
| 805 |
+
]
|
| 806 |
+
},
|
| 807 |
+
"mid1.t_proj.weight": {
|
| 808 |
+
"w_norm": 16.017248153686523,
|
| 809 |
+
"g_norm": 1.0194400346108523e-07,
|
| 810 |
+
"shape": [
|
| 811 |
+
768,
|
| 812 |
+
256
|
| 813 |
+
]
|
| 814 |
+
},
|
| 815 |
+
"mid1.t_proj.bias": {
|
| 816 |
+
"w_norm": 1.0305330753326416,
|
| 817 |
+
"g_norm": 2.0257930088973808e-07,
|
| 818 |
+
"shape": [
|
| 819 |
+
768
|
| 820 |
+
]
|
| 821 |
+
},
|
| 822 |
+
"mid1.norm2.weight": {
|
| 823 |
+
"w_norm": 27.515790939331055,
|
| 824 |
+
"g_norm": 2.3792208025952277e-07,
|
| 825 |
+
"shape": [
|
| 826 |
+
768
|
| 827 |
+
]
|
| 828 |
+
},
|
| 829 |
+
"mid1.norm2.bias": {
|
| 830 |
+
"w_norm": 0.2597298324108124,
|
| 831 |
+
"g_norm": 3.9932524487085175e-07,
|
| 832 |
+
"shape": [
|
| 833 |
+
768
|
| 834 |
+
]
|
| 835 |
+
},
|
| 836 |
+
"mid1.conv2.weight": {
|
| 837 |
+
"w_norm": 16.51875877380371,
|
| 838 |
+
"g_norm": 1.753577816998586e-05,
|
| 839 |
+
"shape": [
|
| 840 |
+
768,
|
| 841 |
+
768,
|
| 842 |
+
3,
|
| 843 |
+
3
|
| 844 |
+
]
|
| 845 |
+
},
|
| 846 |
+
"mid1.conv2.bias": {
|
| 847 |
+
"w_norm": 0.19382815062999725,
|
| 848 |
+
"g_norm": 7.970655815370264e-07,
|
| 849 |
+
"shape": [
|
| 850 |
+
768
|
| 851 |
+
]
|
| 852 |
+
},
|
| 853 |
+
"mid_attn.norm.weight": {
|
| 854 |
+
"w_norm": 27.66170883178711,
|
| 855 |
+
"g_norm": 2.1583025500149233e-07,
|
| 856 |
+
"shape": [
|
| 857 |
+
768
|
| 858 |
+
]
|
| 859 |
+
},
|
| 860 |
+
"mid_attn.norm.bias": {
|
| 861 |
+
"w_norm": 0.05416889116168022,
|
| 862 |
+
"g_norm": 2.994936494360445e-07,
|
| 863 |
+
"shape": [
|
| 864 |
+
768
|
| 865 |
+
]
|
| 866 |
+
},
|
| 867 |
+
"mid_attn.qkv.weight": {
|
| 868 |
+
"w_norm": 27.94774627685547,
|
| 869 |
+
"g_norm": 9.899607903207652e-06,
|
| 870 |
+
"shape": [
|
| 871 |
+
2304,
|
| 872 |
+
768,
|
| 873 |
+
1,
|
| 874 |
+
1
|
| 875 |
+
]
|
| 876 |
+
},
|
| 877 |
+
"mid_attn.qkv.bias": {
|
| 878 |
+
"w_norm": 0.9840787053108215,
|
| 879 |
+
"g_norm": 4.803904403161141e-07,
|
| 880 |
+
"shape": [
|
| 881 |
+
2304
|
| 882 |
+
]
|
| 883 |
+
},
|
| 884 |
+
"mid_attn.proj.weight": {
|
| 885 |
+
"w_norm": 16.05330467224121,
|
| 886 |
+
"g_norm": 1.0125810149475e-05,
|
| 887 |
+
"shape": [
|
| 888 |
+
768,
|
| 889 |
+
768,
|
| 890 |
+
1,
|
| 891 |
+
1
|
| 892 |
+
]
|
| 893 |
+
},
|
| 894 |
+
"mid_attn.proj.bias": {
|
| 895 |
+
"w_norm": 0.5761896371841431,
|
| 896 |
+
"g_norm": 7.825606189726386e-07,
|
| 897 |
+
"shape": [
|
| 898 |
+
768
|
| 899 |
+
]
|
| 900 |
+
},
|
| 901 |
+
"mid2.norm1.weight": {
|
| 902 |
+
"w_norm": 27.72232437133789,
|
| 903 |
+
"g_norm": 2.2575868285912293e-07,
|
| 904 |
+
"shape": [
|
| 905 |
+
768
|
| 906 |
+
]
|
| 907 |
+
},
|
| 908 |
+
"mid2.norm1.bias": {
|
| 909 |
+
"w_norm": 0.06341087073087692,
|
| 910 |
+
"g_norm": 1.9701586495557422e-07,
|
| 911 |
+
"shape": [
|
| 912 |
+
768
|
| 913 |
+
]
|
| 914 |
+
},
|
| 915 |
+
"mid2.conv1.weight": {
|
| 916 |
+
"w_norm": 17.024290084838867,
|
| 917 |
+
"g_norm": 7.21612968845875e-06,
|
| 918 |
+
"shape": [
|
| 919 |
+
768,
|
| 920 |
+
768,
|
| 921 |
+
3,
|
| 922 |
+
3
|
| 923 |
+
]
|
| 924 |
+
},
|
| 925 |
+
"mid2.conv1.bias": {
|
| 926 |
+
"w_norm": 0.20101290941238403,
|
| 927 |
+
"g_norm": 1.657937644949925e-07,
|
| 928 |
+
"shape": [
|
| 929 |
+
768
|
| 930 |
+
]
|
| 931 |
+
},
|
| 932 |
+
"mid2.t_proj.weight": {
|
| 933 |
+
"w_norm": 16.00647735595703,
|
| 934 |
+
"g_norm": 1.1699727764380441e-07,
|
| 935 |
+
"shape": [
|
| 936 |
+
768,
|
| 937 |
+
256
|
| 938 |
+
]
|
| 939 |
+
},
|
| 940 |
+
"mid2.t_proj.bias": {
|
| 941 |
+
"w_norm": 0.9998338222503662,
|
| 942 |
+
"g_norm": 1.6574250594203477e-07,
|
| 943 |
+
"shape": [
|
| 944 |
+
768
|
| 945 |
+
]
|
| 946 |
+
},
|
| 947 |
+
"mid2.norm2.weight": {
|
| 948 |
+
"w_norm": 27.492137908935547,
|
| 949 |
+
"g_norm": 1.739176553883226e-07,
|
| 950 |
+
"shape": [
|
| 951 |
+
768
|
| 952 |
+
]
|
| 953 |
+
},
|
| 954 |
+
"mid2.norm2.bias": {
|
| 955 |
+
"w_norm": 0.26778048276901245,
|
| 956 |
+
"g_norm": 3.6507077538772137e-07,
|
| 957 |
+
"shape": [
|
| 958 |
+
768
|
| 959 |
+
]
|
| 960 |
+
},
|
| 961 |
+
"mid2.conv2.weight": {
|
| 962 |
+
"w_norm": 16.478769302368164,
|
| 963 |
+
"g_norm": 1.3940637472842354e-05,
|
| 964 |
+
"shape": [
|
| 965 |
+
768,
|
| 966 |
+
768,
|
| 967 |
+
3,
|
| 968 |
+
3
|
| 969 |
+
]
|
| 970 |
+
},
|
| 971 |
+
"mid2.conv2.bias": {
|
| 972 |
+
"w_norm": 0.1897919923067093,
|
| 973 |
+
"g_norm": 7.798574301887129e-07,
|
| 974 |
+
"shape": [
|
| 975 |
+
768
|
| 976 |
+
]
|
| 977 |
+
},
|
| 978 |
+
"dec_blocks.0.norm1.weight": {
|
| 979 |
+
"w_norm": 39.217708587646484,
|
| 980 |
+
"g_norm": 3.7901781979599036e-07,
|
| 981 |
+
"shape": [
|
| 982 |
+
1536
|
| 983 |
+
]
|
| 984 |
+
},
|
| 985 |
+
"dec_blocks.0.norm1.bias": {
|
| 986 |
+
"w_norm": 0.09732687473297119,
|
| 987 |
+
"g_norm": 3.5458742786431685e-07,
|
| 988 |
+
"shape": [
|
| 989 |
+
1536
|
| 990 |
+
]
|
| 991 |
+
},
|
| 992 |
+
"dec_blocks.0.conv1.weight": {
|
| 993 |
+
"w_norm": 17.912757873535156,
|
| 994 |
+
"g_norm": 1.0468619620951358e-05,
|
| 995 |
+
"shape": [
|
| 996 |
+
768,
|
| 997 |
+
1536,
|
| 998 |
+
3,
|
| 999 |
+
3
|
| 1000 |
+
]
|
| 1001 |
+
},
|
| 1002 |
+
"dec_blocks.0.conv1.bias": {
|
| 1003 |
+
"w_norm": 0.15449513494968414,
|
| 1004 |
+
"g_norm": 1.8222155517833016e-07,
|
| 1005 |
+
"shape": [
|
| 1006 |
+
768
|
| 1007 |
+
]
|
| 1008 |
+
},
|
| 1009 |
+
"dec_blocks.0.t_proj.weight": {
|
| 1010 |
+
"w_norm": 16.024795532226562,
|
| 1011 |
+
"g_norm": 1.1644377195807465e-07,
|
| 1012 |
+
"shape": [
|
| 1013 |
+
768,
|
| 1014 |
+
256
|
| 1015 |
+
]
|
| 1016 |
+
},
|
| 1017 |
+
"dec_blocks.0.t_proj.bias": {
|
| 1018 |
+
"w_norm": 1.0251551866531372,
|
| 1019 |
+
"g_norm": 1.8221867037482298e-07,
|
| 1020 |
+
"shape": [
|
| 1021 |
+
768
|
| 1022 |
+
]
|
| 1023 |
+
},
|
| 1024 |
+
"dec_blocks.0.norm2.weight": {
|
| 1025 |
+
"w_norm": 27.449193954467773,
|
| 1026 |
+
"g_norm": 3.468843203791039e-07,
|
| 1027 |
+
"shape": [
|
| 1028 |
+
768
|
| 1029 |
+
]
|
| 1030 |
+
},
|
| 1031 |
+
"dec_blocks.0.norm2.bias": {
|
| 1032 |
+
"w_norm": 0.3095521032810211,
|
| 1033 |
+
"g_norm": 8.821044730211725e-07,
|
| 1034 |
+
"shape": [
|
| 1035 |
+
768
|
| 1036 |
+
]
|
| 1037 |
+
},
|
| 1038 |
+
"dec_blocks.0.conv2.weight": {
|
| 1039 |
+
"w_norm": 16.268291473388672,
|
| 1040 |
+
"g_norm": 2.5844508854788728e-05,
|
| 1041 |
+
"shape": [
|
| 1042 |
+
768,
|
| 1043 |
+
768,
|
| 1044 |
+
3,
|
| 1045 |
+
3
|
| 1046 |
+
]
|
| 1047 |
+
},
|
| 1048 |
+
"dec_blocks.0.conv2.bias": {
|
| 1049 |
+
"w_norm": 0.19935624301433563,
|
| 1050 |
+
"g_norm": 1.6814665286801755e-06,
|
| 1051 |
+
"shape": [
|
| 1052 |
+
768
|
| 1053 |
+
]
|
| 1054 |
+
},
|
| 1055 |
+
"dec_blocks.0.skip.weight": {
|
| 1056 |
+
"w_norm": 15.761346817016602,
|
| 1057 |
+
"g_norm": 9.896884148474783e-05,
|
| 1058 |
+
"shape": [
|
| 1059 |
+
768,
|
| 1060 |
+
1536,
|
| 1061 |
+
1,
|
| 1062 |
+
1
|
| 1063 |
+
]
|
| 1064 |
+
},
|
| 1065 |
+
"dec_blocks.0.skip.bias": {
|
| 1066 |
+
"w_norm": 0.410925954580307,
|
| 1067 |
+
"g_norm": 1.6814665286801755e-06,
|
| 1068 |
+
"shape": [
|
| 1069 |
+
768
|
| 1070 |
+
]
|
| 1071 |
+
},
|
| 1072 |
+
"dec_blocks.1.norm1.weight": {
|
| 1073 |
+
"w_norm": 39.21905517578125,
|
| 1074 |
+
"g_norm": 6.33988861409307e-07,
|
| 1075 |
+
"shape": [
|
| 1076 |
+
1536
|
| 1077 |
+
]
|
| 1078 |
+
},
|
| 1079 |
+
"dec_blocks.1.norm1.bias": {
|
| 1080 |
+
"w_norm": 0.09087496995925903,
|
| 1081 |
+
"g_norm": 5.014872499486955e-07,
|
| 1082 |
+
"shape": [
|
| 1083 |
+
1536
|
| 1084 |
+
]
|
| 1085 |
+
},
|
| 1086 |
+
"dec_blocks.1.conv1.weight": {
|
| 1087 |
+
"w_norm": 17.78136444091797,
|
| 1088 |
+
"g_norm": 2.920708357123658e-05,
|
| 1089 |
+
"shape": [
|
| 1090 |
+
768,
|
| 1091 |
+
1536,
|
| 1092 |
+
3,
|
| 1093 |
+
3
|
| 1094 |
+
]
|
| 1095 |
+
},
|
| 1096 |
+
"dec_blocks.1.conv1.bias": {
|
| 1097 |
+
"w_norm": 0.14948569238185883,
|
| 1098 |
+
"g_norm": 4.78054118957516e-07,
|
| 1099 |
+
"shape": [
|
| 1100 |
+
768
|
| 1101 |
+
]
|
| 1102 |
+
},
|
| 1103 |
+
"dec_blocks.1.t_proj.weight": {
|
| 1104 |
+
"w_norm": 16.056991577148438,
|
| 1105 |
+
"g_norm": 4.6630319161522493e-07,
|
| 1106 |
+
"shape": [
|
| 1107 |
+
768,
|
| 1108 |
+
256
|
| 1109 |
+
]
|
| 1110 |
+
},
|
| 1111 |
+
"dec_blocks.1.t_proj.bias": {
|
| 1112 |
+
"w_norm": 0.9883320927619934,
|
| 1113 |
+
"g_norm": 4.781649636242946e-07,
|
| 1114 |
+
"shape": [
|
| 1115 |
+
768
|
| 1116 |
+
]
|
| 1117 |
+
},
|
| 1118 |
+
"dec_blocks.1.norm2.weight": {
|
| 1119 |
+
"w_norm": 27.447290420532227,
|
| 1120 |
+
"g_norm": 9.171581609734858e-07,
|
| 1121 |
+
"shape": [
|
| 1122 |
+
768
|
| 1123 |
+
]
|
| 1124 |
+
},
|
| 1125 |
+
"dec_blocks.1.norm2.bias": {
|
| 1126 |
+
"w_norm": 0.3388512134552002,
|
| 1127 |
+
"g_norm": 2.0488243990257615e-06,
|
| 1128 |
+
"shape": [
|
| 1129 |
+
768
|
| 1130 |
+
]
|
| 1131 |
+
},
|
| 1132 |
+
"dec_blocks.1.conv2.weight": {
|
| 1133 |
+
"w_norm": 16.2231502532959,
|
| 1134 |
+
"g_norm": 8.360083302250132e-05,
|
| 1135 |
+
"shape": [
|
| 1136 |
+
768,
|
| 1137 |
+
768,
|
| 1138 |
+
3,
|
| 1139 |
+
3
|
| 1140 |
+
]
|
| 1141 |
+
},
|
| 1142 |
+
"dec_blocks.1.conv2.bias": {
|
| 1143 |
+
"w_norm": 0.19709675014019012,
|
| 1144 |
+
"g_norm": 5.03153023601044e-06,
|
| 1145 |
+
"shape": [
|
| 1146 |
+
768
|
| 1147 |
+
]
|
| 1148 |
+
},
|
| 1149 |
+
"dec_blocks.1.skip.weight": {
|
| 1150 |
+
"w_norm": 15.74288272857666,
|
| 1151 |
+
"g_norm": 0.00020004258840344846,
|
| 1152 |
+
"shape": [
|
| 1153 |
+
768,
|
| 1154 |
+
1536,
|
| 1155 |
+
1,
|
| 1156 |
+
1
|
| 1157 |
+
]
|
| 1158 |
+
},
|
| 1159 |
+
"dec_blocks.1.skip.bias": {
|
| 1160 |
+
"w_norm": 0.4062535762786865,
|
| 1161 |
+
"g_norm": 5.03153023601044e-06,
|
| 1162 |
+
"shape": [
|
| 1163 |
+
768
|
| 1164 |
+
]
|
| 1165 |
+
},
|
| 1166 |
+
"dec_blocks.2.norm1.weight": {
|
| 1167 |
+
"w_norm": 33.958805084228516,
|
| 1168 |
+
"g_norm": 1.8269259953740402e-06,
|
| 1169 |
+
"shape": [
|
| 1170 |
+
1152
|
| 1171 |
+
]
|
| 1172 |
+
},
|
| 1173 |
+
"dec_blocks.2.norm1.bias": {
|
| 1174 |
+
"w_norm": 0.09479564428329468,
|
| 1175 |
+
"g_norm": 1.4436129731620895e-06,
|
| 1176 |
+
"shape": [
|
| 1177 |
+
1152
|
| 1178 |
+
]
|
| 1179 |
+
},
|
| 1180 |
+
"dec_blocks.2.conv1.weight": {
|
| 1181 |
+
"w_norm": 12.608702659606934,
|
| 1182 |
+
"g_norm": 8.076793164946139e-05,
|
| 1183 |
+
"shape": [
|
| 1184 |
+
384,
|
| 1185 |
+
1152,
|
| 1186 |
+
3,
|
| 1187 |
+
3
|
| 1188 |
+
]
|
| 1189 |
+
},
|
| 1190 |
+
"dec_blocks.2.conv1.bias": {
|
| 1191 |
+
"w_norm": 0.1214609444141388,
|
| 1192 |
+
"g_norm": 1.3795322502119234e-06,
|
| 1193 |
+
"shape": [
|
| 1194 |
+
384
|
| 1195 |
+
]
|
| 1196 |
+
},
|
| 1197 |
+
"dec_blocks.2.t_proj.weight": {
|
| 1198 |
+
"w_norm": 11.346697807312012,
|
| 1199 |
+
"g_norm": 1.288618136641162e-06,
|
| 1200 |
+
"shape": [
|
| 1201 |
+
384,
|
| 1202 |
+
256
|
| 1203 |
+
]
|
| 1204 |
+
},
|
| 1205 |
+
"dec_blocks.2.t_proj.bias": {
|
| 1206 |
+
"w_norm": 0.7093044519424438,
|
| 1207 |
+
"g_norm": 1.3801052318740403e-06,
|
| 1208 |
+
"shape": [
|
| 1209 |
+
384
|
| 1210 |
+
]
|
| 1211 |
+
},
|
| 1212 |
+
"dec_blocks.2.norm2.weight": {
|
| 1213 |
+
"w_norm": 19.435365676879883,
|
| 1214 |
+
"g_norm": 3.010250793522573e-06,
|
| 1215 |
+
"shape": [
|
| 1216 |
+
384
|
| 1217 |
+
]
|
| 1218 |
+
},
|
| 1219 |
+
"dec_blocks.2.norm2.bias": {
|
| 1220 |
+
"w_norm": 0.22179819643497467,
|
| 1221 |
+
"g_norm": 5.038673407398164e-06,
|
| 1222 |
+
"shape": [
|
| 1223 |
+
384
|
| 1224 |
+
]
|
| 1225 |
+
},
|
| 1226 |
+
"dec_blocks.2.conv2.weight": {
|
| 1227 |
+
"w_norm": 11.456557273864746,
|
| 1228 |
+
"g_norm": 0.00020378570479806513,
|
| 1229 |
+
"shape": [
|
| 1230 |
+
384,
|
| 1231 |
+
384,
|
| 1232 |
+
3,
|
| 1233 |
+
3
|
| 1234 |
+
]
|
| 1235 |
+
},
|
| 1236 |
+
"dec_blocks.2.conv2.bias": {
|
| 1237 |
+
"w_norm": 0.1886361539363861,
|
| 1238 |
+
"g_norm": 1.2030825928377453e-05,
|
| 1239 |
+
"shape": [
|
| 1240 |
+
384
|
| 1241 |
+
]
|
| 1242 |
+
},
|
| 1243 |
+
"dec_blocks.2.skip.weight": {
|
| 1244 |
+
"w_norm": 10.977109909057617,
|
| 1245 |
+
"g_norm": 0.0004246227617841214,
|
| 1246 |
+
"shape": [
|
| 1247 |
+
384,
|
| 1248 |
+
1152,
|
| 1249 |
+
1,
|
| 1250 |
+
1
|
| 1251 |
+
]
|
| 1252 |
+
},
|
| 1253 |
+
"dec_blocks.2.skip.bias": {
|
| 1254 |
+
"w_norm": 0.331463485956192,
|
| 1255 |
+
"g_norm": 1.2030825928377453e-05,
|
| 1256 |
+
"shape": [
|
| 1257 |
+
384
|
| 1258 |
+
]
|
| 1259 |
+
},
|
| 1260 |
+
"dec_blocks.3.norm1.weight": {
|
| 1261 |
+
"w_norm": 27.710662841796875,
|
| 1262 |
+
"g_norm": 1.555925700813532e-05,
|
| 1263 |
+
"shape": [
|
| 1264 |
+
768
|
| 1265 |
+
]
|
| 1266 |
+
},
|
| 1267 |
+
"dec_blocks.3.norm1.bias": {
|
| 1268 |
+
"w_norm": 0.08933005481958389,
|
| 1269 |
+
"g_norm": 1.3325834515853785e-05,
|
| 1270 |
+
"shape": [
|
| 1271 |
+
768
|
| 1272 |
+
]
|
| 1273 |
+
},
|
| 1274 |
+
"dec_blocks.3.conv1.weight": {
|
| 1275 |
+
"w_norm": 12.44627571105957,
|
| 1276 |
+
"g_norm": 0.0004514533211477101,
|
| 1277 |
+
"shape": [
|
| 1278 |
+
384,
|
| 1279 |
+
768,
|
| 1280 |
+
3,
|
| 1281 |
+
3
|
| 1282 |
+
]
|
| 1283 |
+
},
|
| 1284 |
+
"dec_blocks.3.conv1.bias": {
|
| 1285 |
+
"w_norm": 0.14480382204055786,
|
| 1286 |
+
"g_norm": 8.666344001539983e-06,
|
| 1287 |
+
"shape": [
|
| 1288 |
+
384
|
| 1289 |
+
]
|
| 1290 |
+
},
|
| 1291 |
+
"dec_blocks.3.t_proj.weight": {
|
| 1292 |
+
"w_norm": 11.391429901123047,
|
| 1293 |
+
"g_norm": 9.20170441531809e-06,
|
| 1294 |
+
"shape": [
|
| 1295 |
+
384,
|
| 1296 |
+
256
|
| 1297 |
+
]
|
| 1298 |
+
},
|
| 1299 |
+
"dec_blocks.3.t_proj.bias": {
|
| 1300 |
+
"w_norm": 0.7092675566673279,
|
| 1301 |
+
"g_norm": 8.667771908221766e-06,
|
| 1302 |
+
"shape": [
|
| 1303 |
+
384
|
| 1304 |
+
]
|
| 1305 |
+
},
|
| 1306 |
+
"dec_blocks.3.norm2.weight": {
|
| 1307 |
+
"w_norm": 19.412303924560547,
|
| 1308 |
+
"g_norm": 2.933637188107241e-05,
|
| 1309 |
+
"shape": [
|
| 1310 |
+
384
|
| 1311 |
+
]
|
| 1312 |
+
},
|
| 1313 |
+
"dec_blocks.3.norm2.bias": {
|
| 1314 |
+
"w_norm": 0.22528940439224243,
|
| 1315 |
+
"g_norm": 2.264487739012111e-05,
|
| 1316 |
+
"shape": [
|
| 1317 |
+
384
|
| 1318 |
+
]
|
| 1319 |
+
},
|
| 1320 |
+
"dec_blocks.3.conv2.weight": {
|
| 1321 |
+
"w_norm": 11.52460765838623,
|
| 1322 |
+
"g_norm": 0.0008822944364510477,
|
| 1323 |
+
"shape": [
|
| 1324 |
+
384,
|
| 1325 |
+
384,
|
| 1326 |
+
3,
|
| 1327 |
+
3
|
| 1328 |
+
]
|
| 1329 |
+
},
|
| 1330 |
+
"dec_blocks.3.conv2.bias": {
|
| 1331 |
+
"w_norm": 0.1936788111925125,
|
| 1332 |
+
"g_norm": 4.2916828533634543e-05,
|
| 1333 |
+
"shape": [
|
| 1334 |
+
384
|
| 1335 |
+
]
|
| 1336 |
+
},
|
| 1337 |
+
"dec_blocks.3.skip.weight": {
|
| 1338 |
+
"w_norm": 11.064810752868652,
|
| 1339 |
+
"g_norm": 0.0010949639836326241,
|
| 1340 |
+
"shape": [
|
| 1341 |
+
384,
|
| 1342 |
+
768,
|
| 1343 |
+
1,
|
| 1344 |
+
1
|
| 1345 |
+
]
|
| 1346 |
+
},
|
| 1347 |
+
"dec_blocks.3.skip.bias": {
|
| 1348 |
+
"w_norm": 0.4060087203979492,
|
| 1349 |
+
"g_norm": 4.2916828533634543e-05,
|
| 1350 |
+
"shape": [
|
| 1351 |
+
384
|
| 1352 |
+
]
|
| 1353 |
+
},
|
| 1354 |
+
"dec_blocks.4.norm1.weight": {
|
| 1355 |
+
"w_norm": 23.974384307861328,
|
| 1356 |
+
"g_norm": 0.00010767428466351703,
|
| 1357 |
+
"shape": [
|
| 1358 |
+
576
|
| 1359 |
+
]
|
| 1360 |
+
},
|
| 1361 |
+
"dec_blocks.4.norm1.bias": {
|
| 1362 |
+
"w_norm": 0.09818897396326065,
|
| 1363 |
+
"g_norm": 0.00010226306039839983,
|
| 1364 |
+
"shape": [
|
| 1365 |
+
576
|
| 1366 |
+
]
|
| 1367 |
+
},
|
| 1368 |
+
"dec_blocks.4.conv1.weight": {
|
| 1369 |
+
"w_norm": 8.616304397583008,
|
| 1370 |
+
"g_norm": 0.002126651583239436,
|
| 1371 |
+
"shape": [
|
| 1372 |
+
192,
|
| 1373 |
+
576,
|
| 1374 |
+
3,
|
| 1375 |
+
3
|
| 1376 |
+
]
|
| 1377 |
+
},
|
| 1378 |
+
"dec_blocks.4.conv1.bias": {
|
| 1379 |
+
"w_norm": 0.11683345586061478,
|
| 1380 |
+
"g_norm": 6.0171176301082596e-05,
|
| 1381 |
+
"shape": [
|
| 1382 |
+
192
|
| 1383 |
+
]
|
| 1384 |
+
},
|
| 1385 |
+
"dec_blocks.4.t_proj.weight": {
|
| 1386 |
+
"w_norm": 8.06219482421875,
|
| 1387 |
+
"g_norm": 6.534672138513997e-05,
|
| 1388 |
+
"shape": [
|
| 1389 |
+
192,
|
| 1390 |
+
256
|
| 1391 |
+
]
|
| 1392 |
+
},
|
| 1393 |
+
"dec_blocks.4.t_proj.bias": {
|
| 1394 |
+
"w_norm": 0.4816136360168457,
|
| 1395 |
+
"g_norm": 6.018929343554191e-05,
|
| 1396 |
+
"shape": [
|
| 1397 |
+
192
|
| 1398 |
+
]
|
| 1399 |
+
},
|
| 1400 |
+
"dec_blocks.4.norm2.weight": {
|
| 1401 |
+
"w_norm": 13.791584968566895,
|
| 1402 |
+
"g_norm": 0.00013399690215010196,
|
| 1403 |
+
"shape": [
|
| 1404 |
+
192
|
| 1405 |
+
]
|
| 1406 |
+
},
|
| 1407 |
+
"dec_blocks.4.norm2.bias": {
|
| 1408 |
+
"w_norm": 0.08628630638122559,
|
| 1409 |
+
"g_norm": 0.00014492537593469024,
|
| 1410 |
+
"shape": [
|
| 1411 |
+
192
|
| 1412 |
+
]
|
| 1413 |
+
},
|
| 1414 |
+
"dec_blocks.4.conv2.weight": {
|
| 1415 |
+
"w_norm": 8.11715030670166,
|
| 1416 |
+
"g_norm": 0.00321268942207098,
|
| 1417 |
+
"shape": [
|
| 1418 |
+
192,
|
| 1419 |
+
192,
|
| 1420 |
+
3,
|
| 1421 |
+
3
|
| 1422 |
+
]
|
| 1423 |
+
},
|
| 1424 |
+
"dec_blocks.4.conv2.bias": {
|
| 1425 |
+
"w_norm": 0.19586454331874847,
|
| 1426 |
+
"g_norm": 0.0001909145648824051,
|
| 1427 |
+
"shape": [
|
| 1428 |
+
192
|
| 1429 |
+
]
|
| 1430 |
+
},
|
| 1431 |
+
"dec_blocks.4.skip.weight": {
|
| 1432 |
+
"w_norm": 7.841552734375,
|
| 1433 |
+
"g_norm": 0.004247691016644239,
|
| 1434 |
+
"shape": [
|
| 1435 |
+
192,
|
| 1436 |
+
576,
|
| 1437 |
+
1,
|
| 1438 |
+
1
|
| 1439 |
+
]
|
| 1440 |
+
},
|
| 1441 |
+
"dec_blocks.4.skip.bias": {
|
| 1442 |
+
"w_norm": 0.3465918302536011,
|
| 1443 |
+
"g_norm": 0.0001909145648824051,
|
| 1444 |
+
"shape": [
|
| 1445 |
+
192
|
| 1446 |
+
]
|
| 1447 |
+
},
|
| 1448 |
+
"dec_blocks.5.norm1.weight": {
|
| 1449 |
+
"w_norm": 19.53300666809082,
|
| 1450 |
+
"g_norm": 0.0025150158908218145,
|
| 1451 |
+
"shape": [
|
| 1452 |
+
384
|
| 1453 |
+
]
|
| 1454 |
+
},
|
| 1455 |
+
"dec_blocks.5.norm1.bias": {
|
| 1456 |
+
"w_norm": 0.12533682584762573,
|
| 1457 |
+
"g_norm": 0.0028404968325048685,
|
| 1458 |
+
"shape": [
|
| 1459 |
+
384
|
| 1460 |
+
]
|
| 1461 |
+
},
|
| 1462 |
+
"dec_blocks.5.conv1.weight": {
|
| 1463 |
+
"w_norm": 8.668478965759277,
|
| 1464 |
+
"g_norm": 0.028973735868930817,
|
| 1465 |
+
"shape": [
|
| 1466 |
+
192,
|
| 1467 |
+
384,
|
| 1468 |
+
3,
|
| 1469 |
+
3
|
| 1470 |
+
]
|
| 1471 |
+
},
|
| 1472 |
+
"dec_blocks.5.conv1.bias": {
|
| 1473 |
+
"w_norm": 0.13875038921833038,
|
| 1474 |
+
"g_norm": 0.0014434006297960877,
|
| 1475 |
+
"shape": [
|
| 1476 |
+
192
|
| 1477 |
+
]
|
| 1478 |
+
},
|
| 1479 |
+
"dec_blocks.5.t_proj.weight": {
|
| 1480 |
+
"w_norm": 8.21015453338623,
|
| 1481 |
+
"g_norm": 0.0008499801042489707,
|
| 1482 |
+
"shape": [
|
| 1483 |
+
192,
|
| 1484 |
+
256
|
| 1485 |
+
]
|
| 1486 |
+
},
|
| 1487 |
+
"dec_blocks.5.t_proj.bias": {
|
| 1488 |
+
"w_norm": 0.5054141283035278,
|
| 1489 |
+
"g_norm": 0.001443629153072834,
|
| 1490 |
+
"shape": [
|
| 1491 |
+
192
|
| 1492 |
+
]
|
| 1493 |
+
},
|
| 1494 |
+
"dec_blocks.5.norm2.weight": {
|
| 1495 |
+
"w_norm": 13.897296905517578,
|
| 1496 |
+
"g_norm": 0.002029021503403783,
|
| 1497 |
+
"shape": [
|
| 1498 |
+
192
|
| 1499 |
+
]
|
| 1500 |
+
},
|
| 1501 |
+
"dec_blocks.5.norm2.bias": {
|
| 1502 |
+
"w_norm": 0.045003172010183334,
|
| 1503 |
+
"g_norm": 0.0019661581609398127,
|
| 1504 |
+
"shape": [
|
| 1505 |
+
192
|
| 1506 |
+
]
|
| 1507 |
+
},
|
| 1508 |
+
"dec_blocks.5.conv2.weight": {
|
| 1509 |
+
"w_norm": 8.435464859008789,
|
| 1510 |
+
"g_norm": 0.029161185026168823,
|
| 1511 |
+
"shape": [
|
| 1512 |
+
192,
|
| 1513 |
+
192,
|
| 1514 |
+
3,
|
| 1515 |
+
3
|
| 1516 |
+
]
|
| 1517 |
+
},
|
| 1518 |
+
"dec_blocks.5.conv2.bias": {
|
| 1519 |
+
"w_norm": 0.20200084149837494,
|
| 1520 |
+
"g_norm": 0.0014317930908873677,
|
| 1521 |
+
"shape": [
|
| 1522 |
+
192
|
| 1523 |
+
]
|
| 1524 |
+
},
|
| 1525 |
+
"dec_blocks.5.skip.weight": {
|
| 1526 |
+
"w_norm": 8.107450485229492,
|
| 1527 |
+
"g_norm": 0.022503562271595,
|
| 1528 |
+
"shape": [
|
| 1529 |
+
192,
|
| 1530 |
+
384,
|
| 1531 |
+
1,
|
| 1532 |
+
1
|
| 1533 |
+
]
|
| 1534 |
+
},
|
| 1535 |
+
"dec_blocks.5.skip.bias": {
|
| 1536 |
+
"w_norm": 0.4229964017868042,
|
| 1537 |
+
"g_norm": 0.0014317930908873677,
|
| 1538 |
+
"shape": [
|
| 1539 |
+
192
|
| 1540 |
+
]
|
| 1541 |
+
},
|
| 1542 |
+
"dec_attns.0.norm.weight": {
|
| 1543 |
+
"w_norm": 27.606098175048828,
|
| 1544 |
+
"g_norm": 1.4698603081342299e-06,
|
| 1545 |
+
"shape": [
|
| 1546 |
+
768
|
| 1547 |
+
]
|
| 1548 |
+
},
|
| 1549 |
+
"dec_attns.0.norm.bias": {
|
| 1550 |
+
"w_norm": 0.05232677236199379,
|
| 1551 |
+
"g_norm": 1.6636099644529168e-06,
|
| 1552 |
+
"shape": [
|
| 1553 |
+
768
|
| 1554 |
+
]
|
| 1555 |
+
},
|
| 1556 |
+
"dec_attns.0.qkv.weight": {
|
| 1557 |
+
"w_norm": 28.047964096069336,
|
| 1558 |
+
"g_norm": 3.253455724916421e-05,
|
| 1559 |
+
"shape": [
|
| 1560 |
+
2304,
|
| 1561 |
+
768,
|
| 1562 |
+
1,
|
| 1563 |
+
1
|
| 1564 |
+
]
|
| 1565 |
+
},
|
| 1566 |
+
"dec_attns.0.qkv.bias": {
|
| 1567 |
+
"w_norm": 1.007122278213501,
|
| 1568 |
+
"g_norm": 1.4771649148315191e-06,
|
| 1569 |
+
"shape": [
|
| 1570 |
+
2304
|
| 1571 |
+
]
|
| 1572 |
+
},
|
| 1573 |
+
"dec_attns.0.proj.weight": {
|
| 1574 |
+
"w_norm": 16.108665466308594,
|
| 1575 |
+
"g_norm": 2.8333268346614204e-05,
|
| 1576 |
+
"shape": [
|
| 1577 |
+
768,
|
| 1578 |
+
768,
|
| 1579 |
+
1,
|
| 1580 |
+
1
|
| 1581 |
+
]
|
| 1582 |
+
},
|
| 1583 |
+
"dec_attns.0.proj.bias": {
|
| 1584 |
+
"w_norm": 0.5780074596405029,
|
| 1585 |
+
"g_norm": 1.8256989733345108e-06,
|
| 1586 |
+
"shape": [
|
| 1587 |
+
768
|
| 1588 |
+
]
|
| 1589 |
+
},
|
| 1590 |
+
"dec_attns.1.norm.weight": {
|
| 1591 |
+
"w_norm": 27.601665496826172,
|
| 1592 |
+
"g_norm": 2.4182666038541356e-06,
|
| 1593 |
+
"shape": [
|
| 1594 |
+
768
|
| 1595 |
+
]
|
| 1596 |
+
},
|
| 1597 |
+
"dec_attns.1.norm.bias": {
|
| 1598 |
+
"w_norm": 0.049121566116809845,
|
| 1599 |
+
"g_norm": 2.8192550871608546e-06,
|
| 1600 |
+
"shape": [
|
| 1601 |
+
768
|
| 1602 |
+
]
|
| 1603 |
+
},
|
| 1604 |
+
"dec_attns.1.qkv.weight": {
|
| 1605 |
+
"w_norm": 28.07448387145996,
|
| 1606 |
+
"g_norm": 7.991576421773061e-05,
|
| 1607 |
+
"shape": [
|
| 1608 |
+
2304,
|
| 1609 |
+
768,
|
| 1610 |
+
1,
|
| 1611 |
+
1
|
| 1612 |
+
]
|
| 1613 |
+
},
|
| 1614 |
+
"dec_attns.1.qkv.bias": {
|
| 1615 |
+
"w_norm": 1.0039130449295044,
|
| 1616 |
+
"g_norm": 3.495277724141488e-06,
|
| 1617 |
+
"shape": [
|
| 1618 |
+
2304
|
| 1619 |
+
]
|
| 1620 |
+
},
|
| 1621 |
+
"dec_attns.1.proj.weight": {
|
| 1622 |
+
"w_norm": 16.064023971557617,
|
| 1623 |
+
"g_norm": 6.958303129067644e-05,
|
| 1624 |
+
"shape": [
|
| 1625 |
+
768,
|
| 1626 |
+
768,
|
| 1627 |
+
1,
|
| 1628 |
+
1
|
| 1629 |
+
]
|
| 1630 |
+
},
|
| 1631 |
+
"dec_attns.1.proj.bias": {
|
| 1632 |
+
"w_norm": 0.5789670944213867,
|
| 1633 |
+
"g_norm": 5.107773631607415e-06,
|
| 1634 |
+
"shape": [
|
| 1635 |
+
768
|
| 1636 |
+
]
|
| 1637 |
+
},
|
| 1638 |
+
"dec_attns.2.norm.weight": {
|
| 1639 |
+
"w_norm": 19.495100021362305,
|
| 1640 |
+
"g_norm": 3.4630834306881297e-06,
|
| 1641 |
+
"shape": [
|
| 1642 |
+
384
|
| 1643 |
+
]
|
| 1644 |
+
},
|
| 1645 |
+
"dec_attns.2.norm.bias": {
|
| 1646 |
+
"w_norm": 0.026043370366096497,
|
| 1647 |
+
"g_norm": 5.35019989911234e-06,
|
| 1648 |
+
"shape": [
|
| 1649 |
+
384
|
| 1650 |
+
]
|
| 1651 |
+
},
|
| 1652 |
+
"dec_attns.2.qkv.weight": {
|
| 1653 |
+
"w_norm": 19.710468292236328,
|
| 1654 |
+
"g_norm": 0.00010689659393392503,
|
| 1655 |
+
"shape": [
|
| 1656 |
+
1152,
|
| 1657 |
+
384,
|
| 1658 |
+
1,
|
| 1659 |
+
1
|
| 1660 |
+
]
|
| 1661 |
+
},
|
| 1662 |
+
"dec_attns.2.qkv.bias": {
|
| 1663 |
+
"w_norm": 0.98268061876297,
|
| 1664 |
+
"g_norm": 7.844309038773645e-06,
|
| 1665 |
+
"shape": [
|
| 1666 |
+
1152
|
| 1667 |
+
]
|
| 1668 |
+
},
|
| 1669 |
+
"dec_attns.2.proj.weight": {
|
| 1670 |
+
"w_norm": 11.305863380432129,
|
| 1671 |
+
"g_norm": 0.00010106317495228723,
|
| 1672 |
+
"shape": [
|
| 1673 |
+
384,
|
| 1674 |
+
384,
|
| 1675 |
+
1,
|
| 1676 |
+
1
|
| 1677 |
+
]
|
| 1678 |
+
},
|
| 1679 |
+
"dec_attns.2.proj.bias": {
|
| 1680 |
+
"w_norm": 0.5879626274108887,
|
| 1681 |
+
"g_norm": 1.2567536941787694e-05,
|
| 1682 |
+
"shape": [
|
| 1683 |
+
384
|
| 1684 |
+
]
|
| 1685 |
+
},
|
| 1686 |
+
"dec_ups.0.op.weight": {
|
| 1687 |
+
"w_norm": 16.159915924072266,
|
| 1688 |
+
"g_norm": 0.00017089844914153218,
|
| 1689 |
+
"shape": [
|
| 1690 |
+
768,
|
| 1691 |
+
768,
|
| 1692 |
+
3,
|
| 1693 |
+
3
|
| 1694 |
+
]
|
| 1695 |
+
},
|
| 1696 |
+
"dec_ups.0.op.bias": {
|
| 1697 |
+
"w_norm": 0.20129258930683136,
|
| 1698 |
+
"g_norm": 2.0403990674822126e-06,
|
| 1699 |
+
"shape": [
|
| 1700 |
+
768
|
| 1701 |
+
]
|
| 1702 |
+
},
|
| 1703 |
+
"dec_ups.1.op.weight": {
|
| 1704 |
+
"w_norm": 16.166606903076172,
|
| 1705 |
+
"g_norm": 0.00038573183701373637,
|
| 1706 |
+
"shape": [
|
| 1707 |
+
768,
|
| 1708 |
+
768,
|
| 1709 |
+
3,
|
| 1710 |
+
3
|
| 1711 |
+
]
|
| 1712 |
+
},
|
| 1713 |
+
"dec_ups.1.op.bias": {
|
| 1714 |
+
"w_norm": 0.1936124861240387,
|
| 1715 |
+
"g_norm": 5.565851552091772e-06,
|
| 1716 |
+
"shape": [
|
| 1717 |
+
768
|
| 1718 |
+
]
|
| 1719 |
+
},
|
| 1720 |
+
"dec_ups.2.op.weight": {
|
| 1721 |
+
"w_norm": 11.31502914428711,
|
| 1722 |
+
"g_norm": 0.0009087788057513535,
|
| 1723 |
+
"shape": [
|
| 1724 |
+
384,
|
| 1725 |
+
384,
|
| 1726 |
+
3,
|
| 1727 |
+
3
|
| 1728 |
+
]
|
| 1729 |
+
},
|
| 1730 |
+
"dec_ups.2.op.bias": {
|
| 1731 |
+
"w_norm": 0.197477325797081,
|
| 1732 |
+
"g_norm": 1.888297265395522e-05,
|
| 1733 |
+
"shape": [
|
| 1734 |
+
384
|
| 1735 |
+
]
|
| 1736 |
+
},
|
| 1737 |
+
"dec_ups.3.op.weight": {
|
| 1738 |
+
"w_norm": 11.352828979492188,
|
| 1739 |
+
"g_norm": 0.002423783065751195,
|
| 1740 |
+
"shape": [
|
| 1741 |
+
384,
|
| 1742 |
+
384,
|
| 1743 |
+
3,
|
| 1744 |
+
3
|
| 1745 |
+
]
|
| 1746 |
+
},
|
| 1747 |
+
"dec_ups.3.op.bias": {
|
| 1748 |
+
"w_norm": 0.19884906709194183,
|
| 1749 |
+
"g_norm": 6.680266233161092e-05,
|
| 1750 |
+
"shape": [
|
| 1751 |
+
384
|
| 1752 |
+
]
|
| 1753 |
+
},
|
| 1754 |
+
"dec_ups.4.op.weight": {
|
| 1755 |
+
"w_norm": 8.010747909545898,
|
| 1756 |
+
"g_norm": 0.010360135696828365,
|
| 1757 |
+
"shape": [
|
| 1758 |
+
192,
|
| 1759 |
+
192,
|
| 1760 |
+
3,
|
| 1761 |
+
3
|
| 1762 |
+
]
|
| 1763 |
+
},
|
| 1764 |
+
"dec_ups.4.op.bias": {
|
| 1765 |
+
"w_norm": 0.19764398038387299,
|
| 1766 |
+
"g_norm": 0.00024076507543213665,
|
| 1767 |
+
"shape": [
|
| 1768 |
+
192
|
| 1769 |
+
]
|
| 1770 |
+
},
|
| 1771 |
+
"out_norm.weight": {
|
| 1772 |
+
"w_norm": 13.761808395385742,
|
| 1773 |
+
"g_norm": 0.003240334102883935,
|
| 1774 |
+
"shape": [
|
| 1775 |
+
192
|
| 1776 |
+
]
|
| 1777 |
+
},
|
| 1778 |
+
"out_norm.bias": {
|
| 1779 |
+
"w_norm": 0.11472446471452713,
|
| 1780 |
+
"g_norm": 0.004786268807947636,
|
| 1781 |
+
"shape": [
|
| 1782 |
+
192
|
| 1783 |
+
]
|
| 1784 |
+
},
|
| 1785 |
+
"out_conv.weight": {
|
| 1786 |
+
"w_norm": 0.15211158990859985,
|
| 1787 |
+
"g_norm": 0.5555833578109741,
|
| 1788 |
+
"shape": [
|
| 1789 |
+
3,
|
| 1790 |
+
192,
|
| 1791 |
+
3,
|
| 1792 |
+
3
|
| 1793 |
+
]
|
| 1794 |
+
},
|
| 1795 |
+
"out_conv.bias": {
|
| 1796 |
+
"w_norm": 0.009061155840754509,
|
| 1797 |
+
"g_norm": 0.03287523612380028,
|
| 1798 |
+
"shape": [
|
| 1799 |
+
3
|
| 1800 |
+
]
|
| 1801 |
+
}
|
| 1802 |
+
},
|
| 1803 |
+
"activations": {
|
| 1804 |
+
"input_conv": {
|
| 1805 |
+
"mean": 0.00040025782072916627,
|
| 1806 |
+
"std": 0.3693671226501465,
|
| 1807 |
+
"abs_max": 2.671875,
|
| 1808 |
+
"frac_dead": 0.00047539008664898574,
|
| 1809 |
+
"frac_sat": 0.0
|
| 1810 |
+
},
|
| 1811 |
+
"time_mlp.1": {
|
| 1812 |
+
"mean": 0.010746731422841549,
|
| 1813 |
+
"std": 0.2820868194103241,
|
| 1814 |
+
"abs_max": 4.25,
|
| 1815 |
+
"frac_dead": 0.0,
|
| 1816 |
+
"frac_sat": 0.0
|
| 1817 |
+
},
|
| 1818 |
+
"time_mlp.3": {
|
| 1819 |
+
"mean": 0.02136988565325737,
|
| 1820 |
+
"std": 0.21643811464309692,
|
| 1821 |
+
"abs_max": 1.0625,
|
| 1822 |
+
"frac_dead": 0.00016276042151730508,
|
| 1823 |
+
"frac_sat": 0.0
|
| 1824 |
+
},
|
| 1825 |
+
"enc_blocks.0.norm1": {
|
| 1826 |
+
"mean": -0.006252565421164036,
|
| 1827 |
+
"std": 0.9951425790786743,
|
| 1828 |
+
"abs_max": 6.74277925491333,
|
| 1829 |
+
"frac_dead": 4.76837158203125e-07,
|
| 1830 |
+
"frac_sat": 0.0
|
| 1831 |
+
},
|
| 1832 |
+
"enc_blocks.0.conv1": {
|
| 1833 |
+
"mean": -0.041000474244356155,
|
| 1834 |
+
"std": 0.6137309670448303,
|
| 1835 |
+
"abs_max": 10.75,
|
| 1836 |
+
"frac_dead": 6.946259236428887e-05,
|
| 1837 |
+
"frac_sat": 1.0265244299034748e-07
|
| 1838 |
+
},
|
| 1839 |
+
"enc_blocks.0.t_proj": {
|
| 1840 |
+
"mean": 0.0007395787979476154,
|
| 1841 |
+
"std": 0.2251749038696289,
|
| 1842 |
+
"abs_max": 2.40625,
|
| 1843 |
+
"frac_dead": 0.0,
|
| 1844 |
+
"frac_sat": 0.0
|
| 1845 |
+
},
|
| 1846 |
+
"enc_blocks.0.norm2": {
|
| 1847 |
+
"mean": -0.004515793640166521,
|
| 1848 |
+
"std": 0.9990665316581726,
|
| 1849 |
+
"abs_max": 12.43277359008789,
|
| 1850 |
+
"frac_dead": 1.3841523696100921e-06,
|
| 1851 |
+
"frac_sat": 4.801485147254425e-07
|
| 1852 |
+
},
|
| 1853 |
+
"enc_blocks.0.conv2": {
|
| 1854 |
+
"mean": -0.030521171167492867,
|
| 1855 |
+
"std": 0.8776020407676697,
|
| 1856 |
+
"abs_max": 8.75,
|
| 1857 |
+
"frac_dead": 3.275275230407715e-05,
|
| 1858 |
+
"frac_sat": 0.0
|
| 1859 |
+
},
|
| 1860 |
+
"enc_downs.0.op": {
|
| 1861 |
+
"mean": -0.035549554973840714,
|
| 1862 |
+
"std": 1.359336495399475,
|
| 1863 |
+
"abs_max": 13.75,
|
| 1864 |
+
"frac_dead": 2.3351774871116504e-05,
|
| 1865 |
+
"frac_sat": 3.378921246621758e-05
|
| 1866 |
+
},
|
| 1867 |
+
"enc_blocks.1.norm1": {
|
| 1868 |
+
"mean": -0.0025818953290581703,
|
| 1869 |
+
"std": 0.9994575381278992,
|
| 1870 |
+
"abs_max": 9.819805145263672,
|
| 1871 |
+
"frac_dead": 9.5367431640625e-07,
|
| 1872 |
+
"frac_sat": 0.0
|
| 1873 |
+
},
|
| 1874 |
+
"enc_blocks.1.conv1": {
|
| 1875 |
+
"mean": -0.2658883333206177,
|
| 1876 |
+
"std": 1.0592257976531982,
|
| 1877 |
+
"abs_max": 12.5,
|
| 1878 |
+
"frac_dead": 2.8875139832962304e-05,
|
| 1879 |
+
"frac_sat": 1.3642841167893494e-06
|
| 1880 |
+
},
|
| 1881 |
+
"enc_blocks.1.t_proj": {
|
| 1882 |
+
"mean": 0.0059111700393259525,
|
| 1883 |
+
"std": 0.24908672273159027,
|
| 1884 |
+
"abs_max": 1.9453125,
|
| 1885 |
+
"frac_dead": 0.0,
|
| 1886 |
+
"frac_sat": 0.0
|
| 1887 |
+
},
|
| 1888 |
+
"enc_blocks.1.norm2": {
|
| 1889 |
+
"mean": -0.005134095437824726,
|
| 1890 |
+
"std": 0.9981088042259216,
|
| 1891 |
+
"abs_max": 12.5274658203125,
|
| 1892 |
+
"frac_dead": 6.22537413619284e-07,
|
| 1893 |
+
"frac_sat": 6.490283794846619e-07
|
| 1894 |
+
},
|
| 1895 |
+
"enc_blocks.1.conv2": {
|
| 1896 |
+
"mean": 0.02617700956761837,
|
| 1897 |
+
"std": 0.9424678087234497,
|
| 1898 |
+
"abs_max": 10.875,
|
| 1899 |
+
"frac_dead": 4.083580643055029e-05,
|
| 1900 |
+
"frac_sat": 2.2517311037972831e-07
|
| 1901 |
+
},
|
| 1902 |
+
"enc_downs.1.op": {
|
| 1903 |
+
"mean": -0.0690908432006836,
|
| 1904 |
+
"std": 1.943811058998108,
|
| 1905 |
+
"abs_max": 14.0625,
|
| 1906 |
+
"frac_dead": 1.811981201171875e-05,
|
| 1907 |
+
"frac_sat": 4.725986218545586e-05
|
| 1908 |
+
},
|
| 1909 |
+
"enc_blocks.2.norm1": {
|
| 1910 |
+
"mean": -0.0037061066832393408,
|
| 1911 |
+
"std": 0.9978165626525879,
|
| 1912 |
+
"abs_max": 10.55722713470459,
|
| 1913 |
+
"frac_dead": 6.35782896551973e-07,
|
| 1914 |
+
"frac_sat": 5.298190686175985e-08
|
| 1915 |
+
},
|
| 1916 |
+
"enc_blocks.2.conv1": {
|
| 1917 |
+
"mean": -0.13071851432323456,
|
| 1918 |
+
"std": 1.3800336122512817,
|
| 1919 |
+
"abs_max": 22.375,
|
| 1920 |
+
"frac_dead": 3.655751424957998e-05,
|
| 1921 |
+
"frac_sat": 0.00047132704639807343
|
| 1922 |
+
},
|
| 1923 |
+
"enc_blocks.2.t_proj": {
|
| 1924 |
+
"mean": -0.0055412352085113525,
|
| 1925 |
+
"std": 0.332256555557251,
|
| 1926 |
+
"abs_max": 3.015625,
|
| 1927 |
+
"frac_dead": 0.0,
|
| 1928 |
+
"frac_sat": 0.0
|
| 1929 |
+
},
|
| 1930 |
+
"enc_blocks.2.norm2": {
|
| 1931 |
+
"mean": -0.011643648147583008,
|
| 1932 |
+
"std": 0.9939536452293396,
|
| 1933 |
+
"abs_max": 12.937939643859863,
|
| 1934 |
+
"frac_dead": 1.2185838613731903e-06,
|
| 1935 |
+
"frac_sat": 1.3589859008789062e-05
|
| 1936 |
+
},
|
| 1937 |
+
"enc_blocks.2.conv2": {
|
| 1938 |
+
"mean": -0.0342453196644783,
|
| 1939 |
+
"std": 0.7672107219696045,
|
| 1940 |
+
"abs_max": 8.8125,
|
| 1941 |
+
"frac_dead": 2.961688551295083e-05,
|
| 1942 |
+
"frac_sat": 0.0
|
| 1943 |
+
},
|
| 1944 |
+
"enc_blocks.2.skip": {
|
| 1945 |
+
"mean": -0.037515006959438324,
|
| 1946 |
+
"std": 1.0088363885879517,
|
| 1947 |
+
"abs_max": 9.625,
|
| 1948 |
+
"frac_dead": 0.00010739432764239609,
|
| 1949 |
+
"frac_sat": 0.0
|
| 1950 |
+
},
|
| 1951 |
+
"enc_downs.2.op": {
|
| 1952 |
+
"mean": -0.199848935008049,
|
| 1953 |
+
"std": 1.5087478160858154,
|
| 1954 |
+
"abs_max": 11.75,
|
| 1955 |
+
"frac_dead": 1.4623005881730933e-05,
|
| 1956 |
+
"frac_sat": 2.1192763597355224e-06
|
| 1957 |
+
},
|
| 1958 |
+
"enc_blocks.3.norm1": {
|
| 1959 |
+
"mean": -0.00175758870318532,
|
| 1960 |
+
"std": 0.99953293800354,
|
| 1961 |
+
"abs_max": 8.614314079284668,
|
| 1962 |
+
"frac_dead": 1.0596381798677612e-06,
|
| 1963 |
+
"frac_sat": 0.0
|
| 1964 |
+
},
|
| 1965 |
+
"enc_blocks.3.conv1": {
|
| 1966 |
+
"mean": -0.10498303920030594,
|
| 1967 |
+
"std": 1.746254563331604,
|
| 1968 |
+
"abs_max": 30.75,
|
| 1969 |
+
"frac_dead": 2.1616617232211865e-05,
|
| 1970 |
+
"frac_sat": 0.002836121479049325
|
| 1971 |
+
},
|
| 1972 |
+
"enc_blocks.3.t_proj": {
|
| 1973 |
+
"mean": 0.002910944866016507,
|
| 1974 |
+
"std": 0.14892221987247467,
|
| 1975 |
+
"abs_max": 1.1875,
|
| 1976 |
+
"frac_dead": 0.0,
|
| 1977 |
+
"frac_sat": 0.0
|
| 1978 |
+
},
|
| 1979 |
+
"enc_blocks.3.norm2": {
|
| 1980 |
+
"mean": -0.01072887796908617,
|
| 1981 |
+
"std": 0.9917898774147034,
|
| 1982 |
+
"abs_max": 16.629051208496094,
|
| 1983 |
+
"frac_dead": 3.602769766075653e-06,
|
| 1984 |
+
"frac_sat": 6.114112329669297e-05
|
| 1985 |
+
},
|
| 1986 |
+
"enc_blocks.3.conv2": {
|
| 1987 |
+
"mean": -0.02169557847082615,
|
| 1988 |
+
"std": 0.7376501560211182,
|
| 1989 |
+
"abs_max": 8.375,
|
| 1990 |
+
"frac_dead": 3.549787652445957e-05,
|
| 1991 |
+
"frac_sat": 0.0
|
| 1992 |
+
},
|
| 1993 |
+
"enc_attns.3.norm": {
|
| 1994 |
+
"mean": 0.0005119277630001307,
|
| 1995 |
+
"std": 0.9960955381393433,
|
| 1996 |
+
"abs_max": 8.947876930236816,
|
| 1997 |
+
"frac_dead": 6.35782896551973e-07,
|
| 1998 |
+
"frac_sat": 0.0
|
| 1999 |
+
},
|
| 2000 |
+
"enc_attns.3.qkv": {
|
| 2001 |
+
"mean": -0.00030917784897610545,
|
| 2002 |
+
"std": 0.6075136661529541,
|
| 2003 |
+
"abs_max": 5.34375,
|
| 2004 |
+
"frac_dead": 0.00010931933502433822,
|
| 2005 |
+
"frac_sat": 0.0
|
| 2006 |
+
},
|
| 2007 |
+
"enc_attns.3.proj": {
|
| 2008 |
+
"mean": 0.007649092935025692,
|
| 2009 |
+
"std": 0.26445072889328003,
|
| 2010 |
+
"abs_max": 1.3046875,
|
| 2011 |
+
"frac_dead": 0.000232696533203125,
|
| 2012 |
+
"frac_sat": 0.0
|
| 2013 |
+
},
|
| 2014 |
+
"enc_downs.3.op": {
|
| 2015 |
+
"mean": -0.23308219015598297,
|
| 2016 |
+
"std": 1.8356003761291504,
|
| 2017 |
+
"abs_max": 16.875,
|
| 2018 |
+
"frac_dead": 1.5682644516346045e-05,
|
| 2019 |
+
"frac_sat": 0.00038316514110192657
|
| 2020 |
+
},
|
| 2021 |
+
"enc_blocks.4.norm1": {
|
| 2022 |
+
"mean": -0.0006192077999003232,
|
| 2023 |
+
"std": 1.0010056495666504,
|
| 2024 |
+
"abs_max": 11.152945518493652,
|
| 2025 |
+
"frac_dead": 8.477105097881577e-07,
|
| 2026 |
+
"frac_sat": 4.238552548940788e-07
|
| 2027 |
+
},
|
| 2028 |
+
"enc_blocks.4.conv1": {
|
| 2029 |
+
"mean": -0.2928019165992737,
|
| 2030 |
+
"std": 1.740490198135376,
|
| 2031 |
+
"abs_max": 31.125,
|
| 2032 |
+
"frac_dead": 2.8822158128605224e-05,
|
| 2033 |
+
"frac_sat": 0.0039316811598837376
|
| 2034 |
+
},
|
| 2035 |
+
"enc_blocks.4.t_proj": {
|
| 2036 |
+
"mean": -0.0017857596976682544,
|
| 2037 |
+
"std": 0.10373267531394958,
|
| 2038 |
+
"abs_max": 0.76171875,
|
| 2039 |
+
"frac_dead": 5.425347262644209e-05,
|
| 2040 |
+
"frac_sat": 0.0
|
| 2041 |
+
},
|
| 2042 |
+
"enc_blocks.4.norm2": {
|
| 2043 |
+
"mean": -0.011213724501430988,
|
| 2044 |
+
"std": 0.9912196397781372,
|
| 2045 |
+
"abs_max": 16.70511245727539,
|
| 2046 |
+
"frac_dead": 8.477105097881577e-07,
|
| 2047 |
+
"frac_sat": 0.00014156765246298164
|
| 2048 |
+
},
|
| 2049 |
+
"enc_blocks.4.conv2": {
|
| 2050 |
+
"mean": 0.031458597630262375,
|
| 2051 |
+
"std": 0.4912275969982147,
|
| 2052 |
+
"abs_max": 3.84375,
|
| 2053 |
+
"frac_dead": 3.369649130036123e-05,
|
| 2054 |
+
"frac_sat": 0.0
|
| 2055 |
+
},
|
| 2056 |
+
"enc_blocks.4.skip": {
|
| 2057 |
+
"mean": -0.03792019560933113,
|
| 2058 |
+
"std": 0.958321213722229,
|
| 2059 |
+
"abs_max": 7.15625,
|
| 2060 |
+
"frac_dead": 8.138021075865254e-05,
|
| 2061 |
+
"frac_sat": 0.0
|
| 2062 |
+
},
|
| 2063 |
+
"enc_attns.4.norm": {
|
| 2064 |
+
"mean": 9.616216266294941e-05,
|
| 2065 |
+
"std": 0.9960263967514038,
|
| 2066 |
+
"abs_max": 8.730345726013184,
|
| 2067 |
+
"frac_dead": 6.35782896551973e-07,
|
| 2068 |
+
"frac_sat": 0.0
|
| 2069 |
+
},
|
| 2070 |
+
"enc_attns.4.qkv": {
|
| 2071 |
+
"mean": -0.006408622022718191,
|
| 2072 |
+
"std": 0.685590386390686,
|
| 2073 |
+
"abs_max": 6.5,
|
| 2074 |
+
"frac_dead": 6.696912896586582e-05,
|
| 2075 |
+
"frac_sat": 0.0
|
| 2076 |
+
},
|
| 2077 |
+
"enc_attns.4.proj": {
|
| 2078 |
+
"mean": 0.003732851007953286,
|
| 2079 |
+
"std": 0.4339083433151245,
|
| 2080 |
+
"abs_max": 3.515625,
|
| 2081 |
+
"frac_dead": 0.00012736850476358086,
|
| 2082 |
+
"frac_sat": 0.0
|
| 2083 |
+
},
|
| 2084 |
+
"enc_downs.4.op": {
|
| 2085 |
+
"mean": 0.0982859805226326,
|
| 2086 |
+
"std": 1.6599680185317993,
|
| 2087 |
+
"abs_max": 12.25,
|
| 2088 |
+
"frac_dead": 1.4411079064302612e-05,
|
| 2089 |
+
"frac_sat": 7.968478894326836e-05
|
| 2090 |
+
},
|
| 2091 |
+
"enc_blocks.5.norm1": {
|
| 2092 |
+
"mean": 2.0242399841663428e-05,
|
| 2093 |
+
"std": 1.0001006126403809,
|
| 2094 |
+
"abs_max": 5.786524295806885,
|
| 2095 |
+
"frac_dead": 8.477105097881577e-07,
|
| 2096 |
+
"frac_sat": 0.0
|
| 2097 |
+
},
|
| 2098 |
+
"enc_blocks.5.conv1": {
|
| 2099 |
+
"mean": -0.39854559302330017,
|
| 2100 |
+
"std": 1.6837373971939087,
|
| 2101 |
+
"abs_max": 14.8125,
|
| 2102 |
+
"frac_dead": 1.610649997019209e-05,
|
| 2103 |
+
"frac_sat": 0.0010392931289970875
|
| 2104 |
+
},
|
| 2105 |
+
"enc_blocks.5.t_proj": {
|
| 2106 |
+
"mean": -0.0011938775423914194,
|
| 2107 |
+
"std": 0.09263734519481659,
|
| 2108 |
+
"abs_max": 0.77734375,
|
| 2109 |
+
"frac_dead": 0.0,
|
| 2110 |
+
"frac_sat": 0.0
|
| 2111 |
+
},
|
| 2112 |
+
"enc_blocks.5.norm2": {
|
| 2113 |
+
"mean": -0.008912770077586174,
|
| 2114 |
+
"std": 0.9928245544433594,
|
| 2115 |
+
"abs_max": 8.65130615234375,
|
| 2116 |
+
"frac_dead": 1.6954210195763153e-06,
|
| 2117 |
+
"frac_sat": 0.0
|
| 2118 |
+
},
|
| 2119 |
+
"enc_blocks.5.conv2": {
|
| 2120 |
+
"mean": 0.19011344015598297,
|
| 2121 |
+
"std": 1.201659917831421,
|
| 2122 |
+
"abs_max": 7.6875,
|
| 2123 |
+
"frac_dead": 7.62939453125e-06,
|
| 2124 |
+
"frac_sat": 0.0
|
| 2125 |
+
},
|
| 2126 |
+
"enc_attns.5.norm": {
|
| 2127 |
+
"mean": 0.00032190897036343813,
|
| 2128 |
+
"std": 0.9978728294372559,
|
| 2129 |
+
"abs_max": 5.934600353240967,
|
| 2130 |
+
"frac_dead": 0.0,
|
| 2131 |
+
"frac_sat": 0.0
|
| 2132 |
+
},
|
| 2133 |
+
"enc_attns.5.qkv": {
|
| 2134 |
+
"mean": -0.015497934073209763,
|
| 2135 |
+
"std": 0.8438277840614319,
|
| 2136 |
+
"abs_max": 5.46875,
|
| 2137 |
+
"frac_dead": 5.849202352692373e-05,
|
| 2138 |
+
"frac_sat": 0.0
|
| 2139 |
+
},
|
| 2140 |
+
"enc_attns.5.proj": {
|
| 2141 |
+
"mean": -0.0038925441913306713,
|
| 2142 |
+
"std": 0.32418742775917053,
|
| 2143 |
+
"abs_max": 2.453125,
|
| 2144 |
+
"frac_dead": 0.00017038981604855508,
|
| 2145 |
+
"frac_sat": 0.0
|
| 2146 |
+
},
|
| 2147 |
+
"mid1.norm1": {
|
| 2148 |
+
"mean": 0.0001421187917003408,
|
| 2149 |
+
"std": 1.0004252195358276,
|
| 2150 |
+
"abs_max": 5.862927436828613,
|
| 2151 |
+
"frac_dead": 0.0,
|
| 2152 |
+
"frac_sat": 0.0
|
| 2153 |
+
},
|
| 2154 |
+
"mid1.conv1": {
|
| 2155 |
+
"mean": -0.8414751291275024,
|
| 2156 |
+
"std": 2.068591833114624,
|
| 2157 |
+
"abs_max": 20.375,
|
| 2158 |
+
"frac_dead": 2.1192763597355224e-05,
|
| 2159 |
+
"frac_sat": 0.0095986258238554
|
| 2160 |
+
},
|
| 2161 |
+
"mid1.t_proj": {
|
| 2162 |
+
"mean": 0.0026207775808870792,
|
| 2163 |
+
"std": 0.09048116207122803,
|
| 2164 |
+
"abs_max": 0.62109375,
|
| 2165 |
+
"frac_dead": 0.0,
|
| 2166 |
+
"frac_sat": 0.0
|
| 2167 |
+
},
|
| 2168 |
+
"mid1.norm2": {
|
| 2169 |
+
"mean": -0.009563598781824112,
|
| 2170 |
+
"std": 0.994515061378479,
|
| 2171 |
+
"abs_max": 8.810402870178223,
|
| 2172 |
+
"frac_dead": 8.477105097881577e-07,
|
| 2173 |
+
"frac_sat": 0.0
|
| 2174 |
+
},
|
| 2175 |
+
"mid1.conv2": {
|
| 2176 |
+
"mean": 0.041060809046030045,
|
| 2177 |
+
"std": 0.7921715974807739,
|
| 2178 |
+
"abs_max": 4.96875,
|
| 2179 |
+
"frac_dead": 2.8822158128605224e-05,
|
| 2180 |
+
"frac_sat": 0.0
|
| 2181 |
+
},
|
| 2182 |
+
"mid_attn.norm": {
|
| 2183 |
+
"mean": 0.0004211680789012462,
|
| 2184 |
+
"std": 0.9976292252540588,
|
| 2185 |
+
"abs_max": 7.214486122131348,
|
| 2186 |
+
"frac_dead": 8.477105097881577e-07,
|
| 2187 |
+
"frac_sat": 0.0
|
| 2188 |
+
},
|
| 2189 |
+
"mid_attn.qkv": {
|
| 2190 |
+
"mean": 0.005006986204534769,
|
| 2191 |
+
"std": 0.7217532992362976,
|
| 2192 |
+
"abs_max": 4.625,
|
| 2193 |
+
"frac_dead": 6.0470018070191145e-05,
|
| 2194 |
+
"frac_sat": 0.0
|
| 2195 |
+
},
|
| 2196 |
+
"mid_attn.proj": {
|
| 2197 |
+
"mean": -0.010449005290865898,
|
| 2198 |
+
"std": 0.3259406089782715,
|
| 2199 |
+
"abs_max": 2.15625,
|
| 2200 |
+
"frac_dead": 0.00012885199976153672,
|
| 2201 |
+
"frac_sat": 0.0
|
| 2202 |
+
},
|
| 2203 |
+
"mid2.norm1": {
|
| 2204 |
+
"mean": 0.000711227476131171,
|
| 2205 |
+
"std": 1.0010067224502563,
|
| 2206 |
+
"abs_max": 7.27166748046875,
|
| 2207 |
+
"frac_dead": 0.0,
|
| 2208 |
+
"frac_sat": 0.0
|
| 2209 |
+
},
|
| 2210 |
+
"mid2.conv1": {
|
| 2211 |
+
"mean": -0.9361047744750977,
|
| 2212 |
+
"std": 2.149503707885742,
|
| 2213 |
+
"abs_max": 21.375,
|
| 2214 |
+
"frac_dead": 2.6279025405528955e-05,
|
| 2215 |
+
"frac_sat": 0.014512804336845875
|
| 2216 |
+
},
|
| 2217 |
+
"mid2.t_proj": {
|
| 2218 |
+
"mean": 0.0011170649668201804,
|
| 2219 |
+
"std": 0.09452700614929199,
|
| 2220 |
+
"abs_max": 0.7109375,
|
| 2221 |
+
"frac_dead": 0.0,
|
| 2222 |
+
"frac_sat": 0.0
|
| 2223 |
+
},
|
| 2224 |
+
"mid2.norm2": {
|
| 2225 |
+
"mean": -0.010762469843029976,
|
| 2226 |
+
"std": 0.9965609312057495,
|
| 2227 |
+
"abs_max": 9.152743339538574,
|
| 2228 |
+
"frac_dead": 1.6954210195763153e-06,
|
| 2229 |
+
"frac_sat": 0.0
|
| 2230 |
+
},
|
| 2231 |
+
"mid2.conv2": {
|
| 2232 |
+
"mean": 0.007028906140476465,
|
| 2233 |
+
"std": 0.4812912940979004,
|
| 2234 |
+
"abs_max": 2.875,
|
| 2235 |
+
"frac_dead": 2.7126736313221045e-05,
|
| 2236 |
+
"frac_sat": 0.0
|
| 2237 |
+
},
|
| 2238 |
+
"dec_blocks.0.norm1": {
|
| 2239 |
+
"mean": 0.0011261552572250366,
|
| 2240 |
+
"std": 1.001523494720459,
|
| 2241 |
+
"abs_max": 6.811408519744873,
|
| 2242 |
+
"frac_dead": 4.238552548940788e-07,
|
| 2243 |
+
"frac_sat": 0.0
|
| 2244 |
+
},
|
| 2245 |
+
"dec_blocks.0.conv1": {
|
| 2246 |
+
"mean": -2.7249886989593506,
|
| 2247 |
+
"std": 4.642963886260986,
|
| 2248 |
+
"abs_max": 51.25,
|
| 2249 |
+
"frac_dead": 4.238552719471045e-06,
|
| 2250 |
+
"frac_sat": 0.06366220861673355
|
| 2251 |
+
},
|
| 2252 |
+
"dec_blocks.0.t_proj": {
|
| 2253 |
+
"mean": -0.0014680986059829593,
|
| 2254 |
+
"std": 0.09673027694225311,
|
| 2255 |
+
"abs_max": 0.6171875,
|
| 2256 |
+
"frac_dead": 0.0,
|
| 2257 |
+
"frac_sat": 0.0
|
| 2258 |
+
},
|
| 2259 |
+
"dec_blocks.0.norm2": {
|
| 2260 |
+
"mean": -0.01321282796561718,
|
| 2261 |
+
"std": 0.9973823428153992,
|
| 2262 |
+
"abs_max": 9.541091918945312,
|
| 2263 |
+
"frac_dead": 8.477105097881577e-07,
|
| 2264 |
+
"frac_sat": 0.0
|
| 2265 |
+
},
|
| 2266 |
+
"dec_blocks.0.conv2": {
|
| 2267 |
+
"mean": -0.00643331790342927,
|
| 2268 |
+
"std": 0.37738314270973206,
|
| 2269 |
+
"abs_max": 2.484375,
|
| 2270 |
+
"frac_dead": 3.899468356394209e-05,
|
| 2271 |
+
"frac_sat": 0.0
|
| 2272 |
+
},
|
| 2273 |
+
"dec_blocks.0.skip": {
|
| 2274 |
+
"mean": 0.005534242372959852,
|
| 2275 |
+
"std": 1.2259471416473389,
|
| 2276 |
+
"abs_max": 7.625,
|
| 2277 |
+
"frac_dead": 2.5431314497836865e-05,
|
| 2278 |
+
"frac_sat": 0.0
|
| 2279 |
+
},
|
| 2280 |
+
"dec_attns.0.norm": {
|
| 2281 |
+
"mean": 7.251939678099006e-05,
|
| 2282 |
+
"std": 0.9961729049682617,
|
| 2283 |
+
"abs_max": 4.991007328033447,
|
| 2284 |
+
"frac_dead": 0.0,
|
| 2285 |
+
"frac_sat": 0.0
|
| 2286 |
+
},
|
| 2287 |
+
"dec_attns.0.qkv": {
|
| 2288 |
+
"mean": -0.032485075294971466,
|
| 2289 |
+
"std": 0.6778285503387451,
|
| 2290 |
+
"abs_max": 4.375,
|
| 2291 |
+
"frac_dead": 6.668655987596139e-05,
|
| 2292 |
+
"frac_sat": 0.0
|
| 2293 |
+
},
|
| 2294 |
+
"dec_attns.0.proj": {
|
| 2295 |
+
"mean": -0.00447096349671483,
|
| 2296 |
+
"std": 0.5198463797569275,
|
| 2297 |
+
"abs_max": 2.59375,
|
| 2298 |
+
"frac_dead": 8.053249621298164e-05,
|
| 2299 |
+
"frac_sat": 0.0
|
| 2300 |
+
},
|
| 2301 |
+
"dec_ups.0.op": {
|
| 2302 |
+
"mean": 0.11986494064331055,
|
| 2303 |
+
"std": 1.3657621145248413,
|
| 2304 |
+
"abs_max": 7.71875,
|
| 2305 |
+
"frac_dead": 1.229180270456709e-05,
|
| 2306 |
+
"frac_sat": 0.0
|
| 2307 |
+
},
|
| 2308 |
+
"dec_blocks.1.norm1": {
|
| 2309 |
+
"mean": 0.0005290090339258313,
|
| 2310 |
+
"std": 1.000602126121521,
|
| 2311 |
+
"abs_max": 7.9642744064331055,
|
| 2312 |
+
"frac_dead": 9.5367431640625e-07,
|
| 2313 |
+
"frac_sat": 0.0
|
| 2314 |
+
},
|
| 2315 |
+
"dec_blocks.1.conv1": {
|
| 2316 |
+
"mean": -2.2336854934692383,
|
| 2317 |
+
"std": 3.7236015796661377,
|
| 2318 |
+
"abs_max": 44.25,
|
| 2319 |
+
"frac_dead": 6.145901352283545e-06,
|
| 2320 |
+
"frac_sat": 0.04314698278903961
|
| 2321 |
+
},
|
| 2322 |
+
"dec_blocks.1.t_proj": {
|
| 2323 |
+
"mean": -0.0013965911930426955,
|
| 2324 |
+
"std": 0.09607820212841034,
|
| 2325 |
+
"abs_max": 0.73046875,
|
| 2326 |
+
"frac_dead": 0.0,
|
| 2327 |
+
"frac_sat": 0.0
|
| 2328 |
+
},
|
| 2329 |
+
"dec_blocks.1.norm2": {
|
| 2330 |
+
"mean": -0.014699429273605347,
|
| 2331 |
+
"std": 0.9981132745742798,
|
| 2332 |
+
"abs_max": 9.811440467834473,
|
| 2333 |
+
"frac_dead": 1.4834934063401306e-06,
|
| 2334 |
+
"frac_sat": 0.0
|
| 2335 |
+
},
|
| 2336 |
+
"dec_blocks.1.conv2": {
|
| 2337 |
+
"mean": -0.00016551630687899888,
|
| 2338 |
+
"std": 0.3925391137599945,
|
| 2339 |
+
"abs_max": 2.421875,
|
| 2340 |
+
"frac_dead": 3.729926174855791e-05,
|
| 2341 |
+
"frac_sat": 0.0
|
| 2342 |
+
},
|
| 2343 |
+
"dec_blocks.1.skip": {
|
| 2344 |
+
"mean": -0.011327608488500118,
|
| 2345 |
+
"std": 0.8946093916893005,
|
| 2346 |
+
"abs_max": 5.21875,
|
| 2347 |
+
"frac_dead": 4.217359673930332e-05,
|
| 2348 |
+
"frac_sat": 0.0
|
| 2349 |
+
},
|
| 2350 |
+
"dec_attns.1.norm": {
|
| 2351 |
+
"mean": 7.759490108583122e-05,
|
| 2352 |
+
"std": 0.9962604641914368,
|
| 2353 |
+
"abs_max": 6.363803863525391,
|
| 2354 |
+
"frac_dead": 4.238552548940788e-07,
|
| 2355 |
+
"frac_sat": 0.0
|
| 2356 |
+
},
|
| 2357 |
+
"dec_attns.1.qkv": {
|
| 2358 |
+
"mean": 0.01140259113162756,
|
| 2359 |
+
"std": 0.6308833956718445,
|
| 2360 |
+
"abs_max": 4.125,
|
| 2361 |
+
"frac_dead": 7.212603668449447e-05,
|
| 2362 |
+
"frac_sat": 0.0
|
| 2363 |
+
},
|
| 2364 |
+
"dec_attns.1.proj": {
|
| 2365 |
+
"mean": 0.009470382705330849,
|
| 2366 |
+
"std": 0.3589651584625244,
|
| 2367 |
+
"abs_max": 1.8203125,
|
| 2368 |
+
"frac_dead": 0.0001373291015625,
|
| 2369 |
+
"frac_sat": 0.0
|
| 2370 |
+
},
|
| 2371 |
+
"dec_ups.1.op": {
|
| 2372 |
+
"mean": 0.05434713140130043,
|
| 2373 |
+
"std": 0.9426736235618591,
|
| 2374 |
+
"abs_max": 5.375,
|
| 2375 |
+
"frac_dead": 1.7589993149158545e-05,
|
| 2376 |
+
"frac_sat": 0.0
|
| 2377 |
+
},
|
| 2378 |
+
"dec_blocks.2.norm1": {
|
| 2379 |
+
"mean": -0.00016650006000418216,
|
| 2380 |
+
"std": 1.0008327960968018,
|
| 2381 |
+
"abs_max": 9.631314277648926,
|
| 2382 |
+
"frac_dead": 7.417467031700653e-07,
|
| 2383 |
+
"frac_sat": 0.0
|
| 2384 |
+
},
|
| 2385 |
+
"dec_blocks.2.conv1": {
|
| 2386 |
+
"mean": -1.0471075773239136,
|
| 2387 |
+
"std": 3.5208213329315186,
|
| 2388 |
+
"abs_max": 40.25,
|
| 2389 |
+
"frac_dead": 9.112887710216455e-06,
|
| 2390 |
+
"frac_sat": 0.041029613465070724
|
| 2391 |
+
},
|
| 2392 |
+
"dec_blocks.2.t_proj": {
|
| 2393 |
+
"mean": -0.004079055972397327,
|
| 2394 |
+
"std": 0.10816413164138794,
|
| 2395 |
+
"abs_max": 0.6796875,
|
| 2396 |
+
"frac_dead": 0.0,
|
| 2397 |
+
"frac_sat": 0.0
|
| 2398 |
+
},
|
| 2399 |
+
"dec_blocks.2.norm2": {
|
| 2400 |
+
"mean": -0.014768723398447037,
|
| 2401 |
+
"std": 1.0005854368209839,
|
| 2402 |
+
"abs_max": 9.347208976745605,
|
| 2403 |
+
"frac_dead": 1.1656019296424347e-06,
|
| 2404 |
+
"frac_sat": 0.0
|
| 2405 |
+
},
|
| 2406 |
+
"dec_blocks.2.conv2": {
|
| 2407 |
+
"mean": 0.023853028193116188,
|
| 2408 |
+
"std": 0.3805305063724518,
|
| 2409 |
+
"abs_max": 2.25,
|
| 2410 |
+
"frac_dead": 5.361768853617832e-05,
|
| 2411 |
+
"frac_sat": 0.0
|
| 2412 |
+
},
|
| 2413 |
+
"dec_blocks.2.skip": {
|
| 2414 |
+
"mean": 0.032520074397325516,
|
| 2415 |
+
"std": 0.8015388250350952,
|
| 2416 |
+
"abs_max": 6.1875,
|
| 2417 |
+
"frac_dead": 4.355112832854502e-05,
|
| 2418 |
+
"frac_sat": 0.0
|
| 2419 |
+
},
|
| 2420 |
+
"dec_attns.2.norm": {
|
| 2421 |
+
"mean": -0.00016280759882647544,
|
| 2422 |
+
"std": 0.9950010776519775,
|
| 2423 |
+
"abs_max": 8.573103904724121,
|
| 2424 |
+
"frac_dead": 7.417467031700653e-07,
|
| 2425 |
+
"frac_sat": 0.0
|
| 2426 |
+
},
|
| 2427 |
+
"dec_attns.2.qkv": {
|
| 2428 |
+
"mean": 0.005614753346890211,
|
| 2429 |
+
"std": 0.6067090034484863,
|
| 2430 |
+
"abs_max": 6.34375,
|
| 2431 |
+
"frac_dead": 0.0001014780136756599,
|
| 2432 |
+
"frac_sat": 0.0
|
| 2433 |
+
},
|
| 2434 |
+
"dec_attns.2.proj": {
|
| 2435 |
+
"mean": 0.00559053011238575,
|
| 2436 |
+
"std": 0.301848441362381,
|
| 2437 |
+
"abs_max": 1.984375,
|
| 2438 |
+
"frac_dead": 0.0001836352894315496,
|
| 2439 |
+
"frac_sat": 0.0
|
| 2440 |
+
},
|
| 2441 |
+
"dec_ups.2.op": {
|
| 2442 |
+
"mean": 0.00250375852920115,
|
| 2443 |
+
"std": 0.7678435444831848,
|
| 2444 |
+
"abs_max": 8.5625,
|
| 2445 |
+
"frac_dead": 2.8689702958217822e-05,
|
| 2446 |
+
"frac_sat": 0.0
|
| 2447 |
+
},
|
| 2448 |
+
"dec_blocks.3.norm1": {
|
| 2449 |
+
"mean": -0.0012076408602297306,
|
| 2450 |
+
"std": 1.0004695653915405,
|
| 2451 |
+
"abs_max": 10.6675443649292,
|
| 2452 |
+
"frac_dead": 5.695554818885284e-07,
|
| 2453 |
+
"frac_sat": 5.298190686175985e-08
|
| 2454 |
+
},
|
| 2455 |
+
"dec_blocks.3.conv1": {
|
| 2456 |
+
"mean": -0.9750791192054749,
|
| 2457 |
+
"std": 2.384197950363159,
|
| 2458 |
+
"abs_max": 42.25,
|
| 2459 |
+
"frac_dead": 1.3007058441871777e-05,
|
| 2460 |
+
"frac_sat": 0.009697622619569302
|
| 2461 |
+
},
|
| 2462 |
+
"dec_blocks.3.t_proj": {
|
| 2463 |
+
"mean": -0.000648355926387012,
|
| 2464 |
+
"std": 0.13717307150363922,
|
| 2465 |
+
"abs_max": 0.94140625,
|
| 2466 |
+
"frac_dead": 0.0,
|
| 2467 |
+
"frac_sat": 0.0
|
| 2468 |
+
},
|
| 2469 |
+
"dec_blocks.3.norm2": {
|
| 2470 |
+
"mean": -0.014358947984874249,
|
| 2471 |
+
"std": 0.9960097074508667,
|
| 2472 |
+
"abs_max": 12.777533531188965,
|
| 2473 |
+
"frac_dead": 6.35782896551973e-07,
|
| 2474 |
+
"frac_sat": 2.2596783310291357e-05
|
| 2475 |
+
},
|
| 2476 |
+
"dec_blocks.3.conv2": {
|
| 2477 |
+
"mean": 0.018594268709421158,
|
| 2478 |
+
"std": 0.43747326731681824,
|
| 2479 |
+
"abs_max": 6.71875,
|
| 2480 |
+
"frac_dead": 5.581643927143887e-05,
|
| 2481 |
+
"frac_sat": 0.0
|
| 2482 |
+
},
|
| 2483 |
+
"dec_blocks.3.skip": {
|
| 2484 |
+
"mean": -6.380594277288765e-05,
|
| 2485 |
+
"std": 0.6293055415153503,
|
| 2486 |
+
"abs_max": 6.375,
|
| 2487 |
+
"frac_dead": 7.343292236328125e-05,
|
| 2488 |
+
"frac_sat": 0.0
|
| 2489 |
+
},
|
| 2490 |
+
"dec_ups.3.op": {
|
| 2491 |
+
"mean": -0.05254688486456871,
|
| 2492 |
+
"std": 0.7604173421859741,
|
| 2493 |
+
"abs_max": 10.375,
|
| 2494 |
+
"frac_dead": 3.1656691135140136e-05,
|
| 2495 |
+
"frac_sat": 1.3245476715439963e-08
|
| 2496 |
+
},
|
| 2497 |
+
"dec_blocks.4.norm1": {
|
| 2498 |
+
"mean": -0.0019422242185100913,
|
| 2499 |
+
"std": 0.9994417428970337,
|
| 2500 |
+
"abs_max": 11.856742858886719,
|
| 2501 |
+
"frac_dead": 4.76837158203125e-07,
|
| 2502 |
+
"frac_sat": 2.0751247120642802e-07
|
| 2503 |
+
},
|
| 2504 |
+
"dec_blocks.4.conv1": {
|
| 2505 |
+
"mean": -0.7406669855117798,
|
| 2506 |
+
"std": 1.986746072769165,
|
| 2507 |
+
"abs_max": 25.25,
|
| 2508 |
+
"frac_dead": 1.1139445632579736e-05,
|
| 2509 |
+
"frac_sat": 0.0016398695297539234
|
| 2510 |
+
},
|
| 2511 |
+
"dec_blocks.4.t_proj": {
|
| 2512 |
+
"mean": -0.004709389992058277,
|
| 2513 |
+
"std": 0.15576112270355225,
|
| 2514 |
+
"abs_max": 0.96484375,
|
| 2515 |
+
"frac_dead": 0.0,
|
| 2516 |
+
"frac_sat": 0.0
|
| 2517 |
+
},
|
| 2518 |
+
"dec_blocks.4.norm2": {
|
| 2519 |
+
"mean": -0.007681197486817837,
|
| 2520 |
+
"std": 0.9972299933433533,
|
| 2521 |
+
"abs_max": 12.993465423583984,
|
| 2522 |
+
"frac_dead": 1.284811219193216e-06,
|
| 2523 |
+
"frac_sat": 2.211994569734088e-06
|
| 2524 |
+
},
|
| 2525 |
+
"dec_blocks.4.conv2": {
|
| 2526 |
+
"mean": -0.046417709439992905,
|
| 2527 |
+
"std": 0.5319167971611023,
|
| 2528 |
+
"abs_max": 5.875,
|
| 2529 |
+
"frac_dead": 6.185637903399765e-05,
|
| 2530 |
+
"frac_sat": 0.0
|
| 2531 |
+
},
|
| 2532 |
+
"dec_blocks.4.skip": {
|
| 2533 |
+
"mean": -0.02933773770928383,
|
| 2534 |
+
"std": 0.7752041816711426,
|
| 2535 |
+
"abs_max": 7.90625,
|
| 2536 |
+
"frac_dead": 7.099575304891914e-05,
|
| 2537 |
+
"frac_sat": 0.0
|
| 2538 |
+
},
|
| 2539 |
+
"dec_ups.4.op": {
|
| 2540 |
+
"mean": -0.261995792388916,
|
| 2541 |
+
"std": 1.4503177404403687,
|
| 2542 |
+
"abs_max": 26.75,
|
| 2543 |
+
"frac_dead": 3.314018249511719e-05,
|
| 2544 |
+
"frac_sat": 0.000735077599529177
|
| 2545 |
+
},
|
| 2546 |
+
"dec_blocks.5.norm1": {
|
| 2547 |
+
"mean": -0.005145529750734568,
|
| 2548 |
+
"std": 0.997856080532074,
|
| 2549 |
+
"abs_max": 13.626863479614258,
|
| 2550 |
+
"frac_dead": 8.940696716308594e-07,
|
| 2551 |
+
"frac_sat": 1.8460883666193695e-06
|
| 2552 |
+
},
|
| 2553 |
+
"dec_blocks.5.conv1": {
|
| 2554 |
+
"mean": -0.1504458636045456,
|
| 2555 |
+
"std": 1.501868724822998,
|
| 2556 |
+
"abs_max": 17.125,
|
| 2557 |
+
"frac_dead": 1.5503830582019873e-05,
|
| 2558 |
+
"frac_sat": 2.1060308426967822e-05
|
| 2559 |
+
},
|
| 2560 |
+
"dec_blocks.5.t_proj": {
|
| 2561 |
+
"mean": 0.012189735658466816,
|
| 2562 |
+
"std": 0.24086928367614746,
|
| 2563 |
+
"abs_max": 1.859375,
|
| 2564 |
+
"frac_dead": 0.0,
|
| 2565 |
+
"frac_sat": 0.0
|
| 2566 |
+
},
|
| 2567 |
+
"dec_blocks.5.norm2": {
|
| 2568 |
+
"mean": -0.0016006923979148269,
|
| 2569 |
+
"std": 1.0046899318695068,
|
| 2570 |
+
"abs_max": 12.903286933898926,
|
| 2571 |
+
"frac_dead": 6.324715400296554e-07,
|
| 2572 |
+
"frac_sat": 6.920761848050461e-07
|
| 2573 |
+
},
|
| 2574 |
+
"dec_blocks.5.conv2": {
|
| 2575 |
+
"mean": -0.23159591853618622,
|
| 2576 |
+
"std": 1.682431697845459,
|
| 2577 |
+
"abs_max": 21.25,
|
| 2578 |
+
"frac_dead": 2.7279058485873975e-05,
|
| 2579 |
+
"frac_sat": 0.0008494390640407801
|
| 2580 |
+
},
|
| 2581 |
+
"dec_blocks.5.skip": {
|
| 2582 |
+
"mean": -0.16567134857177734,
|
| 2583 |
+
"std": 1.1772466897964478,
|
| 2584 |
+
"abs_max": 12.4375,
|
| 2585 |
+
"frac_dead": 6.788306927774101e-05,
|
| 2586 |
+
"frac_sat": 9.238719940185547e-07
|
| 2587 |
+
},
|
| 2588 |
+
"out_norm": {
|
| 2589 |
+
"mean": -0.008237053640186787,
|
| 2590 |
+
"std": 0.9936524033546448,
|
| 2591 |
+
"abs_max": 11.3441743850708,
|
| 2592 |
+
"frac_dead": 1.417266048520105e-06,
|
| 2593 |
+
"frac_sat": 3.8743019104003906e-07
|
| 2594 |
+
},
|
| 2595 |
+
"out_conv": {
|
| 2596 |
+
"mean": 0.16196849942207336,
|
| 2597 |
+
"std": 0.25798922777175903,
|
| 2598 |
+
"abs_max": 0.9609375,
|
| 2599 |
+
"frac_dead": 1.4834933608653955e-05,
|
| 2600 |
+
"frac_sat": 0.0
|
| 2601 |
+
}
|
| 2602 |
+
}
|
| 2603 |
+
}
|
diagnostics/per_layer_iter3000.json
ADDED
|
@@ -0,0 +1,2603 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"iter": 3000,
|
| 3 |
+
"per_layer": {
|
| 4 |
+
"time_mlp.1.weight": {
|
| 5 |
+
"w_norm": 19.0517578125,
|
| 6 |
+
"g_norm": 0.14865441620349884,
|
| 7 |
+
"shape": [
|
| 8 |
+
1024,
|
| 9 |
+
256
|
| 10 |
+
]
|
| 11 |
+
},
|
| 12 |
+
"time_mlp.1.bias": {
|
| 13 |
+
"w_norm": 1.1601998805999756,
|
| 14 |
+
"g_norm": 0.013137980364263058,
|
| 15 |
+
"shape": [
|
| 16 |
+
1024
|
| 17 |
+
]
|
| 18 |
+
},
|
| 19 |
+
"time_mlp.3.weight": {
|
| 20 |
+
"w_norm": 9.77436351776123,
|
| 21 |
+
"g_norm": 0.022276513278484344,
|
| 22 |
+
"shape": [
|
| 23 |
+
256,
|
| 24 |
+
1024
|
| 25 |
+
]
|
| 26 |
+
},
|
| 27 |
+
"time_mlp.3.bias": {
|
| 28 |
+
"w_norm": 0.2890886664390564,
|
| 29 |
+
"g_norm": 0.011864181607961655,
|
| 30 |
+
"shape": [
|
| 31 |
+
256
|
| 32 |
+
]
|
| 33 |
+
},
|
| 34 |
+
"input_conv.weight": {
|
| 35 |
+
"w_norm": 7.956737995147705,
|
| 36 |
+
"g_norm": 0.19501574337482452,
|
| 37 |
+
"shape": [
|
| 38 |
+
192,
|
| 39 |
+
6,
|
| 40 |
+
3,
|
| 41 |
+
3
|
| 42 |
+
]
|
| 43 |
+
},
|
| 44 |
+
"input_conv.bias": {
|
| 45 |
+
"w_norm": 1.1898951530456543,
|
| 46 |
+
"g_norm": 0.19526450335979462,
|
| 47 |
+
"shape": [
|
| 48 |
+
192
|
| 49 |
+
]
|
| 50 |
+
},
|
| 51 |
+
"enc_blocks.0.norm1.weight": {
|
| 52 |
+
"w_norm": 13.789409637451172,
|
| 53 |
+
"g_norm": 0.05440613627433777,
|
| 54 |
+
"shape": [
|
| 55 |
+
192
|
| 56 |
+
]
|
| 57 |
+
},
|
| 58 |
+
"enc_blocks.0.norm1.bias": {
|
| 59 |
+
"w_norm": 0.12104981392621994,
|
| 60 |
+
"g_norm": 0.08626538515090942,
|
| 61 |
+
"shape": [
|
| 62 |
+
192
|
| 63 |
+
]
|
| 64 |
+
},
|
| 65 |
+
"enc_blocks.0.conv1.weight": {
|
| 66 |
+
"w_norm": 8.61393928527832,
|
| 67 |
+
"g_norm": 0.4161059260368347,
|
| 68 |
+
"shape": [
|
| 69 |
+
192,
|
| 70 |
+
192,
|
| 71 |
+
3,
|
| 72 |
+
3
|
| 73 |
+
]
|
| 74 |
+
},
|
| 75 |
+
"enc_blocks.0.conv1.bias": {
|
| 76 |
+
"w_norm": 0.20222757756710052,
|
| 77 |
+
"g_norm": 0.04245798662304878,
|
| 78 |
+
"shape": [
|
| 79 |
+
192
|
| 80 |
+
]
|
| 81 |
+
},
|
| 82 |
+
"enc_blocks.0.t_proj.weight": {
|
| 83 |
+
"w_norm": 8.108176231384277,
|
| 84 |
+
"g_norm": 0.04685848206281662,
|
| 85 |
+
"shape": [
|
| 86 |
+
192,
|
| 87 |
+
256
|
| 88 |
+
]
|
| 89 |
+
},
|
| 90 |
+
"enc_blocks.0.t_proj.bias": {
|
| 91 |
+
"w_norm": 0.5027391910552979,
|
| 92 |
+
"g_norm": 0.042459018528461456,
|
| 93 |
+
"shape": [
|
| 94 |
+
192
|
| 95 |
+
]
|
| 96 |
+
},
|
| 97 |
+
"enc_blocks.0.norm2.weight": {
|
| 98 |
+
"w_norm": 13.823328971862793,
|
| 99 |
+
"g_norm": 0.026325175538659096,
|
| 100 |
+
"shape": [
|
| 101 |
+
192
|
| 102 |
+
]
|
| 103 |
+
},
|
| 104 |
+
"enc_blocks.0.norm2.bias": {
|
| 105 |
+
"w_norm": 0.08441212773323059,
|
| 106 |
+
"g_norm": 0.033365774899721146,
|
| 107 |
+
"shape": [
|
| 108 |
+
192
|
| 109 |
+
]
|
| 110 |
+
},
|
| 111 |
+
"enc_blocks.0.conv2.weight": {
|
| 112 |
+
"w_norm": 8.440913200378418,
|
| 113 |
+
"g_norm": 0.1690007895231247,
|
| 114 |
+
"shape": [
|
| 115 |
+
192,
|
| 116 |
+
192,
|
| 117 |
+
3,
|
| 118 |
+
3
|
| 119 |
+
]
|
| 120 |
+
},
|
| 121 |
+
"enc_blocks.0.conv2.bias": {
|
| 122 |
+
"w_norm": 0.18897441029548645,
|
| 123 |
+
"g_norm": 0.01589149981737137,
|
| 124 |
+
"shape": [
|
| 125 |
+
192
|
| 126 |
+
]
|
| 127 |
+
},
|
| 128 |
+
"enc_blocks.1.norm1.weight": {
|
| 129 |
+
"w_norm": 13.84272575378418,
|
| 130 |
+
"g_norm": 0.0005544356536120176,
|
| 131 |
+
"shape": [
|
| 132 |
+
192
|
| 133 |
+
]
|
| 134 |
+
},
|
| 135 |
+
"enc_blocks.1.norm1.bias": {
|
| 136 |
+
"w_norm": 0.053813960403203964,
|
| 137 |
+
"g_norm": 0.000708088162355125,
|
| 138 |
+
"shape": [
|
| 139 |
+
192
|
| 140 |
+
]
|
| 141 |
+
},
|
| 142 |
+
"enc_blocks.1.conv1.weight": {
|
| 143 |
+
"w_norm": 8.352360725402832,
|
| 144 |
+
"g_norm": 0.009228352457284927,
|
| 145 |
+
"shape": [
|
| 146 |
+
192,
|
| 147 |
+
192,
|
| 148 |
+
3,
|
| 149 |
+
3
|
| 150 |
+
]
|
| 151 |
+
},
|
| 152 |
+
"enc_blocks.1.conv1.bias": {
|
| 153 |
+
"w_norm": 0.1925705522298813,
|
| 154 |
+
"g_norm": 0.0006506195059046149,
|
| 155 |
+
"shape": [
|
| 156 |
+
192
|
| 157 |
+
]
|
| 158 |
+
},
|
| 159 |
+
"enc_blocks.1.t_proj.weight": {
|
| 160 |
+
"w_norm": 8.264411926269531,
|
| 161 |
+
"g_norm": 0.0007487694965675473,
|
| 162 |
+
"shape": [
|
| 163 |
+
192,
|
| 164 |
+
256
|
| 165 |
+
]
|
| 166 |
+
},
|
| 167 |
+
"enc_blocks.1.t_proj.bias": {
|
| 168 |
+
"w_norm": 0.523145854473114,
|
| 169 |
+
"g_norm": 0.0006508316146209836,
|
| 170 |
+
"shape": [
|
| 171 |
+
192
|
| 172 |
+
]
|
| 173 |
+
},
|
| 174 |
+
"enc_blocks.1.norm2.weight": {
|
| 175 |
+
"w_norm": 13.815474510192871,
|
| 176 |
+
"g_norm": 0.0007219835533760488,
|
| 177 |
+
"shape": [
|
| 178 |
+
192
|
| 179 |
+
]
|
| 180 |
+
},
|
| 181 |
+
"enc_blocks.1.norm2.bias": {
|
| 182 |
+
"w_norm": 0.08295512944459915,
|
| 183 |
+
"g_norm": 0.0007739425054751337,
|
| 184 |
+
"shape": [
|
| 185 |
+
192
|
| 186 |
+
]
|
| 187 |
+
},
|
| 188 |
+
"enc_blocks.1.conv2.weight": {
|
| 189 |
+
"w_norm": 8.333471298217773,
|
| 190 |
+
"g_norm": 0.012200208380818367,
|
| 191 |
+
"shape": [
|
| 192 |
+
192,
|
| 193 |
+
192,
|
| 194 |
+
3,
|
| 195 |
+
3
|
| 196 |
+
]
|
| 197 |
+
},
|
| 198 |
+
"enc_blocks.1.conv2.bias": {
|
| 199 |
+
"w_norm": 0.20990578830242157,
|
| 200 |
+
"g_norm": 0.0010111451847478747,
|
| 201 |
+
"shape": [
|
| 202 |
+
192
|
| 203 |
+
]
|
| 204 |
+
},
|
| 205 |
+
"enc_blocks.2.norm1.weight": {
|
| 206 |
+
"w_norm": 13.804555892944336,
|
| 207 |
+
"g_norm": 0.0002658558951225132,
|
| 208 |
+
"shape": [
|
| 209 |
+
192
|
| 210 |
+
]
|
| 211 |
+
},
|
| 212 |
+
"enc_blocks.2.norm1.bias": {
|
| 213 |
+
"w_norm": 0.07251661270856857,
|
| 214 |
+
"g_norm": 0.00028446578653529286,
|
| 215 |
+
"shape": [
|
| 216 |
+
192
|
| 217 |
+
]
|
| 218 |
+
},
|
| 219 |
+
"enc_blocks.2.conv1.weight": {
|
| 220 |
+
"w_norm": 11.796462059020996,
|
| 221 |
+
"g_norm": 0.00543182622641325,
|
| 222 |
+
"shape": [
|
| 223 |
+
384,
|
| 224 |
+
192,
|
| 225 |
+
3,
|
| 226 |
+
3
|
| 227 |
+
]
|
| 228 |
+
},
|
| 229 |
+
"enc_blocks.2.conv1.bias": {
|
| 230 |
+
"w_norm": 0.26768797636032104,
|
| 231 |
+
"g_norm": 0.00036368807195685804,
|
| 232 |
+
"shape": [
|
| 233 |
+
384
|
| 234 |
+
]
|
| 235 |
+
},
|
| 236 |
+
"enc_blocks.2.t_proj.weight": {
|
| 237 |
+
"w_norm": 12.015633583068848,
|
| 238 |
+
"g_norm": 0.0003876331902574748,
|
| 239 |
+
"shape": [
|
| 240 |
+
384,
|
| 241 |
+
256
|
| 242 |
+
]
|
| 243 |
+
},
|
| 244 |
+
"enc_blocks.2.t_proj.bias": {
|
| 245 |
+
"w_norm": 0.7215366363525391,
|
| 246 |
+
"g_norm": 0.00036360954982228577,
|
| 247 |
+
"shape": [
|
| 248 |
+
384
|
| 249 |
+
]
|
| 250 |
+
},
|
| 251 |
+
"enc_blocks.2.norm2.weight": {
|
| 252 |
+
"w_norm": 19.397293090820312,
|
| 253 |
+
"g_norm": 0.0003247189160902053,
|
| 254 |
+
"shape": [
|
| 255 |
+
384
|
| 256 |
+
]
|
| 257 |
+
},
|
| 258 |
+
"enc_blocks.2.norm2.bias": {
|
| 259 |
+
"w_norm": 0.26001405715942383,
|
| 260 |
+
"g_norm": 0.00037615434848703444,
|
| 261 |
+
"shape": [
|
| 262 |
+
384
|
| 263 |
+
]
|
| 264 |
+
},
|
| 265 |
+
"enc_blocks.2.conv2.weight": {
|
| 266 |
+
"w_norm": 11.833823204040527,
|
| 267 |
+
"g_norm": 0.009932780638337135,
|
| 268 |
+
"shape": [
|
| 269 |
+
384,
|
| 270 |
+
384,
|
| 271 |
+
3,
|
| 272 |
+
3
|
| 273 |
+
]
|
| 274 |
+
},
|
| 275 |
+
"enc_blocks.2.conv2.bias": {
|
| 276 |
+
"w_norm": 0.20211109519004822,
|
| 277 |
+
"g_norm": 0.0006463375175371766,
|
| 278 |
+
"shape": [
|
| 279 |
+
384
|
| 280 |
+
]
|
| 281 |
+
},
|
| 282 |
+
"enc_blocks.2.skip.weight": {
|
| 283 |
+
"w_norm": 11.166708946228027,
|
| 284 |
+
"g_norm": 0.010496451519429684,
|
| 285 |
+
"shape": [
|
| 286 |
+
384,
|
| 287 |
+
192,
|
| 288 |
+
1,
|
| 289 |
+
1
|
| 290 |
+
]
|
| 291 |
+
},
|
| 292 |
+
"enc_blocks.2.skip.bias": {
|
| 293 |
+
"w_norm": 0.8282492756843567,
|
| 294 |
+
"g_norm": 0.0006463375175371766,
|
| 295 |
+
"shape": [
|
| 296 |
+
384
|
| 297 |
+
]
|
| 298 |
+
},
|
| 299 |
+
"enc_blocks.3.norm1.weight": {
|
| 300 |
+
"w_norm": 19.57082176208496,
|
| 301 |
+
"g_norm": 0.00017050265159923583,
|
| 302 |
+
"shape": [
|
| 303 |
+
384
|
| 304 |
+
]
|
| 305 |
+
},
|
| 306 |
+
"enc_blocks.3.norm1.bias": {
|
| 307 |
+
"w_norm": 0.07634127885103226,
|
| 308 |
+
"g_norm": 0.000175074819708243,
|
| 309 |
+
"shape": [
|
| 310 |
+
384
|
| 311 |
+
]
|
| 312 |
+
},
|
| 313 |
+
"enc_blocks.3.conv1.weight": {
|
| 314 |
+
"w_norm": 12.135543823242188,
|
| 315 |
+
"g_norm": 0.003038321156054735,
|
| 316 |
+
"shape": [
|
| 317 |
+
384,
|
| 318 |
+
384,
|
| 319 |
+
3,
|
| 320 |
+
3
|
| 321 |
+
]
|
| 322 |
+
},
|
| 323 |
+
"enc_blocks.3.conv1.bias": {
|
| 324 |
+
"w_norm": 0.19764570891857147,
|
| 325 |
+
"g_norm": 0.00012538516602944583,
|
| 326 |
+
"shape": [
|
| 327 |
+
384
|
| 328 |
+
]
|
| 329 |
+
},
|
| 330 |
+
"enc_blocks.3.t_proj.weight": {
|
| 331 |
+
"w_norm": 11.44118881225586,
|
| 332 |
+
"g_norm": 0.00013460988702718168,
|
| 333 |
+
"shape": [
|
| 334 |
+
384,
|
| 335 |
+
256
|
| 336 |
+
]
|
| 337 |
+
},
|
| 338 |
+
"enc_blocks.3.t_proj.bias": {
|
| 339 |
+
"w_norm": 0.692889392375946,
|
| 340 |
+
"g_norm": 0.00012540096940938383,
|
| 341 |
+
"shape": [
|
| 342 |
+
384
|
| 343 |
+
]
|
| 344 |
+
},
|
| 345 |
+
"enc_blocks.3.norm2.weight": {
|
| 346 |
+
"w_norm": 19.37941551208496,
|
| 347 |
+
"g_norm": 0.0001833721762523055,
|
| 348 |
+
"shape": [
|
| 349 |
+
384
|
| 350 |
+
]
|
| 351 |
+
},
|
| 352 |
+
"enc_blocks.3.norm2.bias": {
|
| 353 |
+
"w_norm": 0.2353246659040451,
|
| 354 |
+
"g_norm": 0.00028943736106157303,
|
| 355 |
+
"shape": [
|
| 356 |
+
384
|
| 357 |
+
]
|
| 358 |
+
},
|
| 359 |
+
"enc_blocks.3.conv2.weight": {
|
| 360 |
+
"w_norm": 11.664400100708008,
|
| 361 |
+
"g_norm": 0.006047896109521389,
|
| 362 |
+
"shape": [
|
| 363 |
+
384,
|
| 364 |
+
384,
|
| 365 |
+
3,
|
| 366 |
+
3
|
| 367 |
+
]
|
| 368 |
+
},
|
| 369 |
+
"enc_blocks.3.conv2.bias": {
|
| 370 |
+
"w_norm": 0.1913631409406662,
|
| 371 |
+
"g_norm": 0.0004047580005135387,
|
| 372 |
+
"shape": [
|
| 373 |
+
384
|
| 374 |
+
]
|
| 375 |
+
},
|
| 376 |
+
"enc_blocks.4.norm1.weight": {
|
| 377 |
+
"w_norm": 19.603595733642578,
|
| 378 |
+
"g_norm": 0.0004483657539822161,
|
| 379 |
+
"shape": [
|
| 380 |
+
384
|
| 381 |
+
]
|
| 382 |
+
},
|
| 383 |
+
"enc_blocks.4.norm1.bias": {
|
| 384 |
+
"w_norm": 0.07635044306516647,
|
| 385 |
+
"g_norm": 0.00047479913337156177,
|
| 386 |
+
"shape": [
|
| 387 |
+
384
|
| 388 |
+
]
|
| 389 |
+
},
|
| 390 |
+
"enc_blocks.4.conv1.weight": {
|
| 391 |
+
"w_norm": 16.962387084960938,
|
| 392 |
+
"g_norm": 0.010729203931987286,
|
| 393 |
+
"shape": [
|
| 394 |
+
768,
|
| 395 |
+
384,
|
| 396 |
+
3,
|
| 397 |
+
3
|
| 398 |
+
]
|
| 399 |
+
},
|
| 400 |
+
"enc_blocks.4.conv1.bias": {
|
| 401 |
+
"w_norm": 0.2939714789390564,
|
| 402 |
+
"g_norm": 0.00041833348223008215,
|
| 403 |
+
"shape": [
|
| 404 |
+
768
|
| 405 |
+
]
|
| 406 |
+
},
|
| 407 |
+
"enc_blocks.4.t_proj.weight": {
|
| 408 |
+
"w_norm": 16.067256927490234,
|
| 409 |
+
"g_norm": 0.00043840549187734723,
|
| 410 |
+
"shape": [
|
| 411 |
+
768,
|
| 412 |
+
256
|
| 413 |
+
]
|
| 414 |
+
},
|
| 415 |
+
"enc_blocks.4.t_proj.bias": {
|
| 416 |
+
"w_norm": 0.9943748712539673,
|
| 417 |
+
"g_norm": 0.00041832224815152586,
|
| 418 |
+
"shape": [
|
| 419 |
+
768
|
| 420 |
+
]
|
| 421 |
+
},
|
| 422 |
+
"enc_blocks.4.norm2.weight": {
|
| 423 |
+
"w_norm": 27.357318878173828,
|
| 424 |
+
"g_norm": 0.0007662373827770352,
|
| 425 |
+
"shape": [
|
| 426 |
+
768
|
| 427 |
+
]
|
| 428 |
+
},
|
| 429 |
+
"enc_blocks.4.norm2.bias": {
|
| 430 |
+
"w_norm": 0.4109998047351837,
|
| 431 |
+
"g_norm": 0.0009103051270358264,
|
| 432 |
+
"shape": [
|
| 433 |
+
768
|
| 434 |
+
]
|
| 435 |
+
},
|
| 436 |
+
"enc_blocks.4.conv2.weight": {
|
| 437 |
+
"w_norm": 16.50996208190918,
|
| 438 |
+
"g_norm": 0.011206631548702717,
|
| 439 |
+
"shape": [
|
| 440 |
+
768,
|
| 441 |
+
768,
|
| 442 |
+
3,
|
| 443 |
+
3
|
| 444 |
+
]
|
| 445 |
+
},
|
| 446 |
+
"enc_blocks.4.conv2.bias": {
|
| 447 |
+
"w_norm": 0.19958554208278656,
|
| 448 |
+
"g_norm": 0.00041942455573007464,
|
| 449 |
+
"shape": [
|
| 450 |
+
768
|
| 451 |
+
]
|
| 452 |
+
},
|
| 453 |
+
"enc_blocks.4.skip.weight": {
|
| 454 |
+
"w_norm": 15.74730110168457,
|
| 455 |
+
"g_norm": 0.011705640703439713,
|
| 456 |
+
"shape": [
|
| 457 |
+
768,
|
| 458 |
+
384,
|
| 459 |
+
1,
|
| 460 |
+
1
|
| 461 |
+
]
|
| 462 |
+
},
|
| 463 |
+
"enc_blocks.4.skip.bias": {
|
| 464 |
+
"w_norm": 0.8086597919464111,
|
| 465 |
+
"g_norm": 0.00041942455573007464,
|
| 466 |
+
"shape": [
|
| 467 |
+
768
|
| 468 |
+
]
|
| 469 |
+
},
|
| 470 |
+
"enc_blocks.5.norm1.weight": {
|
| 471 |
+
"w_norm": 27.715965270996094,
|
| 472 |
+
"g_norm": 0.00013690636842511594,
|
| 473 |
+
"shape": [
|
| 474 |
+
768
|
| 475 |
+
]
|
| 476 |
+
},
|
| 477 |
+
"enc_blocks.5.norm1.bias": {
|
| 478 |
+
"w_norm": 0.07511666417121887,
|
| 479 |
+
"g_norm": 0.00011273950076429173,
|
| 480 |
+
"shape": [
|
| 481 |
+
768
|
| 482 |
+
]
|
| 483 |
+
},
|
| 484 |
+
"enc_blocks.5.conv1.weight": {
|
| 485 |
+
"w_norm": 17.259342193603516,
|
| 486 |
+
"g_norm": 0.002783330390229821,
|
| 487 |
+
"shape": [
|
| 488 |
+
768,
|
| 489 |
+
768,
|
| 490 |
+
3,
|
| 491 |
+
3
|
| 492 |
+
]
|
| 493 |
+
},
|
| 494 |
+
"enc_blocks.5.conv1.bias": {
|
| 495 |
+
"w_norm": 0.20967231690883636,
|
| 496 |
+
"g_norm": 5.966354729025625e-05,
|
| 497 |
+
"shape": [
|
| 498 |
+
768
|
| 499 |
+
]
|
| 500 |
+
},
|
| 501 |
+
"enc_blocks.5.t_proj.weight": {
|
| 502 |
+
"w_norm": 16.016868591308594,
|
| 503 |
+
"g_norm": 6.135379226179793e-05,
|
| 504 |
+
"shape": [
|
| 505 |
+
768,
|
| 506 |
+
256
|
| 507 |
+
]
|
| 508 |
+
},
|
| 509 |
+
"enc_blocks.5.t_proj.bias": {
|
| 510 |
+
"w_norm": 1.0008527040481567,
|
| 511 |
+
"g_norm": 5.964525553281419e-05,
|
| 512 |
+
"shape": [
|
| 513 |
+
768
|
| 514 |
+
]
|
| 515 |
+
},
|
| 516 |
+
"enc_blocks.5.norm2.weight": {
|
| 517 |
+
"w_norm": 27.47331428527832,
|
| 518 |
+
"g_norm": 0.00027751424931921065,
|
| 519 |
+
"shape": [
|
| 520 |
+
768
|
| 521 |
+
]
|
| 522 |
+
},
|
| 523 |
+
"enc_blocks.5.norm2.bias": {
|
| 524 |
+
"w_norm": 0.28648436069488525,
|
| 525 |
+
"g_norm": 0.00025248766178265214,
|
| 526 |
+
"shape": [
|
| 527 |
+
768
|
| 528 |
+
]
|
| 529 |
+
},
|
| 530 |
+
"enc_blocks.5.conv2.weight": {
|
| 531 |
+
"w_norm": 16.705930709838867,
|
| 532 |
+
"g_norm": 0.007712048012763262,
|
| 533 |
+
"shape": [
|
| 534 |
+
768,
|
| 535 |
+
768,
|
| 536 |
+
3,
|
| 537 |
+
3
|
| 538 |
+
]
|
| 539 |
+
},
|
| 540 |
+
"enc_blocks.5.conv2.bias": {
|
| 541 |
+
"w_norm": 0.19320331513881683,
|
| 542 |
+
"g_norm": 0.0002017767692450434,
|
| 543 |
+
"shape": [
|
| 544 |
+
768
|
| 545 |
+
]
|
| 546 |
+
},
|
| 547 |
+
"enc_attns.3.norm.weight": {
|
| 548 |
+
"w_norm": 19.50695037841797,
|
| 549 |
+
"g_norm": 9.197030158247799e-05,
|
| 550 |
+
"shape": [
|
| 551 |
+
384
|
| 552 |
+
]
|
| 553 |
+
},
|
| 554 |
+
"enc_attns.3.norm.bias": {
|
| 555 |
+
"w_norm": 0.04814915731549263,
|
| 556 |
+
"g_norm": 0.00017879036022350192,
|
| 557 |
+
"shape": [
|
| 558 |
+
384
|
| 559 |
+
]
|
| 560 |
+
},
|
| 561 |
+
"enc_attns.3.qkv.weight": {
|
| 562 |
+
"w_norm": 19.677698135375977,
|
| 563 |
+
"g_norm": 0.0026700582820922136,
|
| 564 |
+
"shape": [
|
| 565 |
+
1152,
|
| 566 |
+
384,
|
| 567 |
+
1,
|
| 568 |
+
1
|
| 569 |
+
]
|
| 570 |
+
},
|
| 571 |
+
"enc_attns.3.qkv.bias": {
|
| 572 |
+
"w_norm": 1.0045832395553589,
|
| 573 |
+
"g_norm": 0.0002563674352131784,
|
| 574 |
+
"shape": [
|
| 575 |
+
1152
|
| 576 |
+
]
|
| 577 |
+
},
|
| 578 |
+
"enc_attns.3.proj.weight": {
|
| 579 |
+
"w_norm": 11.314918518066406,
|
| 580 |
+
"g_norm": 0.002640425693243742,
|
| 581 |
+
"shape": [
|
| 582 |
+
384,
|
| 583 |
+
384,
|
| 584 |
+
1,
|
| 585 |
+
1
|
| 586 |
+
]
|
| 587 |
+
},
|
| 588 |
+
"enc_attns.3.proj.bias": {
|
| 589 |
+
"w_norm": 0.561999499797821,
|
| 590 |
+
"g_norm": 0.0004093097522854805,
|
| 591 |
+
"shape": [
|
| 592 |
+
384
|
| 593 |
+
]
|
| 594 |
+
},
|
| 595 |
+
"enc_attns.4.norm.weight": {
|
| 596 |
+
"w_norm": 27.55660629272461,
|
| 597 |
+
"g_norm": 0.00014296467998065054,
|
| 598 |
+
"shape": [
|
| 599 |
+
768
|
| 600 |
+
]
|
| 601 |
+
},
|
| 602 |
+
"enc_attns.4.norm.bias": {
|
| 603 |
+
"w_norm": 0.07683202624320984,
|
| 604 |
+
"g_norm": 0.00021652992290910333,
|
| 605 |
+
"shape": [
|
| 606 |
+
768
|
| 607 |
+
]
|
| 608 |
+
},
|
| 609 |
+
"enc_attns.4.qkv.weight": {
|
| 610 |
+
"w_norm": 28.08203887939453,
|
| 611 |
+
"g_norm": 0.004884236492216587,
|
| 612 |
+
"shape": [
|
| 613 |
+
2304,
|
| 614 |
+
768,
|
| 615 |
+
1,
|
| 616 |
+
1
|
| 617 |
+
]
|
| 618 |
+
},
|
| 619 |
+
"enc_attns.4.qkv.bias": {
|
| 620 |
+
"w_norm": 1.0157814025878906,
|
| 621 |
+
"g_norm": 0.0002601943851914257,
|
| 622 |
+
"shape": [
|
| 623 |
+
2304
|
| 624 |
+
]
|
| 625 |
+
},
|
| 626 |
+
"enc_attns.4.proj.weight": {
|
| 627 |
+
"w_norm": 16.036779403686523,
|
| 628 |
+
"g_norm": 0.004596893209964037,
|
| 629 |
+
"shape": [
|
| 630 |
+
768,
|
| 631 |
+
768,
|
| 632 |
+
1,
|
| 633 |
+
1
|
| 634 |
+
]
|
| 635 |
+
},
|
| 636 |
+
"enc_attns.4.proj.bias": {
|
| 637 |
+
"w_norm": 0.5738911628723145,
|
| 638 |
+
"g_norm": 0.000422380689997226,
|
| 639 |
+
"shape": [
|
| 640 |
+
768
|
| 641 |
+
]
|
| 642 |
+
},
|
| 643 |
+
"enc_attns.5.norm.weight": {
|
| 644 |
+
"w_norm": 27.647903442382812,
|
| 645 |
+
"g_norm": 4.3613141315290704e-05,
|
| 646 |
+
"shape": [
|
| 647 |
+
768
|
| 648 |
+
]
|
| 649 |
+
},
|
| 650 |
+
"enc_attns.5.norm.bias": {
|
| 651 |
+
"w_norm": 0.06394332647323608,
|
| 652 |
+
"g_norm": 9.029910142999142e-05,
|
| 653 |
+
"shape": [
|
| 654 |
+
768
|
| 655 |
+
]
|
| 656 |
+
},
|
| 657 |
+
"enc_attns.5.qkv.weight": {
|
| 658 |
+
"w_norm": 28.060827255249023,
|
| 659 |
+
"g_norm": 0.0017676395364105701,
|
| 660 |
+
"shape": [
|
| 661 |
+
2304,
|
| 662 |
+
768,
|
| 663 |
+
1,
|
| 664 |
+
1
|
| 665 |
+
]
|
| 666 |
+
},
|
| 667 |
+
"enc_attns.5.qkv.bias": {
|
| 668 |
+
"w_norm": 1.0001214742660522,
|
| 669 |
+
"g_norm": 0.00012707892165053636,
|
| 670 |
+
"shape": [
|
| 671 |
+
2304
|
| 672 |
+
]
|
| 673 |
+
},
|
| 674 |
+
"enc_attns.5.proj.weight": {
|
| 675 |
+
"w_norm": 16.089628219604492,
|
| 676 |
+
"g_norm": 0.001794372103177011,
|
| 677 |
+
"shape": [
|
| 678 |
+
768,
|
| 679 |
+
768,
|
| 680 |
+
1,
|
| 681 |
+
1
|
| 682 |
+
]
|
| 683 |
+
},
|
| 684 |
+
"enc_attns.5.proj.bias": {
|
| 685 |
+
"w_norm": 0.5971724987030029,
|
| 686 |
+
"g_norm": 0.0002005139976972714,
|
| 687 |
+
"shape": [
|
| 688 |
+
768
|
| 689 |
+
]
|
| 690 |
+
},
|
| 691 |
+
"enc_downs.0.op.weight": {
|
| 692 |
+
"w_norm": 8.235817909240723,
|
| 693 |
+
"g_norm": 0.018652230501174927,
|
| 694 |
+
"shape": [
|
| 695 |
+
192,
|
| 696 |
+
192,
|
| 697 |
+
3,
|
| 698 |
+
3
|
| 699 |
+
]
|
| 700 |
+
},
|
| 701 |
+
"enc_downs.0.op.bias": {
|
| 702 |
+
"w_norm": 0.18452279269695282,
|
| 703 |
+
"g_norm": 0.0009377316455356777,
|
| 704 |
+
"shape": [
|
| 705 |
+
192
|
| 706 |
+
]
|
| 707 |
+
},
|
| 708 |
+
"enc_downs.1.op.weight": {
|
| 709 |
+
"w_norm": 8.140389442443848,
|
| 710 |
+
"g_norm": 0.017915265634655952,
|
| 711 |
+
"shape": [
|
| 712 |
+
192,
|
| 713 |
+
192,
|
| 714 |
+
3,
|
| 715 |
+
3
|
| 716 |
+
]
|
| 717 |
+
},
|
| 718 |
+
"enc_downs.1.op.bias": {
|
| 719 |
+
"w_norm": 0.19395767152309418,
|
| 720 |
+
"g_norm": 0.0005633846158161759,
|
| 721 |
+
"shape": [
|
| 722 |
+
192
|
| 723 |
+
]
|
| 724 |
+
},
|
| 725 |
+
"enc_downs.2.op.weight": {
|
| 726 |
+
"w_norm": 11.588905334472656,
|
| 727 |
+
"g_norm": 0.016342995688319206,
|
| 728 |
+
"shape": [
|
| 729 |
+
384,
|
| 730 |
+
384,
|
| 731 |
+
3,
|
| 732 |
+
3
|
| 733 |
+
]
|
| 734 |
+
},
|
| 735 |
+
"enc_downs.2.op.bias": {
|
| 736 |
+
"w_norm": 0.18809567391872406,
|
| 737 |
+
"g_norm": 0.00040573853766545653,
|
| 738 |
+
"shape": [
|
| 739 |
+
384
|
| 740 |
+
]
|
| 741 |
+
},
|
| 742 |
+
"enc_downs.3.op.weight": {
|
| 743 |
+
"w_norm": 11.490419387817383,
|
| 744 |
+
"g_norm": 0.022241534665226936,
|
| 745 |
+
"shape": [
|
| 746 |
+
384,
|
| 747 |
+
384,
|
| 748 |
+
3,
|
| 749 |
+
3
|
| 750 |
+
]
|
| 751 |
+
},
|
| 752 |
+
"enc_downs.3.op.bias": {
|
| 753 |
+
"w_norm": 0.19505050778388977,
|
| 754 |
+
"g_norm": 0.00036343102692626417,
|
| 755 |
+
"shape": [
|
| 756 |
+
384
|
| 757 |
+
]
|
| 758 |
+
},
|
| 759 |
+
"enc_downs.4.op.weight": {
|
| 760 |
+
"w_norm": 16.517168045043945,
|
| 761 |
+
"g_norm": 0.023726537823677063,
|
| 762 |
+
"shape": [
|
| 763 |
+
768,
|
| 764 |
+
768,
|
| 765 |
+
3,
|
| 766 |
+
3
|
| 767 |
+
]
|
| 768 |
+
},
|
| 769 |
+
"enc_downs.4.op.bias": {
|
| 770 |
+
"w_norm": 0.20217803120613098,
|
| 771 |
+
"g_norm": 0.00020157717517577112,
|
| 772 |
+
"shape": [
|
| 773 |
+
768
|
| 774 |
+
]
|
| 775 |
+
},
|
| 776 |
+
"mid1.norm1.weight": {
|
| 777 |
+
"w_norm": 27.712066650390625,
|
| 778 |
+
"g_norm": 7.81923154136166e-05,
|
| 779 |
+
"shape": [
|
| 780 |
+
768
|
| 781 |
+
]
|
| 782 |
+
},
|
| 783 |
+
"mid1.norm1.bias": {
|
| 784 |
+
"w_norm": 0.05954904481768608,
|
| 785 |
+
"g_norm": 6.75633127684705e-05,
|
| 786 |
+
"shape": [
|
| 787 |
+
768
|
| 788 |
+
]
|
| 789 |
+
},
|
| 790 |
+
"mid1.conv1.weight": {
|
| 791 |
+
"w_norm": 17.084714889526367,
|
| 792 |
+
"g_norm": 0.0018813041970133781,
|
| 793 |
+
"shape": [
|
| 794 |
+
768,
|
| 795 |
+
768,
|
| 796 |
+
3,
|
| 797 |
+
3
|
| 798 |
+
]
|
| 799 |
+
},
|
| 800 |
+
"mid1.conv1.bias": {
|
| 801 |
+
"w_norm": 0.20746949315071106,
|
| 802 |
+
"g_norm": 4.295578764867969e-05,
|
| 803 |
+
"shape": [
|
| 804 |
+
768
|
| 805 |
+
]
|
| 806 |
+
},
|
| 807 |
+
"mid1.t_proj.weight": {
|
| 808 |
+
"w_norm": 16.020248413085938,
|
| 809 |
+
"g_norm": 4.511191582423635e-05,
|
| 810 |
+
"shape": [
|
| 811 |
+
768,
|
| 812 |
+
256
|
| 813 |
+
]
|
| 814 |
+
},
|
| 815 |
+
"mid1.t_proj.bias": {
|
| 816 |
+
"w_norm": 1.030643105506897,
|
| 817 |
+
"g_norm": 4.29449173680041e-05,
|
| 818 |
+
"shape": [
|
| 819 |
+
768
|
| 820 |
+
]
|
| 821 |
+
},
|
| 822 |
+
"mid1.norm2.weight": {
|
| 823 |
+
"w_norm": 27.478395462036133,
|
| 824 |
+
"g_norm": 5.8335594076197594e-05,
|
| 825 |
+
"shape": [
|
| 826 |
+
768
|
| 827 |
+
]
|
| 828 |
+
},
|
| 829 |
+
"mid1.norm2.bias": {
|
| 830 |
+
"w_norm": 0.32949304580688477,
|
| 831 |
+
"g_norm": 8.398802310694009e-05,
|
| 832 |
+
"shape": [
|
| 833 |
+
768
|
| 834 |
+
]
|
| 835 |
+
},
|
| 836 |
+
"mid1.conv2.weight": {
|
| 837 |
+
"w_norm": 16.644107818603516,
|
| 838 |
+
"g_norm": 0.002602463588118553,
|
| 839 |
+
"shape": [
|
| 840 |
+
768,
|
| 841 |
+
768,
|
| 842 |
+
3,
|
| 843 |
+
3
|
| 844 |
+
]
|
| 845 |
+
},
|
| 846 |
+
"mid1.conv2.bias": {
|
| 847 |
+
"w_norm": 0.19403482973575592,
|
| 848 |
+
"g_norm": 0.00012408455950208008,
|
| 849 |
+
"shape": [
|
| 850 |
+
768
|
| 851 |
+
]
|
| 852 |
+
},
|
| 853 |
+
"mid_attn.norm.weight": {
|
| 854 |
+
"w_norm": 27.65032958984375,
|
| 855 |
+
"g_norm": 2.737711838562973e-05,
|
| 856 |
+
"shape": [
|
| 857 |
+
768
|
| 858 |
+
]
|
| 859 |
+
},
|
| 860 |
+
"mid_attn.norm.bias": {
|
| 861 |
+
"w_norm": 0.0695677399635315,
|
| 862 |
+
"g_norm": 4.668824476539157e-05,
|
| 863 |
+
"shape": [
|
| 864 |
+
768
|
| 865 |
+
]
|
| 866 |
+
},
|
| 867 |
+
"mid_attn.qkv.weight": {
|
| 868 |
+
"w_norm": 27.996524810791016,
|
| 869 |
+
"g_norm": 0.001264288672246039,
|
| 870 |
+
"shape": [
|
| 871 |
+
2304,
|
| 872 |
+
768,
|
| 873 |
+
1,
|
| 874 |
+
1
|
| 875 |
+
]
|
| 876 |
+
},
|
| 877 |
+
"mid_attn.qkv.bias": {
|
| 878 |
+
"w_norm": 0.9832267761230469,
|
| 879 |
+
"g_norm": 7.399465539492667e-05,
|
| 880 |
+
"shape": [
|
| 881 |
+
2304
|
| 882 |
+
]
|
| 883 |
+
},
|
| 884 |
+
"mid_attn.proj.weight": {
|
| 885 |
+
"w_norm": 16.07176399230957,
|
| 886 |
+
"g_norm": 0.0011793198063969612,
|
| 887 |
+
"shape": [
|
| 888 |
+
768,
|
| 889 |
+
768,
|
| 890 |
+
1,
|
| 891 |
+
1
|
| 892 |
+
]
|
| 893 |
+
},
|
| 894 |
+
"mid_attn.proj.bias": {
|
| 895 |
+
"w_norm": 0.576002836227417,
|
| 896 |
+
"g_norm": 0.00012308346049394459,
|
| 897 |
+
"shape": [
|
| 898 |
+
768
|
| 899 |
+
]
|
| 900 |
+
},
|
| 901 |
+
"mid2.norm1.weight": {
|
| 902 |
+
"w_norm": 27.723308563232422,
|
| 903 |
+
"g_norm": 0.00023281383619178087,
|
| 904 |
+
"shape": [
|
| 905 |
+
768
|
| 906 |
+
]
|
| 907 |
+
},
|
| 908 |
+
"mid2.norm1.bias": {
|
| 909 |
+
"w_norm": 0.06764264404773712,
|
| 910 |
+
"g_norm": 0.00020761771884281188,
|
| 911 |
+
"shape": [
|
| 912 |
+
768
|
| 913 |
+
]
|
| 914 |
+
},
|
| 915 |
+
"mid2.conv1.weight": {
|
| 916 |
+
"w_norm": 17.205245971679688,
|
| 917 |
+
"g_norm": 0.0032772363629192114,
|
| 918 |
+
"shape": [
|
| 919 |
+
768,
|
| 920 |
+
768,
|
| 921 |
+
3,
|
| 922 |
+
3
|
| 923 |
+
]
|
| 924 |
+
},
|
| 925 |
+
"mid2.conv1.bias": {
|
| 926 |
+
"w_norm": 0.2025199979543686,
|
| 927 |
+
"g_norm": 7.581734098494053e-05,
|
| 928 |
+
"shape": [
|
| 929 |
+
768
|
| 930 |
+
]
|
| 931 |
+
},
|
| 932 |
+
"mid2.t_proj.weight": {
|
| 933 |
+
"w_norm": 16.00699234008789,
|
| 934 |
+
"g_norm": 7.93769650044851e-05,
|
| 935 |
+
"shape": [
|
| 936 |
+
768,
|
| 937 |
+
256
|
| 938 |
+
]
|
| 939 |
+
},
|
| 940 |
+
"mid2.t_proj.bias": {
|
| 941 |
+
"w_norm": 0.9997656941413879,
|
| 942 |
+
"g_norm": 7.582550460938364e-05,
|
| 943 |
+
"shape": [
|
| 944 |
+
768
|
| 945 |
+
]
|
| 946 |
+
},
|
| 947 |
+
"mid2.norm2.weight": {
|
| 948 |
+
"w_norm": 27.45841407775879,
|
| 949 |
+
"g_norm": 0.0001569108135299757,
|
| 950 |
+
"shape": [
|
| 951 |
+
768
|
| 952 |
+
]
|
| 953 |
+
},
|
| 954 |
+
"mid2.norm2.bias": {
|
| 955 |
+
"w_norm": 0.3297288417816162,
|
| 956 |
+
"g_norm": 0.00017398105410393327,
|
| 957 |
+
"shape": [
|
| 958 |
+
768
|
| 959 |
+
]
|
| 960 |
+
},
|
| 961 |
+
"mid2.conv2.weight": {
|
| 962 |
+
"w_norm": 16.620267868041992,
|
| 963 |
+
"g_norm": 0.0026973795611411333,
|
| 964 |
+
"shape": [
|
| 965 |
+
768,
|
| 966 |
+
768,
|
| 967 |
+
3,
|
| 968 |
+
3
|
| 969 |
+
]
|
| 970 |
+
},
|
| 971 |
+
"mid2.conv2.bias": {
|
| 972 |
+
"w_norm": 0.19106969237327576,
|
| 973 |
+
"g_norm": 0.00011876138887600973,
|
| 974 |
+
"shape": [
|
| 975 |
+
768
|
| 976 |
+
]
|
| 977 |
+
},
|
| 978 |
+
"dec_blocks.0.norm1.weight": {
|
| 979 |
+
"w_norm": 39.22458267211914,
|
| 980 |
+
"g_norm": 0.0001079371286323294,
|
| 981 |
+
"shape": [
|
| 982 |
+
1536
|
| 983 |
+
]
|
| 984 |
+
},
|
| 985 |
+
"dec_blocks.0.norm1.bias": {
|
| 986 |
+
"w_norm": 0.1136740893125534,
|
| 987 |
+
"g_norm": 9.537713776808232e-05,
|
| 988 |
+
"shape": [
|
| 989 |
+
1536
|
| 990 |
+
]
|
| 991 |
+
},
|
| 992 |
+
"dec_blocks.0.conv1.weight": {
|
| 993 |
+
"w_norm": 18.268850326538086,
|
| 994 |
+
"g_norm": 0.0026090722531080246,
|
| 995 |
+
"shape": [
|
| 996 |
+
768,
|
| 997 |
+
1536,
|
| 998 |
+
3,
|
| 999 |
+
3
|
| 1000 |
+
]
|
| 1001 |
+
},
|
| 1002 |
+
"dec_blocks.0.conv1.bias": {
|
| 1003 |
+
"w_norm": 0.15742380917072296,
|
| 1004 |
+
"g_norm": 4.251589416526258e-05,
|
| 1005 |
+
"shape": [
|
| 1006 |
+
768
|
| 1007 |
+
]
|
| 1008 |
+
},
|
| 1009 |
+
"dec_blocks.0.t_proj.weight": {
|
| 1010 |
+
"w_norm": 16.02547836303711,
|
| 1011 |
+
"g_norm": 4.457262184587307e-05,
|
| 1012 |
+
"shape": [
|
| 1013 |
+
768,
|
| 1014 |
+
256
|
| 1015 |
+
]
|
| 1016 |
+
},
|
| 1017 |
+
"dec_blocks.0.t_proj.bias": {
|
| 1018 |
+
"w_norm": 1.0245296955108643,
|
| 1019 |
+
"g_norm": 4.250950587447733e-05,
|
| 1020 |
+
"shape": [
|
| 1021 |
+
768
|
| 1022 |
+
]
|
| 1023 |
+
},
|
| 1024 |
+
"dec_blocks.0.norm2.weight": {
|
| 1025 |
+
"w_norm": 27.40232276916504,
|
| 1026 |
+
"g_norm": 0.00011834588804049417,
|
| 1027 |
+
"shape": [
|
| 1028 |
+
768
|
| 1029 |
+
]
|
| 1030 |
+
},
|
| 1031 |
+
"dec_blocks.0.norm2.bias": {
|
| 1032 |
+
"w_norm": 0.4019148647785187,
|
| 1033 |
+
"g_norm": 0.00021984959312248975,
|
| 1034 |
+
"shape": [
|
| 1035 |
+
768
|
| 1036 |
+
]
|
| 1037 |
+
},
|
| 1038 |
+
"dec_blocks.0.conv2.weight": {
|
| 1039 |
+
"w_norm": 16.37566566467285,
|
| 1040 |
+
"g_norm": 0.0035729294177144766,
|
| 1041 |
+
"shape": [
|
| 1042 |
+
768,
|
| 1043 |
+
768,
|
| 1044 |
+
3,
|
| 1045 |
+
3
|
| 1046 |
+
]
|
| 1047 |
+
},
|
| 1048 |
+
"dec_blocks.0.conv2.bias": {
|
| 1049 |
+
"w_norm": 0.19908437132835388,
|
| 1050 |
+
"g_norm": 0.00025187665596604347,
|
| 1051 |
+
"shape": [
|
| 1052 |
+
768
|
| 1053 |
+
]
|
| 1054 |
+
},
|
| 1055 |
+
"dec_blocks.0.skip.weight": {
|
| 1056 |
+
"w_norm": 15.740446090698242,
|
| 1057 |
+
"g_norm": 0.037088118493556976,
|
| 1058 |
+
"shape": [
|
| 1059 |
+
768,
|
| 1060 |
+
1536,
|
| 1061 |
+
1,
|
| 1062 |
+
1
|
| 1063 |
+
]
|
| 1064 |
+
},
|
| 1065 |
+
"dec_blocks.0.skip.bias": {
|
| 1066 |
+
"w_norm": 0.4123978316783905,
|
| 1067 |
+
"g_norm": 0.00025187665596604347,
|
| 1068 |
+
"shape": [
|
| 1069 |
+
768
|
| 1070 |
+
]
|
| 1071 |
+
},
|
| 1072 |
+
"dec_blocks.1.norm1.weight": {
|
| 1073 |
+
"w_norm": 39.235748291015625,
|
| 1074 |
+
"g_norm": 0.00011050483590224758,
|
| 1075 |
+
"shape": [
|
| 1076 |
+
1536
|
| 1077 |
+
]
|
| 1078 |
+
},
|
| 1079 |
+
"dec_blocks.1.norm1.bias": {
|
| 1080 |
+
"w_norm": 0.10630356520414352,
|
| 1081 |
+
"g_norm": 0.00011029136658180505,
|
| 1082 |
+
"shape": [
|
| 1083 |
+
1536
|
| 1084 |
+
]
|
| 1085 |
+
},
|
| 1086 |
+
"dec_blocks.1.conv1.weight": {
|
| 1087 |
+
"w_norm": 18.143503189086914,
|
| 1088 |
+
"g_norm": 0.0031994630116969347,
|
| 1089 |
+
"shape": [
|
| 1090 |
+
768,
|
| 1091 |
+
1536,
|
| 1092 |
+
3,
|
| 1093 |
+
3
|
| 1094 |
+
]
|
| 1095 |
+
},
|
| 1096 |
+
"dec_blocks.1.conv1.bias": {
|
| 1097 |
+
"w_norm": 0.1565992832183838,
|
| 1098 |
+
"g_norm": 5.2350569603731856e-05,
|
| 1099 |
+
"shape": [
|
| 1100 |
+
768
|
| 1101 |
+
]
|
| 1102 |
+
},
|
| 1103 |
+
"dec_blocks.1.t_proj.weight": {
|
| 1104 |
+
"w_norm": 16.061790466308594,
|
| 1105 |
+
"g_norm": 5.610016523860395e-05,
|
| 1106 |
+
"shape": [
|
| 1107 |
+
768,
|
| 1108 |
+
256
|
| 1109 |
+
]
|
| 1110 |
+
},
|
| 1111 |
+
"dec_blocks.1.t_proj.bias": {
|
| 1112 |
+
"w_norm": 0.9904308915138245,
|
| 1113 |
+
"g_norm": 5.23491544299759e-05,
|
| 1114 |
+
"shape": [
|
| 1115 |
+
768
|
| 1116 |
+
]
|
| 1117 |
+
},
|
| 1118 |
+
"dec_blocks.1.norm2.weight": {
|
| 1119 |
+
"w_norm": 27.39215660095215,
|
| 1120 |
+
"g_norm": 9.991656406782568e-05,
|
| 1121 |
+
"shape": [
|
| 1122 |
+
768
|
| 1123 |
+
]
|
| 1124 |
+
},
|
| 1125 |
+
"dec_blocks.1.norm2.bias": {
|
| 1126 |
+
"w_norm": 0.43167370557785034,
|
| 1127 |
+
"g_norm": 0.0002531306236051023,
|
| 1128 |
+
"shape": [
|
| 1129 |
+
768
|
| 1130 |
+
]
|
| 1131 |
+
},
|
| 1132 |
+
"dec_blocks.1.conv2.weight": {
|
| 1133 |
+
"w_norm": 16.279890060424805,
|
| 1134 |
+
"g_norm": 0.003765431232750416,
|
| 1135 |
+
"shape": [
|
| 1136 |
+
768,
|
| 1137 |
+
768,
|
| 1138 |
+
3,
|
| 1139 |
+
3
|
| 1140 |
+
]
|
| 1141 |
+
},
|
| 1142 |
+
"dec_blocks.1.conv2.bias": {
|
| 1143 |
+
"w_norm": 0.19697929918766022,
|
| 1144 |
+
"g_norm": 0.0002718368486966938,
|
| 1145 |
+
"shape": [
|
| 1146 |
+
768
|
| 1147 |
+
]
|
| 1148 |
+
},
|
| 1149 |
+
"dec_blocks.1.skip.weight": {
|
| 1150 |
+
"w_norm": 15.701605796813965,
|
| 1151 |
+
"g_norm": 0.051730845123529434,
|
| 1152 |
+
"shape": [
|
| 1153 |
+
768,
|
| 1154 |
+
1536,
|
| 1155 |
+
1,
|
| 1156 |
+
1
|
| 1157 |
+
]
|
| 1158 |
+
},
|
| 1159 |
+
"dec_blocks.1.skip.bias": {
|
| 1160 |
+
"w_norm": 0.407044380903244,
|
| 1161 |
+
"g_norm": 0.0002718368486966938,
|
| 1162 |
+
"shape": [
|
| 1163 |
+
768
|
| 1164 |
+
]
|
| 1165 |
+
},
|
| 1166 |
+
"dec_blocks.2.norm1.weight": {
|
| 1167 |
+
"w_norm": 33.96385192871094,
|
| 1168 |
+
"g_norm": 3.643078525783494e-05,
|
| 1169 |
+
"shape": [
|
| 1170 |
+
1152
|
| 1171 |
+
]
|
| 1172 |
+
},
|
| 1173 |
+
"dec_blocks.2.norm1.bias": {
|
| 1174 |
+
"w_norm": 0.10771745443344116,
|
| 1175 |
+
"g_norm": 3.162841312587261e-05,
|
| 1176 |
+
"shape": [
|
| 1177 |
+
1152
|
| 1178 |
+
]
|
| 1179 |
+
},
|
| 1180 |
+
"dec_blocks.2.conv1.weight": {
|
| 1181 |
+
"w_norm": 12.829668045043945,
|
| 1182 |
+
"g_norm": 0.0013110062573105097,
|
| 1183 |
+
"shape": [
|
| 1184 |
+
384,
|
| 1185 |
+
1152,
|
| 1186 |
+
3,
|
| 1187 |
+
3
|
| 1188 |
+
]
|
| 1189 |
+
},
|
| 1190 |
+
"dec_blocks.2.conv1.bias": {
|
| 1191 |
+
"w_norm": 0.12390010058879852,
|
| 1192 |
+
"g_norm": 2.300715095771011e-05,
|
| 1193 |
+
"shape": [
|
| 1194 |
+
384
|
| 1195 |
+
]
|
| 1196 |
+
},
|
| 1197 |
+
"dec_blocks.2.t_proj.weight": {
|
| 1198 |
+
"w_norm": 11.354249954223633,
|
| 1199 |
+
"g_norm": 2.4828577807056718e-05,
|
| 1200 |
+
"shape": [
|
| 1201 |
+
384,
|
| 1202 |
+
256
|
| 1203 |
+
]
|
| 1204 |
+
},
|
| 1205 |
+
"dec_blocks.2.t_proj.bias": {
|
| 1206 |
+
"w_norm": 0.7089831829071045,
|
| 1207 |
+
"g_norm": 2.300190317328088e-05,
|
| 1208 |
+
"shape": [
|
| 1209 |
+
384
|
| 1210 |
+
]
|
| 1211 |
+
},
|
| 1212 |
+
"dec_blocks.2.norm2.weight": {
|
| 1213 |
+
"w_norm": 19.416515350341797,
|
| 1214 |
+
"g_norm": 5.5451666412409395e-05,
|
| 1215 |
+
"shape": [
|
| 1216 |
+
384
|
| 1217 |
+
]
|
| 1218 |
+
},
|
| 1219 |
+
"dec_blocks.2.norm2.bias": {
|
| 1220 |
+
"w_norm": 0.26346495747566223,
|
| 1221 |
+
"g_norm": 9.812486678129062e-05,
|
| 1222 |
+
"shape": [
|
| 1223 |
+
384
|
| 1224 |
+
]
|
| 1225 |
+
},
|
| 1226 |
+
"dec_blocks.2.conv2.weight": {
|
| 1227 |
+
"w_norm": 11.493674278259277,
|
| 1228 |
+
"g_norm": 0.003377738408744335,
|
| 1229 |
+
"shape": [
|
| 1230 |
+
384,
|
| 1231 |
+
384,
|
| 1232 |
+
3,
|
| 1233 |
+
3
|
| 1234 |
+
]
|
| 1235 |
+
},
|
| 1236 |
+
"dec_blocks.2.conv2.bias": {
|
| 1237 |
+
"w_norm": 0.18863797187805176,
|
| 1238 |
+
"g_norm": 0.0002339720376767218,
|
| 1239 |
+
"shape": [
|
| 1240 |
+
384
|
| 1241 |
+
]
|
| 1242 |
+
},
|
| 1243 |
+
"dec_blocks.2.skip.weight": {
|
| 1244 |
+
"w_norm": 10.904874801635742,
|
| 1245 |
+
"g_norm": 0.04481441527605057,
|
| 1246 |
+
"shape": [
|
| 1247 |
+
384,
|
| 1248 |
+
1152,
|
| 1249 |
+
1,
|
| 1250 |
+
1
|
| 1251 |
+
]
|
| 1252 |
+
},
|
| 1253 |
+
"dec_blocks.2.skip.bias": {
|
| 1254 |
+
"w_norm": 0.3309810757637024,
|
| 1255 |
+
"g_norm": 0.0002339720376767218,
|
| 1256 |
+
"shape": [
|
| 1257 |
+
384
|
| 1258 |
+
]
|
| 1259 |
+
},
|
| 1260 |
+
"dec_blocks.3.norm1.weight": {
|
| 1261 |
+
"w_norm": 27.707651138305664,
|
| 1262 |
+
"g_norm": 0.00013387756189331412,
|
| 1263 |
+
"shape": [
|
| 1264 |
+
768
|
| 1265 |
+
]
|
| 1266 |
+
},
|
| 1267 |
+
"dec_blocks.3.norm1.bias": {
|
| 1268 |
+
"w_norm": 0.09725356847047806,
|
| 1269 |
+
"g_norm": 0.00011039790115319192,
|
| 1270 |
+
"shape": [
|
| 1271 |
+
768
|
| 1272 |
+
]
|
| 1273 |
+
},
|
| 1274 |
+
"dec_blocks.3.conv1.weight": {
|
| 1275 |
+
"w_norm": 12.639841079711914,
|
| 1276 |
+
"g_norm": 0.004064203705638647,
|
| 1277 |
+
"shape": [
|
| 1278 |
+
384,
|
| 1279 |
+
768,
|
| 1280 |
+
3,
|
| 1281 |
+
3
|
| 1282 |
+
]
|
| 1283 |
+
},
|
| 1284 |
+
"dec_blocks.3.conv1.bias": {
|
| 1285 |
+
"w_norm": 0.14569738507270813,
|
| 1286 |
+
"g_norm": 9.590496483724564e-05,
|
| 1287 |
+
"shape": [
|
| 1288 |
+
384
|
| 1289 |
+
]
|
| 1290 |
+
},
|
| 1291 |
+
"dec_blocks.3.t_proj.weight": {
|
| 1292 |
+
"w_norm": 11.424116134643555,
|
| 1293 |
+
"g_norm": 0.00010441212361911312,
|
| 1294 |
+
"shape": [
|
| 1295 |
+
384,
|
| 1296 |
+
256
|
| 1297 |
+
]
|
| 1298 |
+
},
|
| 1299 |
+
"dec_blocks.3.t_proj.bias": {
|
| 1300 |
+
"w_norm": 0.7090743184089661,
|
| 1301 |
+
"g_norm": 9.588411921868101e-05,
|
| 1302 |
+
"shape": [
|
| 1303 |
+
384
|
| 1304 |
+
]
|
| 1305 |
+
},
|
| 1306 |
+
"dec_blocks.3.norm2.weight": {
|
| 1307 |
+
"w_norm": 19.390981674194336,
|
| 1308 |
+
"g_norm": 0.00021730814478360116,
|
| 1309 |
+
"shape": [
|
| 1310 |
+
384
|
| 1311 |
+
]
|
| 1312 |
+
},
|
| 1313 |
+
"dec_blocks.3.norm2.bias": {
|
| 1314 |
+
"w_norm": 0.2664073705673218,
|
| 1315 |
+
"g_norm": 0.00026408053236082196,
|
| 1316 |
+
"shape": [
|
| 1317 |
+
384
|
| 1318 |
+
]
|
| 1319 |
+
},
|
| 1320 |
+
"dec_blocks.3.conv2.weight": {
|
| 1321 |
+
"w_norm": 11.58509349822998,
|
| 1322 |
+
"g_norm": 0.008606428280472755,
|
| 1323 |
+
"shape": [
|
| 1324 |
+
384,
|
| 1325 |
+
384,
|
| 1326 |
+
3,
|
| 1327 |
+
3
|
| 1328 |
+
]
|
| 1329 |
+
},
|
| 1330 |
+
"dec_blocks.3.conv2.bias": {
|
| 1331 |
+
"w_norm": 0.19408394396305084,
|
| 1332 |
+
"g_norm": 0.0005706407828256488,
|
| 1333 |
+
"shape": [
|
| 1334 |
+
384
|
| 1335 |
+
]
|
| 1336 |
+
},
|
| 1337 |
+
"dec_blocks.3.skip.weight": {
|
| 1338 |
+
"w_norm": 11.021206855773926,
|
| 1339 |
+
"g_norm": 0.05618388578295708,
|
| 1340 |
+
"shape": [
|
| 1341 |
+
384,
|
| 1342 |
+
768,
|
| 1343 |
+
1,
|
| 1344 |
+
1
|
| 1345 |
+
]
|
| 1346 |
+
},
|
| 1347 |
+
"dec_blocks.3.skip.bias": {
|
| 1348 |
+
"w_norm": 0.4036846458911896,
|
| 1349 |
+
"g_norm": 0.0005706407828256488,
|
| 1350 |
+
"shape": [
|
| 1351 |
+
384
|
| 1352 |
+
]
|
| 1353 |
+
},
|
| 1354 |
+
"dec_blocks.4.norm1.weight": {
|
| 1355 |
+
"w_norm": 23.96885108947754,
|
| 1356 |
+
"g_norm": 0.00025959074264392257,
|
| 1357 |
+
"shape": [
|
| 1358 |
+
576
|
| 1359 |
+
]
|
| 1360 |
+
},
|
| 1361 |
+
"dec_blocks.4.norm1.bias": {
|
| 1362 |
+
"w_norm": 0.11969521641731262,
|
| 1363 |
+
"g_norm": 0.00022613562759943306,
|
| 1364 |
+
"shape": [
|
| 1365 |
+
576
|
| 1366 |
+
]
|
| 1367 |
+
},
|
| 1368 |
+
"dec_blocks.4.conv1.weight": {
|
| 1369 |
+
"w_norm": 8.764872550964355,
|
| 1370 |
+
"g_norm": 0.009391983039677143,
|
| 1371 |
+
"shape": [
|
| 1372 |
+
192,
|
| 1373 |
+
576,
|
| 1374 |
+
3,
|
| 1375 |
+
3
|
| 1376 |
+
]
|
| 1377 |
+
},
|
| 1378 |
+
"dec_blocks.4.conv1.bias": {
|
| 1379 |
+
"w_norm": 0.11739768087863922,
|
| 1380 |
+
"g_norm": 0.00023739114112686366,
|
| 1381 |
+
"shape": [
|
| 1382 |
+
192
|
| 1383 |
+
]
|
| 1384 |
+
},
|
| 1385 |
+
"dec_blocks.4.t_proj.weight": {
|
| 1386 |
+
"w_norm": 8.095752716064453,
|
| 1387 |
+
"g_norm": 0.00026289356173947453,
|
| 1388 |
+
"shape": [
|
| 1389 |
+
192,
|
| 1390 |
+
256
|
| 1391 |
+
]
|
| 1392 |
+
},
|
| 1393 |
+
"dec_blocks.4.t_proj.bias": {
|
| 1394 |
+
"w_norm": 0.4813956618309021,
|
| 1395 |
+
"g_norm": 0.0002374465111643076,
|
| 1396 |
+
"shape": [
|
| 1397 |
+
192
|
| 1398 |
+
]
|
| 1399 |
+
},
|
| 1400 |
+
"dec_blocks.4.norm2.weight": {
|
| 1401 |
+
"w_norm": 13.783257484436035,
|
| 1402 |
+
"g_norm": 0.0007113862666301429,
|
| 1403 |
+
"shape": [
|
| 1404 |
+
192
|
| 1405 |
+
]
|
| 1406 |
+
},
|
| 1407 |
+
"dec_blocks.4.norm2.bias": {
|
| 1408 |
+
"w_norm": 0.1056097000837326,
|
| 1409 |
+
"g_norm": 0.0006580191547982395,
|
| 1410 |
+
"shape": [
|
| 1411 |
+
192
|
| 1412 |
+
]
|
| 1413 |
+
},
|
| 1414 |
+
"dec_blocks.4.conv2.weight": {
|
| 1415 |
+
"w_norm": 8.158761978149414,
|
| 1416 |
+
"g_norm": 0.02111119031906128,
|
| 1417 |
+
"shape": [
|
| 1418 |
+
192,
|
| 1419 |
+
192,
|
| 1420 |
+
3,
|
| 1421 |
+
3
|
| 1422 |
+
]
|
| 1423 |
+
},
|
| 1424 |
+
"dec_blocks.4.conv2.bias": {
|
| 1425 |
+
"w_norm": 0.19651232659816742,
|
| 1426 |
+
"g_norm": 0.0013463678769767284,
|
| 1427 |
+
"shape": [
|
| 1428 |
+
192
|
| 1429 |
+
]
|
| 1430 |
+
},
|
| 1431 |
+
"dec_blocks.4.skip.weight": {
|
| 1432 |
+
"w_norm": 7.813741683959961,
|
| 1433 |
+
"g_norm": 0.09076514840126038,
|
| 1434 |
+
"shape": [
|
| 1435 |
+
192,
|
| 1436 |
+
576,
|
| 1437 |
+
1,
|
| 1438 |
+
1
|
| 1439 |
+
]
|
| 1440 |
+
},
|
| 1441 |
+
"dec_blocks.4.skip.bias": {
|
| 1442 |
+
"w_norm": 0.3467942774295807,
|
| 1443 |
+
"g_norm": 0.0013463678769767284,
|
| 1444 |
+
"shape": [
|
| 1445 |
+
192
|
| 1446 |
+
]
|
| 1447 |
+
},
|
| 1448 |
+
"dec_blocks.5.norm1.weight": {
|
| 1449 |
+
"w_norm": 19.519624710083008,
|
| 1450 |
+
"g_norm": 0.011791425757110119,
|
| 1451 |
+
"shape": [
|
| 1452 |
+
384
|
| 1453 |
+
]
|
| 1454 |
+
},
|
| 1455 |
+
"dec_blocks.5.norm1.bias": {
|
| 1456 |
+
"w_norm": 0.14571304619312286,
|
| 1457 |
+
"g_norm": 0.013299982994794846,
|
| 1458 |
+
"shape": [
|
| 1459 |
+
384
|
| 1460 |
+
]
|
| 1461 |
+
},
|
| 1462 |
+
"dec_blocks.5.conv1.weight": {
|
| 1463 |
+
"w_norm": 8.778403282165527,
|
| 1464 |
+
"g_norm": 0.08446153998374939,
|
| 1465 |
+
"shape": [
|
| 1466 |
+
192,
|
| 1467 |
+
384,
|
| 1468 |
+
3,
|
| 1469 |
+
3
|
| 1470 |
+
]
|
| 1471 |
+
},
|
| 1472 |
+
"dec_blocks.5.conv1.bias": {
|
| 1473 |
+
"w_norm": 0.13901221752166748,
|
| 1474 |
+
"g_norm": 0.00463363341987133,
|
| 1475 |
+
"shape": [
|
| 1476 |
+
192
|
| 1477 |
+
]
|
| 1478 |
+
},
|
| 1479 |
+
"dec_blocks.5.t_proj.weight": {
|
| 1480 |
+
"w_norm": 8.244766235351562,
|
| 1481 |
+
"g_norm": 0.005314541980624199,
|
| 1482 |
+
"shape": [
|
| 1483 |
+
192,
|
| 1484 |
+
256
|
| 1485 |
+
]
|
| 1486 |
+
},
|
| 1487 |
+
"dec_blocks.5.t_proj.bias": {
|
| 1488 |
+
"w_norm": 0.5051465034484863,
|
| 1489 |
+
"g_norm": 0.0046348548494279385,
|
| 1490 |
+
"shape": [
|
| 1491 |
+
192
|
| 1492 |
+
]
|
| 1493 |
+
},
|
| 1494 |
+
"dec_blocks.5.norm2.weight": {
|
| 1495 |
+
"w_norm": 13.902420043945312,
|
| 1496 |
+
"g_norm": 0.0054893940687179565,
|
| 1497 |
+
"shape": [
|
| 1498 |
+
192
|
| 1499 |
+
]
|
| 1500 |
+
},
|
| 1501 |
+
"dec_blocks.5.norm2.bias": {
|
| 1502 |
+
"w_norm": 0.04917939752340317,
|
| 1503 |
+
"g_norm": 0.0060037486255168915,
|
| 1504 |
+
"shape": [
|
| 1505 |
+
192
|
| 1506 |
+
]
|
| 1507 |
+
},
|
| 1508 |
+
"dec_blocks.5.conv2.weight": {
|
| 1509 |
+
"w_norm": 8.520387649536133,
|
| 1510 |
+
"g_norm": 0.0472685769200325,
|
| 1511 |
+
"shape": [
|
| 1512 |
+
192,
|
| 1513 |
+
192,
|
| 1514 |
+
3,
|
| 1515 |
+
3
|
| 1516 |
+
]
|
| 1517 |
+
},
|
| 1518 |
+
"dec_blocks.5.conv2.bias": {
|
| 1519 |
+
"w_norm": 0.2030690759420395,
|
| 1520 |
+
"g_norm": 0.0030816341750323772,
|
| 1521 |
+
"shape": [
|
| 1522 |
+
192
|
| 1523 |
+
]
|
| 1524 |
+
},
|
| 1525 |
+
"dec_blocks.5.skip.weight": {
|
| 1526 |
+
"w_norm": 8.117816925048828,
|
| 1527 |
+
"g_norm": 0.08592767268419266,
|
| 1528 |
+
"shape": [
|
| 1529 |
+
192,
|
| 1530 |
+
384,
|
| 1531 |
+
1,
|
| 1532 |
+
1
|
| 1533 |
+
]
|
| 1534 |
+
},
|
| 1535 |
+
"dec_blocks.5.skip.bias": {
|
| 1536 |
+
"w_norm": 0.42368704080581665,
|
| 1537 |
+
"g_norm": 0.0030816341750323772,
|
| 1538 |
+
"shape": [
|
| 1539 |
+
192
|
| 1540 |
+
]
|
| 1541 |
+
},
|
| 1542 |
+
"dec_attns.0.norm.weight": {
|
| 1543 |
+
"w_norm": 27.58348846435547,
|
| 1544 |
+
"g_norm": 0.00026784223155118525,
|
| 1545 |
+
"shape": [
|
| 1546 |
+
768
|
| 1547 |
+
]
|
| 1548 |
+
},
|
| 1549 |
+
"dec_attns.0.norm.bias": {
|
| 1550 |
+
"w_norm": 0.05455893278121948,
|
| 1551 |
+
"g_norm": 0.00027358514489606023,
|
| 1552 |
+
"shape": [
|
| 1553 |
+
768
|
| 1554 |
+
]
|
| 1555 |
+
},
|
| 1556 |
+
"dec_attns.0.qkv.weight": {
|
| 1557 |
+
"w_norm": 28.11470603942871,
|
| 1558 |
+
"g_norm": 0.006553055718541145,
|
| 1559 |
+
"shape": [
|
| 1560 |
+
2304,
|
| 1561 |
+
768,
|
| 1562 |
+
1,
|
| 1563 |
+
1
|
| 1564 |
+
]
|
| 1565 |
+
},
|
| 1566 |
+
"dec_attns.0.qkv.bias": {
|
| 1567 |
+
"w_norm": 1.0048754215240479,
|
| 1568 |
+
"g_norm": 0.00023312140547204763,
|
| 1569 |
+
"shape": [
|
| 1570 |
+
2304
|
| 1571 |
+
]
|
| 1572 |
+
},
|
| 1573 |
+
"dec_attns.0.proj.weight": {
|
| 1574 |
+
"w_norm": 16.170562744140625,
|
| 1575 |
+
"g_norm": 0.005232008174061775,
|
| 1576 |
+
"shape": [
|
| 1577 |
+
768,
|
| 1578 |
+
768,
|
| 1579 |
+
1,
|
| 1580 |
+
1
|
| 1581 |
+
]
|
| 1582 |
+
},
|
| 1583 |
+
"dec_attns.0.proj.bias": {
|
| 1584 |
+
"w_norm": 0.5775165557861328,
|
| 1585 |
+
"g_norm": 0.00028337317053228617,
|
| 1586 |
+
"shape": [
|
| 1587 |
+
768
|
| 1588 |
+
]
|
| 1589 |
+
},
|
| 1590 |
+
"dec_attns.1.norm.weight": {
|
| 1591 |
+
"w_norm": 27.570045471191406,
|
| 1592 |
+
"g_norm": 0.000315696292091161,
|
| 1593 |
+
"shape": [
|
| 1594 |
+
768
|
| 1595 |
+
]
|
| 1596 |
+
},
|
| 1597 |
+
"dec_attns.1.norm.bias": {
|
| 1598 |
+
"w_norm": 0.05546895042061806,
|
| 1599 |
+
"g_norm": 0.0002729733823798597,
|
| 1600 |
+
"shape": [
|
| 1601 |
+
768
|
| 1602 |
+
]
|
| 1603 |
+
},
|
| 1604 |
+
"dec_attns.1.qkv.weight": {
|
| 1605 |
+
"w_norm": 28.15424156188965,
|
| 1606 |
+
"g_norm": 0.008273596875369549,
|
| 1607 |
+
"shape": [
|
| 1608 |
+
2304,
|
| 1609 |
+
768,
|
| 1610 |
+
1,
|
| 1611 |
+
1
|
| 1612 |
+
]
|
| 1613 |
+
},
|
| 1614 |
+
"dec_attns.1.qkv.bias": {
|
| 1615 |
+
"w_norm": 1.0058194398880005,
|
| 1616 |
+
"g_norm": 0.00025404043844901025,
|
| 1617 |
+
"shape": [
|
| 1618 |
+
2304
|
| 1619 |
+
]
|
| 1620 |
+
},
|
| 1621 |
+
"dec_attns.1.proj.weight": {
|
| 1622 |
+
"w_norm": 16.085893630981445,
|
| 1623 |
+
"g_norm": 0.007718106731772423,
|
| 1624 |
+
"shape": [
|
| 1625 |
+
768,
|
| 1626 |
+
768,
|
| 1627 |
+
1,
|
| 1628 |
+
1
|
| 1629 |
+
]
|
| 1630 |
+
},
|
| 1631 |
+
"dec_attns.1.proj.bias": {
|
| 1632 |
+
"w_norm": 0.5793389081954956,
|
| 1633 |
+
"g_norm": 0.0003180239873472601,
|
| 1634 |
+
"shape": [
|
| 1635 |
+
768
|
| 1636 |
+
]
|
| 1637 |
+
},
|
| 1638 |
+
"dec_attns.2.norm.weight": {
|
| 1639 |
+
"w_norm": 19.477664947509766,
|
| 1640 |
+
"g_norm": 9.047827188624069e-05,
|
| 1641 |
+
"shape": [
|
| 1642 |
+
384
|
| 1643 |
+
]
|
| 1644 |
+
},
|
| 1645 |
+
"dec_attns.2.norm.bias": {
|
| 1646 |
+
"w_norm": 0.031263552606105804,
|
| 1647 |
+
"g_norm": 8.09360426501371e-05,
|
| 1648 |
+
"shape": [
|
| 1649 |
+
384
|
| 1650 |
+
]
|
| 1651 |
+
},
|
| 1652 |
+
"dec_attns.2.qkv.weight": {
|
| 1653 |
+
"w_norm": 19.72080421447754,
|
| 1654 |
+
"g_norm": 0.002651409013196826,
|
| 1655 |
+
"shape": [
|
| 1656 |
+
1152,
|
| 1657 |
+
384,
|
| 1658 |
+
1,
|
| 1659 |
+
1
|
| 1660 |
+
]
|
| 1661 |
+
},
|
| 1662 |
+
"dec_attns.2.qkv.bias": {
|
| 1663 |
+
"w_norm": 0.9826018810272217,
|
| 1664 |
+
"g_norm": 0.00012897119449917227,
|
| 1665 |
+
"shape": [
|
| 1666 |
+
1152
|
| 1667 |
+
]
|
| 1668 |
+
},
|
| 1669 |
+
"dec_attns.2.proj.weight": {
|
| 1670 |
+
"w_norm": 11.312171936035156,
|
| 1671 |
+
"g_norm": 0.002945007523521781,
|
| 1672 |
+
"shape": [
|
| 1673 |
+
384,
|
| 1674 |
+
384,
|
| 1675 |
+
1,
|
| 1676 |
+
1
|
| 1677 |
+
]
|
| 1678 |
+
},
|
| 1679 |
+
"dec_attns.2.proj.bias": {
|
| 1680 |
+
"w_norm": 0.5890069007873535,
|
| 1681 |
+
"g_norm": 0.00023511593462899327,
|
| 1682 |
+
"shape": [
|
| 1683 |
+
384
|
| 1684 |
+
]
|
| 1685 |
+
},
|
| 1686 |
+
"dec_ups.0.op.weight": {
|
| 1687 |
+
"w_norm": 16.251949310302734,
|
| 1688 |
+
"g_norm": 0.02308141626417637,
|
| 1689 |
+
"shape": [
|
| 1690 |
+
768,
|
| 1691 |
+
768,
|
| 1692 |
+
3,
|
| 1693 |
+
3
|
| 1694 |
+
]
|
| 1695 |
+
},
|
| 1696 |
+
"dec_ups.0.op.bias": {
|
| 1697 |
+
"w_norm": 0.20316992700099945,
|
| 1698 |
+
"g_norm": 0.00010308789205737412,
|
| 1699 |
+
"shape": [
|
| 1700 |
+
768
|
| 1701 |
+
]
|
| 1702 |
+
},
|
| 1703 |
+
"dec_ups.1.op.weight": {
|
| 1704 |
+
"w_norm": 16.251108169555664,
|
| 1705 |
+
"g_norm": 0.03540193662047386,
|
| 1706 |
+
"shape": [
|
| 1707 |
+
768,
|
| 1708 |
+
768,
|
| 1709 |
+
3,
|
| 1710 |
+
3
|
| 1711 |
+
]
|
| 1712 |
+
},
|
| 1713 |
+
"dec_ups.1.op.bias": {
|
| 1714 |
+
"w_norm": 0.1951359361410141,
|
| 1715 |
+
"g_norm": 0.0001347730722045526,
|
| 1716 |
+
"shape": [
|
| 1717 |
+
768
|
| 1718 |
+
]
|
| 1719 |
+
},
|
| 1720 |
+
"dec_ups.2.op.weight": {
|
| 1721 |
+
"w_norm": 11.344901084899902,
|
| 1722 |
+
"g_norm": 0.06848753243684769,
|
| 1723 |
+
"shape": [
|
| 1724 |
+
384,
|
| 1725 |
+
384,
|
| 1726 |
+
3,
|
| 1727 |
+
3
|
| 1728 |
+
]
|
| 1729 |
+
},
|
| 1730 |
+
"dec_ups.2.op.bias": {
|
| 1731 |
+
"w_norm": 0.19678521156311035,
|
| 1732 |
+
"g_norm": 0.00026137454551644623,
|
| 1733 |
+
"shape": [
|
| 1734 |
+
384
|
| 1735 |
+
]
|
| 1736 |
+
},
|
| 1737 |
+
"dec_ups.3.op.weight": {
|
| 1738 |
+
"w_norm": 11.392304420471191,
|
| 1739 |
+
"g_norm": 0.0678146630525589,
|
| 1740 |
+
"shape": [
|
| 1741 |
+
384,
|
| 1742 |
+
384,
|
| 1743 |
+
3,
|
| 1744 |
+
3
|
| 1745 |
+
]
|
| 1746 |
+
},
|
| 1747 |
+
"dec_ups.3.op.bias": {
|
| 1748 |
+
"w_norm": 0.2005225270986557,
|
| 1749 |
+
"g_norm": 0.0004707111802417785,
|
| 1750 |
+
"shape": [
|
| 1751 |
+
384
|
| 1752 |
+
]
|
| 1753 |
+
},
|
| 1754 |
+
"dec_ups.4.op.weight": {
|
| 1755 |
+
"w_norm": 8.021990776062012,
|
| 1756 |
+
"g_norm": 0.0939137414097786,
|
| 1757 |
+
"shape": [
|
| 1758 |
+
192,
|
| 1759 |
+
192,
|
| 1760 |
+
3,
|
| 1761 |
+
3
|
| 1762 |
+
]
|
| 1763 |
+
},
|
| 1764 |
+
"dec_ups.4.op.bias": {
|
| 1765 |
+
"w_norm": 0.19830958545207977,
|
| 1766 |
+
"g_norm": 0.0010575436754152179,
|
| 1767 |
+
"shape": [
|
| 1768 |
+
192
|
| 1769 |
+
]
|
| 1770 |
+
},
|
| 1771 |
+
"out_norm.weight": {
|
| 1772 |
+
"w_norm": 13.740449905395508,
|
| 1773 |
+
"g_norm": 0.009818742983043194,
|
| 1774 |
+
"shape": [
|
| 1775 |
+
192
|
| 1776 |
+
]
|
| 1777 |
+
},
|
| 1778 |
+
"out_norm.bias": {
|
| 1779 |
+
"w_norm": 0.13754217326641083,
|
| 1780 |
+
"g_norm": 0.013854917138814926,
|
| 1781 |
+
"shape": [
|
| 1782 |
+
192
|
| 1783 |
+
]
|
| 1784 |
+
},
|
| 1785 |
+
"out_conv.weight": {
|
| 1786 |
+
"w_norm": 0.15520590543746948,
|
| 1787 |
+
"g_norm": 0.785935640335083,
|
| 1788 |
+
"shape": [
|
| 1789 |
+
3,
|
| 1790 |
+
192,
|
| 1791 |
+
3,
|
| 1792 |
+
3
|
| 1793 |
+
]
|
| 1794 |
+
},
|
| 1795 |
+
"out_conv.bias": {
|
| 1796 |
+
"w_norm": 0.009202906861901283,
|
| 1797 |
+
"g_norm": 0.06661342829465866,
|
| 1798 |
+
"shape": [
|
| 1799 |
+
3
|
| 1800 |
+
]
|
| 1801 |
+
}
|
| 1802 |
+
},
|
| 1803 |
+
"activations": {
|
| 1804 |
+
"input_conv": {
|
| 1805 |
+
"mean": 0.007421940565109253,
|
| 1806 |
+
"std": 0.45184287428855896,
|
| 1807 |
+
"abs_max": 3.21875,
|
| 1808 |
+
"frac_dead": 0.00039866898441687226,
|
| 1809 |
+
"frac_sat": 0.0
|
| 1810 |
+
},
|
| 1811 |
+
"time_mlp.1": {
|
| 1812 |
+
"mean": 0.007643388584256172,
|
| 1813 |
+
"std": 0.2923099100589752,
|
| 1814 |
+
"abs_max": 4.53125,
|
| 1815 |
+
"frac_dead": 4.069010537932627e-05,
|
| 1816 |
+
"frac_sat": 0.0
|
| 1817 |
+
},
|
| 1818 |
+
"time_mlp.3": {
|
| 1819 |
+
"mean": 0.025249801576137543,
|
| 1820 |
+
"std": 0.23188936710357666,
|
| 1821 |
+
"abs_max": 1.1484375,
|
| 1822 |
+
"frac_dead": 0.00016276042151730508,
|
| 1823 |
+
"frac_sat": 0.0
|
| 1824 |
+
},
|
| 1825 |
+
"enc_blocks.0.norm1": {
|
| 1826 |
+
"mean": -0.007203842978924513,
|
| 1827 |
+
"std": 0.9947881102561951,
|
| 1828 |
+
"abs_max": 6.314060688018799,
|
| 1829 |
+
"frac_dead": 4.072984154390724e-07,
|
| 1830 |
+
"frac_sat": 0.0
|
| 1831 |
+
},
|
| 1832 |
+
"enc_blocks.0.conv1": {
|
| 1833 |
+
"mean": 0.1297854334115982,
|
| 1834 |
+
"std": 1.0878294706344604,
|
| 1835 |
+
"abs_max": 10.1875,
|
| 1836 |
+
"frac_dead": 3.8295984268188477e-05,
|
| 1837 |
+
"frac_sat": 3.311369178859991e-09
|
| 1838 |
+
},
|
| 1839 |
+
"enc_blocks.0.t_proj": {
|
| 1840 |
+
"mean": 0.0019384645856916904,
|
| 1841 |
+
"std": 0.2523488402366638,
|
| 1842 |
+
"abs_max": 2.78125,
|
| 1843 |
+
"frac_dead": 0.0,
|
| 1844 |
+
"frac_sat": 0.0
|
| 1845 |
+
},
|
| 1846 |
+
"enc_blocks.0.norm2": {
|
| 1847 |
+
"mean": -0.005775494500994682,
|
| 1848 |
+
"std": 0.9985000491142273,
|
| 1849 |
+
"abs_max": 9.170462608337402,
|
| 1850 |
+
"frac_dead": 8.245309004450974e-07,
|
| 1851 |
+
"frac_sat": 0.0
|
| 1852 |
+
},
|
| 1853 |
+
"enc_blocks.0.conv2": {
|
| 1854 |
+
"mean": -0.035144317895174026,
|
| 1855 |
+
"std": 1.0234105587005615,
|
| 1856 |
+
"abs_max": 13.1875,
|
| 1857 |
+
"frac_dead": 3.355410444783047e-05,
|
| 1858 |
+
"frac_sat": 1.430511474609375e-06
|
| 1859 |
+
},
|
| 1860 |
+
"enc_downs.0.op": {
|
| 1861 |
+
"mean": -0.014735549688339233,
|
| 1862 |
+
"std": 2.3309483528137207,
|
| 1863 |
+
"abs_max": 33.25,
|
| 1864 |
+
"frac_dead": 1.627869096409995e-05,
|
| 1865 |
+
"frac_sat": 0.0026156241074204445
|
| 1866 |
+
},
|
| 1867 |
+
"enc_blocks.1.norm1": {
|
| 1868 |
+
"mean": -0.0033612640108913183,
|
| 1869 |
+
"std": 0.999249279499054,
|
| 1870 |
+
"abs_max": 13.021814346313477,
|
| 1871 |
+
"frac_dead": 5.828009648212173e-07,
|
| 1872 |
+
"frac_sat": 1.5894572413799324e-07
|
| 1873 |
+
},
|
| 1874 |
+
"enc_blocks.1.conv1": {
|
| 1875 |
+
"mean": 0.024152027443051338,
|
| 1876 |
+
"std": 1.0423327684402466,
|
| 1877 |
+
"abs_max": 15.3125,
|
| 1878 |
+
"frac_dead": 3.337860107421875e-05,
|
| 1879 |
+
"frac_sat": 7.973777428560425e-06
|
| 1880 |
+
},
|
| 1881 |
+
"enc_blocks.1.t_proj": {
|
| 1882 |
+
"mean": 0.006402192637324333,
|
| 1883 |
+
"std": 0.3085084557533264,
|
| 1884 |
+
"abs_max": 2.359375,
|
| 1885 |
+
"frac_dead": 0.0,
|
| 1886 |
+
"frac_sat": 0.0
|
| 1887 |
+
},
|
| 1888 |
+
"enc_blocks.1.norm2": {
|
| 1889 |
+
"mean": -0.00576779106631875,
|
| 1890 |
+
"std": 0.9974737763404846,
|
| 1891 |
+
"abs_max": 15.784193992614746,
|
| 1892 |
+
"frac_dead": 2.3444492853741394e-06,
|
| 1893 |
+
"frac_sat": 1.4702478665640228e-06
|
| 1894 |
+
},
|
| 1895 |
+
"enc_blocks.1.conv2": {
|
| 1896 |
+
"mean": -0.022583598271012306,
|
| 1897 |
+
"std": 0.932984471321106,
|
| 1898 |
+
"abs_max": 12.3125,
|
| 1899 |
+
"frac_dead": 3.710058081196621e-05,
|
| 1900 |
+
"frac_sat": 3.8411883451772155e-07
|
| 1901 |
+
},
|
| 1902 |
+
"enc_downs.1.op": {
|
| 1903 |
+
"mean": -0.06834577023983002,
|
| 1904 |
+
"std": 2.8865416049957275,
|
| 1905 |
+
"abs_max": 37.0,
|
| 1906 |
+
"frac_dead": 1.2132856682001147e-05,
|
| 1907 |
+
"frac_sat": 0.006111780647188425
|
| 1908 |
+
},
|
| 1909 |
+
"enc_blocks.2.norm1": {
|
| 1910 |
+
"mean": -0.004561815410852432,
|
| 1911 |
+
"std": 0.9969245791435242,
|
| 1912 |
+
"abs_max": 9.989190101623535,
|
| 1913 |
+
"frac_dead": 2.649095449669403e-07,
|
| 1914 |
+
"frac_sat": 0.0
|
| 1915 |
+
},
|
| 1916 |
+
"enc_blocks.2.conv1": {
|
| 1917 |
+
"mean": -0.1135186105966568,
|
| 1918 |
+
"std": 1.2132693529129028,
|
| 1919 |
+
"abs_max": 18.625,
|
| 1920 |
+
"frac_dead": 4.283587259124033e-05,
|
| 1921 |
+
"frac_sat": 0.00017280048632528633
|
| 1922 |
+
},
|
| 1923 |
+
"enc_blocks.2.t_proj": {
|
| 1924 |
+
"mean": -0.0073701548390090466,
|
| 1925 |
+
"std": 0.4103415012359619,
|
| 1926 |
+
"abs_max": 3.734375,
|
| 1927 |
+
"frac_dead": 0.0,
|
| 1928 |
+
"frac_sat": 0.0
|
| 1929 |
+
},
|
| 1930 |
+
"enc_blocks.2.norm2": {
|
| 1931 |
+
"mean": -0.014336032792925835,
|
| 1932 |
+
"std": 0.9936417937278748,
|
| 1933 |
+
"abs_max": 12.799908638000488,
|
| 1934 |
+
"frac_dead": 5.033281240685028e-07,
|
| 1935 |
+
"frac_sat": 7.523430667788489e-06
|
| 1936 |
+
},
|
| 1937 |
+
"enc_blocks.2.conv2": {
|
| 1938 |
+
"mean": -0.015473869629204273,
|
| 1939 |
+
"std": 1.0459991693496704,
|
| 1940 |
+
"abs_max": 14.1875,
|
| 1941 |
+
"frac_dead": 2.0371542632346973e-05,
|
| 1942 |
+
"frac_sat": 2.9934776648588013e-06
|
| 1943 |
+
},
|
| 1944 |
+
"enc_blocks.2.skip": {
|
| 1945 |
+
"mean": -0.03707383945584297,
|
| 1946 |
+
"std": 1.4708917140960693,
|
| 1947 |
+
"abs_max": 17.25,
|
| 1948 |
+
"frac_dead": 7.65588556532748e-05,
|
| 1949 |
+
"frac_sat": 4.4345855712890625e-05
|
| 1950 |
+
},
|
| 1951 |
+
"enc_downs.2.op": {
|
| 1952 |
+
"mean": -0.27560582756996155,
|
| 1953 |
+
"std": 2.0658843517303467,
|
| 1954 |
+
"abs_max": 35.0,
|
| 1955 |
+
"frac_dead": 1.0914272934314795e-05,
|
| 1956 |
+
"frac_sat": 0.0011461046524345875
|
| 1957 |
+
},
|
| 1958 |
+
"enc_blocks.3.norm1": {
|
| 1959 |
+
"mean": -0.002862681169062853,
|
| 1960 |
+
"std": 0.9991037249565125,
|
| 1961 |
+
"abs_max": 9.527902603149414,
|
| 1962 |
+
"frac_dead": 6.35782896551973e-07,
|
| 1963 |
+
"frac_sat": 0.0
|
| 1964 |
+
},
|
| 1965 |
+
"enc_blocks.3.conv1": {
|
| 1966 |
+
"mean": -0.22325478494167328,
|
| 1967 |
+
"std": 1.7933238744735718,
|
| 1968 |
+
"abs_max": 34.75,
|
| 1969 |
+
"frac_dead": 1.706017428659834e-05,
|
| 1970 |
+
"frac_sat": 0.0021811590995639563
|
| 1971 |
+
},
|
| 1972 |
+
"enc_blocks.3.t_proj": {
|
| 1973 |
+
"mean": 0.001510739792138338,
|
| 1974 |
+
"std": 0.1843785047531128,
|
| 1975 |
+
"abs_max": 1.34375,
|
| 1976 |
+
"frac_dead": 0.0,
|
| 1977 |
+
"frac_sat": 0.0
|
| 1978 |
+
},
|
| 1979 |
+
"enc_blocks.3.norm2": {
|
| 1980 |
+
"mean": -0.01221651490777731,
|
| 1981 |
+
"std": 0.990524411201477,
|
| 1982 |
+
"abs_max": 19.570201873779297,
|
| 1983 |
+
"frac_dead": 8.477105097881577e-07,
|
| 1984 |
+
"frac_sat": 4.76837158203125e-05
|
| 1985 |
+
},
|
| 1986 |
+
"enc_blocks.3.conv2": {
|
| 1987 |
+
"mean": 0.11187144368886948,
|
| 1988 |
+
"std": 1.2421133518218994,
|
| 1989 |
+
"abs_max": 11.3125,
|
| 1990 |
+
"frac_dead": 2.3312039047596045e-05,
|
| 1991 |
+
"frac_sat": 1.9073486328125e-06
|
| 1992 |
+
},
|
| 1993 |
+
"enc_attns.3.norm": {
|
| 1994 |
+
"mean": 0.00036654993891716003,
|
| 1995 |
+
"std": 0.9955237507820129,
|
| 1996 |
+
"abs_max": 10.021247863769531,
|
| 1997 |
+
"frac_dead": 9.5367431640625e-07,
|
| 1998 |
+
"frac_sat": 1.059638137235197e-07
|
| 1999 |
+
},
|
| 2000 |
+
"enc_attns.3.qkv": {
|
| 2001 |
+
"mean": 0.005369497928768396,
|
| 2002 |
+
"std": 0.6125555634498596,
|
| 2003 |
+
"abs_max": 6.71875,
|
| 2004 |
+
"frac_dead": 0.00010663492139428854,
|
| 2005 |
+
"frac_sat": 0.0
|
| 2006 |
+
},
|
| 2007 |
+
"enc_attns.3.proj": {
|
| 2008 |
+
"mean": 0.0026778418105095625,
|
| 2009 |
+
"std": 0.3171274662017822,
|
| 2010 |
+
"abs_max": 1.953125,
|
| 2011 |
+
"frac_dead": 0.00018797980737872422,
|
| 2012 |
+
"frac_sat": 0.0
|
| 2013 |
+
},
|
| 2014 |
+
"enc_downs.3.op": {
|
| 2015 |
+
"mean": -0.23350858688354492,
|
| 2016 |
+
"std": 2.7067246437072754,
|
| 2017 |
+
"abs_max": 41.25,
|
| 2018 |
+
"frac_dead": 1.229180270456709e-05,
|
| 2019 |
+
"frac_sat": 0.004200829658657312
|
| 2020 |
+
},
|
| 2021 |
+
"enc_blocks.4.norm1": {
|
| 2022 |
+
"mean": -0.00010409951210021973,
|
| 2023 |
+
"std": 1.0004417896270752,
|
| 2024 |
+
"abs_max": 9.303688049316406,
|
| 2025 |
+
"frac_dead": 1.6954210195763153e-06,
|
| 2026 |
+
"frac_sat": 0.0
|
| 2027 |
+
},
|
| 2028 |
+
"enc_blocks.4.conv1": {
|
| 2029 |
+
"mean": -0.30328503251075745,
|
| 2030 |
+
"std": 1.6799970865249634,
|
| 2031 |
+
"abs_max": 31.625,
|
| 2032 |
+
"frac_dead": 2.373589450144209e-05,
|
| 2033 |
+
"frac_sat": 0.0029724969062954187
|
| 2034 |
+
},
|
| 2035 |
+
"enc_blocks.4.t_proj": {
|
| 2036 |
+
"mean": -0.0012046759948134422,
|
| 2037 |
+
"std": 0.11991295218467712,
|
| 2038 |
+
"abs_max": 1.203125,
|
| 2039 |
+
"frac_dead": 0.0,
|
| 2040 |
+
"frac_sat": 0.0
|
| 2041 |
+
},
|
| 2042 |
+
"enc_blocks.4.norm2": {
|
| 2043 |
+
"mean": -0.01589149236679077,
|
| 2044 |
+
"std": 0.9891142845153809,
|
| 2045 |
+
"abs_max": 17.085819244384766,
|
| 2046 |
+
"frac_dead": 1.0596381798677612e-06,
|
| 2047 |
+
"frac_sat": 6.76049166941084e-05
|
| 2048 |
+
},
|
| 2049 |
+
"enc_blocks.4.conv2": {
|
| 2050 |
+
"mean": 0.14119984209537506,
|
| 2051 |
+
"std": 1.3356584310531616,
|
| 2052 |
+
"abs_max": 8.9375,
|
| 2053 |
+
"frac_dead": 1.229180270456709e-05,
|
| 2054 |
+
"frac_sat": 0.0
|
| 2055 |
+
},
|
| 2056 |
+
"enc_blocks.4.skip": {
|
| 2057 |
+
"mean": -0.0065156240016222,
|
| 2058 |
+
"std": 1.5490919351577759,
|
| 2059 |
+
"abs_max": 31.0,
|
| 2060 |
+
"frac_dead": 4.789564263774082e-05,
|
| 2061 |
+
"frac_sat": 0.00041135153151117265
|
| 2062 |
+
},
|
| 2063 |
+
"enc_attns.4.norm": {
|
| 2064 |
+
"mean": 0.00013783077883999795,
|
| 2065 |
+
"std": 0.994690477848053,
|
| 2066 |
+
"abs_max": 10.197798728942871,
|
| 2067 |
+
"frac_dead": 8.477105097881577e-07,
|
| 2068 |
+
"frac_sat": 4.238552548940788e-07
|
| 2069 |
+
},
|
| 2070 |
+
"enc_attns.4.qkv": {
|
| 2071 |
+
"mean": 0.012843161821365356,
|
| 2072 |
+
"std": 0.6926348209381104,
|
| 2073 |
+
"abs_max": 8.1875,
|
| 2074 |
+
"frac_dead": 6.852326623629779e-05,
|
| 2075 |
+
"frac_sat": 0.0
|
| 2076 |
+
},
|
| 2077 |
+
"enc_attns.4.proj": {
|
| 2078 |
+
"mean": 0.002519653644412756,
|
| 2079 |
+
"std": 0.4470694065093994,
|
| 2080 |
+
"abs_max": 5.0,
|
| 2081 |
+
"frac_dead": 0.00011083814752055332,
|
| 2082 |
+
"frac_sat": 0.0
|
| 2083 |
+
},
|
| 2084 |
+
"enc_downs.4.op": {
|
| 2085 |
+
"mean": 0.2122226357460022,
|
| 2086 |
+
"std": 3.1655666828155518,
|
| 2087 |
+
"abs_max": 45.0,
|
| 2088 |
+
"frac_dead": 3.3908420391526306e-06,
|
| 2089 |
+
"frac_sat": 0.006535000167787075
|
| 2090 |
+
},
|
| 2091 |
+
"enc_blocks.5.norm1": {
|
| 2092 |
+
"mean": 0.00028194073820486665,
|
| 2093 |
+
"std": 1.0010477304458618,
|
| 2094 |
+
"abs_max": 9.194173812866211,
|
| 2095 |
+
"frac_dead": 0.0,
|
| 2096 |
+
"frac_sat": 0.0
|
| 2097 |
+
},
|
| 2098 |
+
"enc_blocks.5.conv1": {
|
| 2099 |
+
"mean": -0.44563060998916626,
|
| 2100 |
+
"std": 2.8831522464752197,
|
| 2101 |
+
"abs_max": 17.75,
|
| 2102 |
+
"frac_dead": 1.0172526344831567e-05,
|
| 2103 |
+
"frac_sat": 0.00896453857421875
|
| 2104 |
+
},
|
| 2105 |
+
"enc_blocks.5.t_proj": {
|
| 2106 |
+
"mean": -0.0013833555858582258,
|
| 2107 |
+
"std": 0.10456022620201111,
|
| 2108 |
+
"abs_max": 0.859375,
|
| 2109 |
+
"frac_dead": 5.425347262644209e-05,
|
| 2110 |
+
"frac_sat": 0.0
|
| 2111 |
+
},
|
| 2112 |
+
"enc_blocks.5.norm2": {
|
| 2113 |
+
"mean": -0.010284450836479664,
|
| 2114 |
+
"std": 0.9922280311584473,
|
| 2115 |
+
"abs_max": 5.930111885070801,
|
| 2116 |
+
"frac_dead": 0.0,
|
| 2117 |
+
"frac_sat": 0.0
|
| 2118 |
+
},
|
| 2119 |
+
"enc_blocks.5.conv2": {
|
| 2120 |
+
"mean": 0.49203333258628845,
|
| 2121 |
+
"std": 2.673565626144409,
|
| 2122 |
+
"abs_max": 17.375,
|
| 2123 |
+
"frac_dead": 5.086263172415784e-06,
|
| 2124 |
+
"frac_sat": 0.00267791748046875
|
| 2125 |
+
},
|
| 2126 |
+
"enc_attns.5.norm": {
|
| 2127 |
+
"mean": 0.0005484223365783691,
|
| 2128 |
+
"std": 0.9975232481956482,
|
| 2129 |
+
"abs_max": 8.068681716918945,
|
| 2130 |
+
"frac_dead": 8.477105097881577e-07,
|
| 2131 |
+
"frac_sat": 0.0
|
| 2132 |
+
},
|
| 2133 |
+
"enc_attns.5.qkv": {
|
| 2134 |
+
"mean": 0.0058145360089838505,
|
| 2135 |
+
"std": 0.7659077644348145,
|
| 2136 |
+
"abs_max": 6.75,
|
| 2137 |
+
"frac_dead": 6.668655987596139e-05,
|
| 2138 |
+
"frac_sat": 0.0
|
| 2139 |
+
},
|
| 2140 |
+
"enc_attns.5.proj": {
|
| 2141 |
+
"mean": -0.004640444181859493,
|
| 2142 |
+
"std": 0.3258810043334961,
|
| 2143 |
+
"abs_max": 4.125,
|
| 2144 |
+
"frac_dead": 0.00016276042151730508,
|
| 2145 |
+
"frac_sat": 0.0
|
| 2146 |
+
},
|
| 2147 |
+
"mid1.norm1": {
|
| 2148 |
+
"mean": -4.7561195970047265e-05,
|
| 2149 |
+
"std": 1.0003793239593506,
|
| 2150 |
+
"abs_max": 7.752286434173584,
|
| 2151 |
+
"frac_dead": 2.543131586207892e-06,
|
| 2152 |
+
"frac_sat": 0.0
|
| 2153 |
+
},
|
| 2154 |
+
"mid1.conv1": {
|
| 2155 |
+
"mean": -0.9362229108810425,
|
| 2156 |
+
"std": 1.927423119544983,
|
| 2157 |
+
"abs_max": 22.125,
|
| 2158 |
+
"frac_dead": 1.4411079064302612e-05,
|
| 2159 |
+
"frac_sat": 0.007679409347474575
|
| 2160 |
+
},
|
| 2161 |
+
"mid1.t_proj": {
|
| 2162 |
+
"mean": 0.002691776491701603,
|
| 2163 |
+
"std": 0.09969361871480942,
|
| 2164 |
+
"abs_max": 0.72265625,
|
| 2165 |
+
"frac_dead": 0.0,
|
| 2166 |
+
"frac_sat": 0.0
|
| 2167 |
+
},
|
| 2168 |
+
"mid1.norm2": {
|
| 2169 |
+
"mean": -0.012542976066470146,
|
| 2170 |
+
"std": 0.9945948123931885,
|
| 2171 |
+
"abs_max": 8.536553382873535,
|
| 2172 |
+
"frac_dead": 0.0,
|
| 2173 |
+
"frac_sat": 0.0
|
| 2174 |
+
},
|
| 2175 |
+
"mid1.conv2": {
|
| 2176 |
+
"mean": -0.007705233991146088,
|
| 2177 |
+
"std": 0.5925304293632507,
|
| 2178 |
+
"abs_max": 3.484375,
|
| 2179 |
+
"frac_dead": 3.221299994038418e-05,
|
| 2180 |
+
"frac_sat": 0.0
|
| 2181 |
+
},
|
| 2182 |
+
"mid_attn.norm": {
|
| 2183 |
+
"mean": 0.0004566659627016634,
|
| 2184 |
+
"std": 0.997485876083374,
|
| 2185 |
+
"abs_max": 7.793497085571289,
|
| 2186 |
+
"frac_dead": 8.477105097881577e-07,
|
| 2187 |
+
"frac_sat": 0.0
|
| 2188 |
+
},
|
| 2189 |
+
"mid_attn.qkv": {
|
| 2190 |
+
"mean": 0.010610872879624367,
|
| 2191 |
+
"std": 0.6843896508216858,
|
| 2192 |
+
"abs_max": 5.8125,
|
| 2193 |
+
"frac_dead": 6.527370715048164e-05,
|
| 2194 |
+
"frac_sat": 0.0
|
| 2195 |
+
},
|
| 2196 |
+
"mid_attn.proj": {
|
| 2197 |
+
"mean": -0.0158980805426836,
|
| 2198 |
+
"std": 0.2606736719608307,
|
| 2199 |
+
"abs_max": 1.4765625,
|
| 2200 |
+
"frac_dead": 0.00015004475426394492,
|
| 2201 |
+
"frac_sat": 0.0
|
| 2202 |
+
},
|
| 2203 |
+
"mid2.norm1": {
|
| 2204 |
+
"mean": 0.0004634691576939076,
|
| 2205 |
+
"std": 1.0005942583084106,
|
| 2206 |
+
"abs_max": 7.7080841064453125,
|
| 2207 |
+
"frac_dead": 0.0,
|
| 2208 |
+
"frac_sat": 0.0
|
| 2209 |
+
},
|
| 2210 |
+
"mid2.conv1": {
|
| 2211 |
+
"mean": -1.05628502368927,
|
| 2212 |
+
"std": 1.9699069261550903,
|
| 2213 |
+
"abs_max": 24.75,
|
| 2214 |
+
"frac_dead": 1.1867947250721045e-05,
|
| 2215 |
+
"frac_sat": 0.011021931655704975
|
| 2216 |
+
},
|
| 2217 |
+
"mid2.t_proj": {
|
| 2218 |
+
"mean": 0.0014389199204742908,
|
| 2219 |
+
"std": 0.10134941339492798,
|
| 2220 |
+
"abs_max": 0.80859375,
|
| 2221 |
+
"frac_dead": 0.0,
|
| 2222 |
+
"frac_sat": 0.0
|
| 2223 |
+
},
|
| 2224 |
+
"mid2.norm2": {
|
| 2225 |
+
"mean": -0.013441530987620354,
|
| 2226 |
+
"std": 0.99652498960495,
|
| 2227 |
+
"abs_max": 10.35103988647461,
|
| 2228 |
+
"frac_dead": 0.0,
|
| 2229 |
+
"frac_sat": 4.238552719471045e-06
|
| 2230 |
+
},
|
| 2231 |
+
"mid2.conv2": {
|
| 2232 |
+
"mean": 0.01876220293343067,
|
| 2233 |
+
"std": 0.8646733164787292,
|
| 2234 |
+
"abs_max": 4.8125,
|
| 2235 |
+
"frac_dead": 1.7801919966586865e-05,
|
| 2236 |
+
"frac_sat": 0.0
|
| 2237 |
+
},
|
| 2238 |
+
"dec_blocks.0.norm1": {
|
| 2239 |
+
"mean": 0.0012762596597895026,
|
| 2240 |
+
"std": 1.001218557357788,
|
| 2241 |
+
"abs_max": 8.005143165588379,
|
| 2242 |
+
"frac_dead": 4.238552548940788e-07,
|
| 2243 |
+
"frac_sat": 0.0
|
| 2244 |
+
},
|
| 2245 |
+
"dec_blocks.0.conv1": {
|
| 2246 |
+
"mean": -2.976426839828491,
|
| 2247 |
+
"std": 4.75129508972168,
|
| 2248 |
+
"abs_max": 59.75,
|
| 2249 |
+
"frac_dead": 4.238552719471045e-06,
|
| 2250 |
+
"frac_sat": 0.0577155239880085
|
| 2251 |
+
},
|
| 2252 |
+
"dec_blocks.0.t_proj": {
|
| 2253 |
+
"mean": -0.0010795136913657188,
|
| 2254 |
+
"std": 0.10378381609916687,
|
| 2255 |
+
"abs_max": 0.6640625,
|
| 2256 |
+
"frac_dead": 0.0,
|
| 2257 |
+
"frac_sat": 0.0
|
| 2258 |
+
},
|
| 2259 |
+
"dec_blocks.0.norm2": {
|
| 2260 |
+
"mean": -0.017285624518990517,
|
| 2261 |
+
"std": 0.9987004995346069,
|
| 2262 |
+
"abs_max": 10.219768524169922,
|
| 2263 |
+
"frac_dead": 8.477105097881577e-07,
|
| 2264 |
+
"frac_sat": 3.3908420391526306e-06
|
| 2265 |
+
},
|
| 2266 |
+
"dec_blocks.0.conv2": {
|
| 2267 |
+
"mean": 0.0011020468082278967,
|
| 2268 |
+
"std": 0.4619784951210022,
|
| 2269 |
+
"abs_max": 2.71875,
|
| 2270 |
+
"frac_dead": 3.645155084086582e-05,
|
| 2271 |
+
"frac_sat": 0.0
|
| 2272 |
+
},
|
| 2273 |
+
"dec_blocks.0.skip": {
|
| 2274 |
+
"mean": 0.05044516175985336,
|
| 2275 |
+
"std": 2.555394411087036,
|
| 2276 |
+
"abs_max": 28.125,
|
| 2277 |
+
"frac_dead": 1.9497341781971045e-05,
|
| 2278 |
+
"frac_sat": 0.00231170654296875
|
| 2279 |
+
},
|
| 2280 |
+
"dec_attns.0.norm": {
|
| 2281 |
+
"mean": 0.00031653710175305605,
|
| 2282 |
+
"std": 0.9951615929603577,
|
| 2283 |
+
"abs_max": 7.695018768310547,
|
| 2284 |
+
"frac_dead": 1.6954210195763153e-06,
|
| 2285 |
+
"frac_sat": 0.0
|
| 2286 |
+
},
|
| 2287 |
+
"dec_attns.0.qkv": {
|
| 2288 |
+
"mean": 0.005484588909894228,
|
| 2289 |
+
"std": 0.659173846244812,
|
| 2290 |
+
"abs_max": 6.28125,
|
| 2291 |
+
"frac_dead": 8.024992712307721e-05,
|
| 2292 |
+
"frac_sat": 0.0
|
| 2293 |
+
},
|
| 2294 |
+
"dec_attns.0.proj": {
|
| 2295 |
+
"mean": 0.009530746378004551,
|
| 2296 |
+
"std": 0.541577935218811,
|
| 2297 |
+
"abs_max": 4.25,
|
| 2298 |
+
"frac_dead": 8.561876165913418e-05,
|
| 2299 |
+
"frac_sat": 0.0
|
| 2300 |
+
},
|
| 2301 |
+
"dec_ups.0.op": {
|
| 2302 |
+
"mean": 0.13259148597717285,
|
| 2303 |
+
"std": 5.4154438972473145,
|
| 2304 |
+
"abs_max": 39.5,
|
| 2305 |
+
"frac_dead": 4.026624992548022e-06,
|
| 2306 |
+
"frac_sat": 0.07323244214057922
|
| 2307 |
+
},
|
| 2308 |
+
"dec_blocks.1.norm1": {
|
| 2309 |
+
"mean": 0.0012307068100199103,
|
| 2310 |
+
"std": 1.0010651350021362,
|
| 2311 |
+
"abs_max": 9.00831127166748,
|
| 2312 |
+
"frac_dead": 1.0596381798677612e-06,
|
| 2313 |
+
"frac_sat": 0.0
|
| 2314 |
+
},
|
| 2315 |
+
"dec_blocks.1.conv1": {
|
| 2316 |
+
"mean": -2.3529160022735596,
|
| 2317 |
+
"std": 4.61397647857666,
|
| 2318 |
+
"abs_max": 79.0,
|
| 2319 |
+
"frac_dead": 8.053249985096045e-06,
|
| 2320 |
+
"frac_sat": 0.0458882637321949
|
| 2321 |
+
},
|
| 2322 |
+
"dec_blocks.1.t_proj": {
|
| 2323 |
+
"mean": -0.0017044859705492854,
|
| 2324 |
+
"std": 0.10656358301639557,
|
| 2325 |
+
"abs_max": 0.8046875,
|
| 2326 |
+
"frac_dead": 0.0,
|
| 2327 |
+
"frac_sat": 0.0
|
| 2328 |
+
},
|
| 2329 |
+
"dec_blocks.1.norm2": {
|
| 2330 |
+
"mean": -0.018469274044036865,
|
| 2331 |
+
"std": 1.0006890296936035,
|
| 2332 |
+
"abs_max": 15.932745933532715,
|
| 2333 |
+
"frac_dead": 1.9073486328125e-06,
|
| 2334 |
+
"frac_sat": 0.00025876364088617265
|
| 2335 |
+
},
|
| 2336 |
+
"dec_blocks.1.conv2": {
|
| 2337 |
+
"mean": -0.001623636344447732,
|
| 2338 |
+
"std": 0.4008197486400604,
|
| 2339 |
+
"abs_max": 2.21875,
|
| 2340 |
+
"frac_dead": 3.645155084086582e-05,
|
| 2341 |
+
"frac_sat": 0.0
|
| 2342 |
+
},
|
| 2343 |
+
"dec_blocks.1.skip": {
|
| 2344 |
+
"mean": 0.03524673357605934,
|
| 2345 |
+
"std": 2.9326562881469727,
|
| 2346 |
+
"abs_max": 18.125,
|
| 2347 |
+
"frac_dead": 1.52587890625e-05,
|
| 2348 |
+
"frac_sat": 0.0042321947403252125
|
| 2349 |
+
},
|
| 2350 |
+
"dec_attns.1.norm": {
|
| 2351 |
+
"mean": 6.92961984896101e-05,
|
| 2352 |
+
"std": 0.9948720932006836,
|
| 2353 |
+
"abs_max": 7.577223300933838,
|
| 2354 |
+
"frac_dead": 1.4834934063401306e-06,
|
| 2355 |
+
"frac_sat": 0.0
|
| 2356 |
+
},
|
| 2357 |
+
"dec_attns.1.qkv": {
|
| 2358 |
+
"mean": 0.01385430246591568,
|
| 2359 |
+
"std": 0.7135639786720276,
|
| 2360 |
+
"abs_max": 5.875,
|
| 2361 |
+
"frac_dead": 8.067378075793386e-05,
|
| 2362 |
+
"frac_sat": 0.0
|
| 2363 |
+
},
|
| 2364 |
+
"dec_attns.1.proj": {
|
| 2365 |
+
"mean": 0.0175122432410717,
|
| 2366 |
+
"std": 0.7606446146965027,
|
| 2367 |
+
"abs_max": 4.0625,
|
| 2368 |
+
"frac_dead": 5.955166489002295e-05,
|
| 2369 |
+
"frac_sat": 0.0
|
| 2370 |
+
},
|
| 2371 |
+
"dec_ups.1.op": {
|
| 2372 |
+
"mean": 0.2810516059398651,
|
| 2373 |
+
"std": 5.511878967285156,
|
| 2374 |
+
"abs_max": 31.25,
|
| 2375 |
+
"frac_dead": 4.132588856009534e-06,
|
| 2376 |
+
"frac_sat": 0.08329842239618301
|
| 2377 |
+
},
|
| 2378 |
+
"dec_blocks.2.norm1": {
|
| 2379 |
+
"mean": 0.0003542789781931788,
|
| 2380 |
+
"std": 1.001099705696106,
|
| 2381 |
+
"abs_max": 10.847487449645996,
|
| 2382 |
+
"frac_dead": 7.770679530949565e-07,
|
| 2383 |
+
"frac_sat": 1.059638137235197e-07
|
| 2384 |
+
},
|
| 2385 |
+
"dec_blocks.2.conv1": {
|
| 2386 |
+
"mean": -1.4827497005462646,
|
| 2387 |
+
"std": 4.093748092651367,
|
| 2388 |
+
"abs_max": 52.75,
|
| 2389 |
+
"frac_dead": 1.3033549294050317e-05,
|
| 2390 |
+
"frac_sat": 0.05096329748630524
|
| 2391 |
+
},
|
| 2392 |
+
"dec_blocks.2.t_proj": {
|
| 2393 |
+
"mean": -0.004567438270896673,
|
| 2394 |
+
"std": 0.12303908169269562,
|
| 2395 |
+
"abs_max": 0.79296875,
|
| 2396 |
+
"frac_dead": 0.0,
|
| 2397 |
+
"frac_sat": 0.0
|
| 2398 |
+
},
|
| 2399 |
+
"dec_blocks.2.norm2": {
|
| 2400 |
+
"mean": -0.017762627452611923,
|
| 2401 |
+
"std": 1.0039026737213135,
|
| 2402 |
+
"abs_max": 12.032835960388184,
|
| 2403 |
+
"frac_dead": 0.0,
|
| 2404 |
+
"frac_sat": 1.3775295883533545e-05
|
| 2405 |
+
},
|
| 2406 |
+
"dec_blocks.2.conv2": {
|
| 2407 |
+
"mean": 0.01796824112534523,
|
| 2408 |
+
"std": 0.32834163308143616,
|
| 2409 |
+
"abs_max": 2.265625,
|
| 2410 |
+
"frac_dead": 5.955166489002295e-05,
|
| 2411 |
+
"frac_sat": 0.0
|
| 2412 |
+
},
|
| 2413 |
+
"dec_blocks.2.skip": {
|
| 2414 |
+
"mean": -0.02365983836352825,
|
| 2415 |
+
"std": 3.8385989665985107,
|
| 2416 |
+
"abs_max": 24.625,
|
| 2417 |
+
"frac_dead": 1.229180270456709e-05,
|
| 2418 |
+
"frac_sat": 0.020438512787222862
|
| 2419 |
+
},
|
| 2420 |
+
"dec_attns.2.norm": {
|
| 2421 |
+
"mean": -0.00018475121760275215,
|
| 2422 |
+
"std": 0.9939836263656616,
|
| 2423 |
+
"abs_max": 6.334526538848877,
|
| 2424 |
+
"frac_dead": 1.271565793103946e-06,
|
| 2425 |
+
"frac_sat": 0.0
|
| 2426 |
+
},
|
| 2427 |
+
"dec_attns.2.qkv": {
|
| 2428 |
+
"mean": -0.009542924351990223,
|
| 2429 |
+
"std": 0.6096933484077454,
|
| 2430 |
+
"abs_max": 4.90625,
|
| 2431 |
+
"frac_dead": 0.00012051617522956803,
|
| 2432 |
+
"frac_sat": 0.0
|
| 2433 |
+
},
|
| 2434 |
+
"dec_attns.2.proj": {
|
| 2435 |
+
"mean": -0.033470239490270615,
|
| 2436 |
+
"std": 0.44429004192352295,
|
| 2437 |
+
"abs_max": 2.03125,
|
| 2438 |
+
"frac_dead": 0.00014177958655636758,
|
| 2439 |
+
"frac_sat": 0.0
|
| 2440 |
+
},
|
| 2441 |
+
"dec_ups.2.op": {
|
| 2442 |
+
"mean": -0.3541581928730011,
|
| 2443 |
+
"std": 4.1034979820251465,
|
| 2444 |
+
"abs_max": 27.125,
|
| 2445 |
+
"frac_dead": 7.920794814708643e-06,
|
| 2446 |
+
"frac_sat": 0.03058568574488163
|
| 2447 |
+
},
|
| 2448 |
+
"dec_blocks.3.norm1": {
|
| 2449 |
+
"mean": -0.0014417987549677491,
|
| 2450 |
+
"std": 1.0000892877578735,
|
| 2451 |
+
"abs_max": 10.300485610961914,
|
| 2452 |
+
"frac_dead": 1.4967388324294006e-06,
|
| 2453 |
+
"frac_sat": 3.973643103449831e-08
|
| 2454 |
+
},
|
| 2455 |
+
"dec_blocks.3.conv1": {
|
| 2456 |
+
"mean": -0.9749405980110168,
|
| 2457 |
+
"std": 2.3514747619628906,
|
| 2458 |
+
"abs_max": 41.5,
|
| 2459 |
+
"frac_dead": 1.3457404747896362e-05,
|
| 2460 |
+
"frac_sat": 0.012159109115600586
|
| 2461 |
+
},
|
| 2462 |
+
"dec_blocks.3.t_proj": {
|
| 2463 |
+
"mean": -0.0004592237819451839,
|
| 2464 |
+
"std": 0.1721496880054474,
|
| 2465 |
+
"abs_max": 1.2265625,
|
| 2466 |
+
"frac_dead": 0.0,
|
| 2467 |
+
"frac_sat": 0.0
|
| 2468 |
+
},
|
| 2469 |
+
"dec_blocks.3.norm2": {
|
| 2470 |
+
"mean": -0.01709667034447193,
|
| 2471 |
+
"std": 0.9970768094062805,
|
| 2472 |
+
"abs_max": 14.025443077087402,
|
| 2473 |
+
"frac_dead": 7.94728578057402e-07,
|
| 2474 |
+
"frac_sat": 2.9987759262439795e-05
|
| 2475 |
+
},
|
| 2476 |
+
"dec_blocks.3.conv2": {
|
| 2477 |
+
"mean": 0.015155193395912647,
|
| 2478 |
+
"std": 0.4408976137638092,
|
| 2479 |
+
"abs_max": 4.8125,
|
| 2480 |
+
"frac_dead": 4.654460644815117e-05,
|
| 2481 |
+
"frac_sat": 0.0
|
| 2482 |
+
},
|
| 2483 |
+
"dec_blocks.3.skip": {
|
| 2484 |
+
"mean": 0.025708815082907677,
|
| 2485 |
+
"std": 2.072154998779297,
|
| 2486 |
+
"abs_max": 14.8125,
|
| 2487 |
+
"frac_dead": 2.8742684662574902e-05,
|
| 2488 |
+
"frac_sat": 0.0005844964180141687
|
| 2489 |
+
},
|
| 2490 |
+
"dec_ups.3.op": {
|
| 2491 |
+
"mean": -0.49303382635116577,
|
| 2492 |
+
"std": 2.7450525760650635,
|
| 2493 |
+
"abs_max": 27.5,
|
| 2494 |
+
"frac_dead": 1.2331538528087549e-05,
|
| 2495 |
+
"frac_sat": 0.004969603382050991
|
| 2496 |
+
},
|
| 2497 |
+
"dec_blocks.4.norm1": {
|
| 2498 |
+
"mean": -0.003529828740283847,
|
| 2499 |
+
"std": 0.9992601871490479,
|
| 2500 |
+
"abs_max": 15.095056533813477,
|
| 2501 |
+
"frac_dead": 1.3598689747595927e-06,
|
| 2502 |
+
"frac_sat": 2.0883701381535502e-06
|
| 2503 |
+
},
|
| 2504 |
+
"dec_blocks.4.conv1": {
|
| 2505 |
+
"mean": -0.6139392256736755,
|
| 2506 |
+
"std": 2.1271767616271973,
|
| 2507 |
+
"abs_max": 26.375,
|
| 2508 |
+
"frac_dead": 1.2238820090715308e-05,
|
| 2509 |
+
"frac_sat": 0.0013043085345998406
|
| 2510 |
+
},
|
| 2511 |
+
"dec_blocks.4.t_proj": {
|
| 2512 |
+
"mean": -0.005459681153297424,
|
| 2513 |
+
"std": 0.1980716437101364,
|
| 2514 |
+
"abs_max": 1.2578125,
|
| 2515 |
+
"frac_dead": 0.0,
|
| 2516 |
+
"frac_sat": 0.0
|
| 2517 |
+
},
|
| 2518 |
+
"dec_blocks.4.norm2": {
|
| 2519 |
+
"mean": -0.008547251112759113,
|
| 2520 |
+
"std": 0.9963138103485107,
|
| 2521 |
+
"abs_max": 11.549351692199707,
|
| 2522 |
+
"frac_dead": 2.2517311037972831e-07,
|
| 2523 |
+
"frac_sat": 2.2517311037972831e-07
|
| 2524 |
+
},
|
| 2525 |
+
"dec_blocks.4.conv2": {
|
| 2526 |
+
"mean": 0.03798123076558113,
|
| 2527 |
+
"std": 0.6906255483627319,
|
| 2528 |
+
"abs_max": 4.375,
|
| 2529 |
+
"frac_dead": 4.7193632781272754e-05,
|
| 2530 |
+
"frac_sat": 0.0
|
| 2531 |
+
},
|
| 2532 |
+
"dec_blocks.4.skip": {
|
| 2533 |
+
"mean": -0.16517576575279236,
|
| 2534 |
+
"std": 1.960332989692688,
|
| 2535 |
+
"abs_max": 19.75,
|
| 2536 |
+
"frac_dead": 3.20937906508334e-05,
|
| 2537 |
+
"frac_sat": 0.00011465284478617832
|
| 2538 |
+
},
|
| 2539 |
+
"dec_ups.4.op": {
|
| 2540 |
+
"mean": -0.42157071828842163,
|
| 2541 |
+
"std": 2.7456705570220947,
|
| 2542 |
+
"abs_max": 41.5,
|
| 2543 |
+
"frac_dead": 1.6805199265945703e-05,
|
| 2544 |
+
"frac_sat": 0.013286677189171314
|
| 2545 |
+
},
|
| 2546 |
+
"dec_blocks.5.norm1": {
|
| 2547 |
+
"mean": -0.00543779693543911,
|
| 2548 |
+
"std": 0.997042715549469,
|
| 2549 |
+
"abs_max": 11.520453453063965,
|
| 2550 |
+
"frac_dead": 1.2483861837608856e-06,
|
| 2551 |
+
"frac_sat": 6.291601550856285e-08
|
| 2552 |
+
},
|
| 2553 |
+
"dec_blocks.5.conv1": {
|
| 2554 |
+
"mean": -0.04659652337431908,
|
| 2555 |
+
"std": 1.7403225898742676,
|
| 2556 |
+
"abs_max": 35.5,
|
| 2557 |
+
"frac_dead": 1.5106465980352368e-05,
|
| 2558 |
+
"frac_sat": 0.000550044933333993
|
| 2559 |
+
},
|
| 2560 |
+
"dec_blocks.5.t_proj": {
|
| 2561 |
+
"mean": 0.015368163585662842,
|
| 2562 |
+
"std": 0.27995017170906067,
|
| 2563 |
+
"abs_max": 2.234375,
|
| 2564 |
+
"frac_dead": 0.0,
|
| 2565 |
+
"frac_sat": 0.0
|
| 2566 |
+
},
|
| 2567 |
+
"dec_blocks.5.norm2": {
|
| 2568 |
+
"mean": 0.000441109761595726,
|
| 2569 |
+
"std": 1.006426215171814,
|
| 2570 |
+
"abs_max": 16.939395904541016,
|
| 2571 |
+
"frac_dead": 4.635916752704361e-07,
|
| 2572 |
+
"frac_sat": 5.116065494803479e-06
|
| 2573 |
+
},
|
| 2574 |
+
"dec_blocks.5.conv2": {
|
| 2575 |
+
"mean": -1.157901406288147,
|
| 2576 |
+
"std": 2.146827220916748,
|
| 2577 |
+
"abs_max": 28.875,
|
| 2578 |
+
"frac_dead": 2.6341942430008203e-05,
|
| 2579 |
+
"frac_sat": 0.004018975887447596
|
| 2580 |
+
},
|
| 2581 |
+
"dec_blocks.5.skip": {
|
| 2582 |
+
"mean": -0.401109904050827,
|
| 2583 |
+
"std": 1.9778581857681274,
|
| 2584 |
+
"abs_max": 26.875,
|
| 2585 |
+
"frac_dead": 4.436572271515615e-05,
|
| 2586 |
+
"frac_sat": 0.0012669430579990149
|
| 2587 |
+
},
|
| 2588 |
+
"out_norm": {
|
| 2589 |
+
"mean": -0.006768459919840097,
|
| 2590 |
+
"std": 0.9930412769317627,
|
| 2591 |
+
"abs_max": 11.80453872680664,
|
| 2592 |
+
"frac_dead": 1.4570024120530434e-07,
|
| 2593 |
+
"frac_sat": 3.344482877309929e-07
|
| 2594 |
+
},
|
| 2595 |
+
"out_conv": {
|
| 2596 |
+
"mean": 0.5060885548591614,
|
| 2597 |
+
"std": 0.23508810997009277,
|
| 2598 |
+
"abs_max": 1.21875,
|
| 2599 |
+
"frac_dead": 2.966986812680261e-06,
|
| 2600 |
+
"frac_sat": 0.0
|
| 2601 |
+
}
|
| 2602 |
+
}
|
| 2603 |
+
}
|
diagnostics/per_layer_iter500.json
ADDED
|
@@ -0,0 +1,2603 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"iter": 500,
|
| 3 |
+
"per_layer": {
|
| 4 |
+
"time_mlp.1.weight": {
|
| 5 |
+
"w_norm": 18.595300674438477,
|
| 6 |
+
"g_norm": 0.0813865214586258,
|
| 7 |
+
"shape": [
|
| 8 |
+
1024,
|
| 9 |
+
256
|
| 10 |
+
]
|
| 11 |
+
},
|
| 12 |
+
"time_mlp.1.bias": {
|
| 13 |
+
"w_norm": 1.1589714288711548,
|
| 14 |
+
"g_norm": 0.007196313701570034,
|
| 15 |
+
"shape": [
|
| 16 |
+
1024
|
| 17 |
+
]
|
| 18 |
+
},
|
| 19 |
+
"time_mlp.3.weight": {
|
| 20 |
+
"w_norm": 9.38232135772705,
|
| 21 |
+
"g_norm": 0.038036804646253586,
|
| 22 |
+
"shape": [
|
| 23 |
+
256,
|
| 24 |
+
1024
|
| 25 |
+
]
|
| 26 |
+
},
|
| 27 |
+
"time_mlp.3.bias": {
|
| 28 |
+
"w_norm": 0.29130446910858154,
|
| 29 |
+
"g_norm": 0.0083179771900177,
|
| 30 |
+
"shape": [
|
| 31 |
+
256
|
| 32 |
+
]
|
| 33 |
+
},
|
| 34 |
+
"input_conv.weight": {
|
| 35 |
+
"w_norm": 7.941638469696045,
|
| 36 |
+
"g_norm": 0.07811103761196136,
|
| 37 |
+
"shape": [
|
| 38 |
+
192,
|
| 39 |
+
6,
|
| 40 |
+
3,
|
| 41 |
+
3
|
| 42 |
+
]
|
| 43 |
+
},
|
| 44 |
+
"input_conv.bias": {
|
| 45 |
+
"w_norm": 1.1801764965057373,
|
| 46 |
+
"g_norm": 0.05344012752175331,
|
| 47 |
+
"shape": [
|
| 48 |
+
192
|
| 49 |
+
]
|
| 50 |
+
},
|
| 51 |
+
"enc_blocks.0.norm1.weight": {
|
| 52 |
+
"w_norm": 13.835872650146484,
|
| 53 |
+
"g_norm": 0.015104694291949272,
|
| 54 |
+
"shape": [
|
| 55 |
+
192
|
| 56 |
+
]
|
| 57 |
+
},
|
| 58 |
+
"enc_blocks.0.norm1.bias": {
|
| 59 |
+
"w_norm": 0.03625425696372986,
|
| 60 |
+
"g_norm": 0.022762326523661613,
|
| 61 |
+
"shape": [
|
| 62 |
+
192
|
| 63 |
+
]
|
| 64 |
+
},
|
| 65 |
+
"enc_blocks.0.conv1.weight": {
|
| 66 |
+
"w_norm": 8.11551570892334,
|
| 67 |
+
"g_norm": 0.1937582790851593,
|
| 68 |
+
"shape": [
|
| 69 |
+
192,
|
| 70 |
+
192,
|
| 71 |
+
3,
|
| 72 |
+
3
|
| 73 |
+
]
|
| 74 |
+
},
|
| 75 |
+
"enc_blocks.0.conv1.bias": {
|
| 76 |
+
"w_norm": 0.2021755576133728,
|
| 77 |
+
"g_norm": 0.019418831914663315,
|
| 78 |
+
"shape": [
|
| 79 |
+
192
|
| 80 |
+
]
|
| 81 |
+
},
|
| 82 |
+
"enc_blocks.0.t_proj.weight": {
|
| 83 |
+
"w_norm": 8.000494956970215,
|
| 84 |
+
"g_norm": 0.014280546456575394,
|
| 85 |
+
"shape": [
|
| 86 |
+
192,
|
| 87 |
+
256
|
| 88 |
+
]
|
| 89 |
+
},
|
| 90 |
+
"enc_blocks.0.t_proj.bias": {
|
| 91 |
+
"w_norm": 0.5028907656669617,
|
| 92 |
+
"g_norm": 0.01941794902086258,
|
| 93 |
+
"shape": [
|
| 94 |
+
192
|
| 95 |
+
]
|
| 96 |
+
},
|
| 97 |
+
"enc_blocks.0.norm2.weight": {
|
| 98 |
+
"w_norm": 13.850933074951172,
|
| 99 |
+
"g_norm": 0.00578329199925065,
|
| 100 |
+
"shape": [
|
| 101 |
+
192
|
| 102 |
+
]
|
| 103 |
+
},
|
| 104 |
+
"enc_blocks.0.norm2.bias": {
|
| 105 |
+
"w_norm": 0.019755441695451736,
|
| 106 |
+
"g_norm": 0.007538977544754744,
|
| 107 |
+
"shape": [
|
| 108 |
+
192
|
| 109 |
+
]
|
| 110 |
+
},
|
| 111 |
+
"enc_blocks.0.conv2.weight": {
|
| 112 |
+
"w_norm": 8.096028327941895,
|
| 113 |
+
"g_norm": 0.06561890244483948,
|
| 114 |
+
"shape": [
|
| 115 |
+
192,
|
| 116 |
+
192,
|
| 117 |
+
3,
|
| 118 |
+
3
|
| 119 |
+
]
|
| 120 |
+
},
|
| 121 |
+
"enc_blocks.0.conv2.bias": {
|
| 122 |
+
"w_norm": 0.187120258808136,
|
| 123 |
+
"g_norm": 0.0059786029160022736,
|
| 124 |
+
"shape": [
|
| 125 |
+
192
|
| 126 |
+
]
|
| 127 |
+
},
|
| 128 |
+
"enc_blocks.1.norm1.weight": {
|
| 129 |
+
"w_norm": 13.851934432983398,
|
| 130 |
+
"g_norm": 9.219303319696337e-05,
|
| 131 |
+
"shape": [
|
| 132 |
+
192
|
| 133 |
+
]
|
| 134 |
+
},
|
| 135 |
+
"enc_blocks.1.norm1.bias": {
|
| 136 |
+
"w_norm": 0.015200300142168999,
|
| 137 |
+
"g_norm": 0.00011359907512087375,
|
| 138 |
+
"shape": [
|
| 139 |
+
192
|
| 140 |
+
]
|
| 141 |
+
},
|
| 142 |
+
"enc_blocks.1.conv1.weight": {
|
| 143 |
+
"w_norm": 8.089102745056152,
|
| 144 |
+
"g_norm": 0.0032792543061077595,
|
| 145 |
+
"shape": [
|
| 146 |
+
192,
|
| 147 |
+
192,
|
| 148 |
+
3,
|
| 149 |
+
3
|
| 150 |
+
]
|
| 151 |
+
},
|
| 152 |
+
"enc_blocks.1.conv1.bias": {
|
| 153 |
+
"w_norm": 0.19240157306194305,
|
| 154 |
+
"g_norm": 0.000176840418134816,
|
| 155 |
+
"shape": [
|
| 156 |
+
192
|
| 157 |
+
]
|
| 158 |
+
},
|
| 159 |
+
"enc_blocks.1.t_proj.weight": {
|
| 160 |
+
"w_norm": 7.998848915100098,
|
| 161 |
+
"g_norm": 0.0001282700541196391,
|
| 162 |
+
"shape": [
|
| 163 |
+
192,
|
| 164 |
+
256
|
| 165 |
+
]
|
| 166 |
+
},
|
| 167 |
+
"enc_blocks.1.t_proj.bias": {
|
| 168 |
+
"w_norm": 0.5218672752380371,
|
| 169 |
+
"g_norm": 0.00017683516489341855,
|
| 170 |
+
"shape": [
|
| 171 |
+
192
|
| 172 |
+
]
|
| 173 |
+
},
|
| 174 |
+
"enc_blocks.1.norm2.weight": {
|
| 175 |
+
"w_norm": 13.833203315734863,
|
| 176 |
+
"g_norm": 0.00020846868574153632,
|
| 177 |
+
"shape": [
|
| 178 |
+
192
|
| 179 |
+
]
|
| 180 |
+
},
|
| 181 |
+
"enc_blocks.1.norm2.bias": {
|
| 182 |
+
"w_norm": 0.02090640738606453,
|
| 183 |
+
"g_norm": 0.00021477529662661254,
|
| 184 |
+
"shape": [
|
| 185 |
+
192
|
| 186 |
+
]
|
| 187 |
+
},
|
| 188 |
+
"enc_blocks.1.conv2.weight": {
|
| 189 |
+
"w_norm": 8.047603607177734,
|
| 190 |
+
"g_norm": 0.006150349508970976,
|
| 191 |
+
"shape": [
|
| 192 |
+
192,
|
| 193 |
+
192,
|
| 194 |
+
3,
|
| 195 |
+
3
|
| 196 |
+
]
|
| 197 |
+
},
|
| 198 |
+
"enc_blocks.1.conv2.bias": {
|
| 199 |
+
"w_norm": 0.2068602740764618,
|
| 200 |
+
"g_norm": 0.0003937286091968417,
|
| 201 |
+
"shape": [
|
| 202 |
+
192
|
| 203 |
+
]
|
| 204 |
+
},
|
| 205 |
+
"enc_blocks.2.norm1.weight": {
|
| 206 |
+
"w_norm": 13.853577613830566,
|
| 207 |
+
"g_norm": 2.653978117450606e-05,
|
| 208 |
+
"shape": [
|
| 209 |
+
192
|
| 210 |
+
]
|
| 211 |
+
},
|
| 212 |
+
"enc_blocks.2.norm1.bias": {
|
| 213 |
+
"w_norm": 0.020222244784235954,
|
| 214 |
+
"g_norm": 3.004993050126359e-05,
|
| 215 |
+
"shape": [
|
| 216 |
+
192
|
| 217 |
+
]
|
| 218 |
+
},
|
| 219 |
+
"enc_blocks.2.conv1.weight": {
|
| 220 |
+
"w_norm": 11.423590660095215,
|
| 221 |
+
"g_norm": 0.0007072354201227427,
|
| 222 |
+
"shape": [
|
| 223 |
+
384,
|
| 224 |
+
192,
|
| 225 |
+
3,
|
| 226 |
+
3
|
| 227 |
+
]
|
| 228 |
+
},
|
| 229 |
+
"enc_blocks.2.conv1.bias": {
|
| 230 |
+
"w_norm": 0.2670600712299347,
|
| 231 |
+
"g_norm": 3.2856267353054136e-05,
|
| 232 |
+
"shape": [
|
| 233 |
+
384
|
| 234 |
+
]
|
| 235 |
+
},
|
| 236 |
+
"enc_blocks.2.t_proj.weight": {
|
| 237 |
+
"w_norm": 11.348468780517578,
|
| 238 |
+
"g_norm": 2.6057350623887032e-05,
|
| 239 |
+
"shape": [
|
| 240 |
+
384,
|
| 241 |
+
256
|
| 242 |
+
]
|
| 243 |
+
},
|
| 244 |
+
"enc_blocks.2.t_proj.bias": {
|
| 245 |
+
"w_norm": 0.7200393080711365,
|
| 246 |
+
"g_norm": 3.286291394033469e-05,
|
| 247 |
+
"shape": [
|
| 248 |
+
384
|
| 249 |
+
]
|
| 250 |
+
},
|
| 251 |
+
"enc_blocks.2.norm2.weight": {
|
| 252 |
+
"w_norm": 19.526947021484375,
|
| 253 |
+
"g_norm": 2.8658503651968203e-05,
|
| 254 |
+
"shape": [
|
| 255 |
+
384
|
| 256 |
+
]
|
| 257 |
+
},
|
| 258 |
+
"enc_blocks.2.norm2.bias": {
|
| 259 |
+
"w_norm": 0.052438266575336456,
|
| 260 |
+
"g_norm": 3.415972969378345e-05,
|
| 261 |
+
"shape": [
|
| 262 |
+
384
|
| 263 |
+
]
|
| 264 |
+
},
|
| 265 |
+
"enc_blocks.2.conv2.weight": {
|
| 266 |
+
"w_norm": 11.376154899597168,
|
| 267 |
+
"g_norm": 0.0018718303181231022,
|
| 268 |
+
"shape": [
|
| 269 |
+
384,
|
| 270 |
+
384,
|
| 271 |
+
3,
|
| 272 |
+
3
|
| 273 |
+
]
|
| 274 |
+
},
|
| 275 |
+
"enc_blocks.2.conv2.bias": {
|
| 276 |
+
"w_norm": 0.20025207102298737,
|
| 277 |
+
"g_norm": 7.625852595083416e-05,
|
| 278 |
+
"shape": [
|
| 279 |
+
384
|
| 280 |
+
]
|
| 281 |
+
},
|
| 282 |
+
"enc_blocks.2.skip.weight": {
|
| 283 |
+
"w_norm": 11.280326843261719,
|
| 284 |
+
"g_norm": 0.0024632003623992205,
|
| 285 |
+
"shape": [
|
| 286 |
+
384,
|
| 287 |
+
192,
|
| 288 |
+
1,
|
| 289 |
+
1
|
| 290 |
+
]
|
| 291 |
+
},
|
| 292 |
+
"enc_blocks.2.skip.bias": {
|
| 293 |
+
"w_norm": 0.8246392607688904,
|
| 294 |
+
"g_norm": 7.625852595083416e-05,
|
| 295 |
+
"shape": [
|
| 296 |
+
384
|
| 297 |
+
]
|
| 298 |
+
},
|
| 299 |
+
"enc_blocks.3.norm1.weight": {
|
| 300 |
+
"w_norm": 19.596302032470703,
|
| 301 |
+
"g_norm": 6.7835412664862815e-06,
|
| 302 |
+
"shape": [
|
| 303 |
+
384
|
| 304 |
+
]
|
| 305 |
+
},
|
| 306 |
+
"enc_blocks.3.norm1.bias": {
|
| 307 |
+
"w_norm": 0.026706550270318985,
|
| 308 |
+
"g_norm": 6.360353381751338e-06,
|
| 309 |
+
"shape": [
|
| 310 |
+
384
|
| 311 |
+
]
|
| 312 |
+
},
|
| 313 |
+
"enc_blocks.3.conv1.weight": {
|
| 314 |
+
"w_norm": 11.475268363952637,
|
| 315 |
+
"g_norm": 0.00031782532460056245,
|
| 316 |
+
"shape": [
|
| 317 |
+
384,
|
| 318 |
+
384,
|
| 319 |
+
3,
|
| 320 |
+
3
|
| 321 |
+
]
|
| 322 |
+
},
|
| 323 |
+
"enc_blocks.3.conv1.bias": {
|
| 324 |
+
"w_norm": 0.1944141387939453,
|
| 325 |
+
"g_norm": 9.2924574346398e-06,
|
| 326 |
+
"shape": [
|
| 327 |
+
384
|
| 328 |
+
]
|
| 329 |
+
},
|
| 330 |
+
"enc_blocks.3.t_proj.weight": {
|
| 331 |
+
"w_norm": 11.324196815490723,
|
| 332 |
+
"g_norm": 7.110313617886277e-06,
|
| 333 |
+
"shape": [
|
| 334 |
+
384,
|
| 335 |
+
256
|
| 336 |
+
]
|
| 337 |
+
},
|
| 338 |
+
"enc_blocks.3.t_proj.bias": {
|
| 339 |
+
"w_norm": 0.691976010799408,
|
| 340 |
+
"g_norm": 9.291392416344024e-06,
|
| 341 |
+
"shape": [
|
| 342 |
+
384
|
| 343 |
+
]
|
| 344 |
+
},
|
| 345 |
+
"enc_blocks.3.norm2.weight": {
|
| 346 |
+
"w_norm": 19.542306900024414,
|
| 347 |
+
"g_norm": 2.2274725779425353e-05,
|
| 348 |
+
"shape": [
|
| 349 |
+
384
|
| 350 |
+
]
|
| 351 |
+
},
|
| 352 |
+
"enc_blocks.3.norm2.bias": {
|
| 353 |
+
"w_norm": 0.04708161577582359,
|
| 354 |
+
"g_norm": 1.5845758753130212e-05,
|
| 355 |
+
"shape": [
|
| 356 |
+
384
|
| 357 |
+
]
|
| 358 |
+
},
|
| 359 |
+
"enc_blocks.3.conv2.weight": {
|
| 360 |
+
"w_norm": 11.365864753723145,
|
| 361 |
+
"g_norm": 0.0007113237516023219,
|
| 362 |
+
"shape": [
|
| 363 |
+
384,
|
| 364 |
+
384,
|
| 365 |
+
3,
|
| 366 |
+
3
|
| 367 |
+
]
|
| 368 |
+
},
|
| 369 |
+
"enc_blocks.3.conv2.bias": {
|
| 370 |
+
"w_norm": 0.1897016316652298,
|
| 371 |
+
"g_norm": 2.332814437977504e-05,
|
| 372 |
+
"shape": [
|
| 373 |
+
384
|
| 374 |
+
]
|
| 375 |
+
},
|
| 376 |
+
"enc_blocks.4.norm1.weight": {
|
| 377 |
+
"w_norm": 19.591983795166016,
|
| 378 |
+
"g_norm": 1.2785332728526555e-05,
|
| 379 |
+
"shape": [
|
| 380 |
+
384
|
| 381 |
+
]
|
| 382 |
+
},
|
| 383 |
+
"enc_blocks.4.norm1.bias": {
|
| 384 |
+
"w_norm": 0.032775815576314926,
|
| 385 |
+
"g_norm": 1.206148044730071e-05,
|
| 386 |
+
"shape": [
|
| 387 |
+
384
|
| 388 |
+
]
|
| 389 |
+
},
|
| 390 |
+
"enc_blocks.4.conv1.weight": {
|
| 391 |
+
"w_norm": 16.21708869934082,
|
| 392 |
+
"g_norm": 0.00042211293475702405,
|
| 393 |
+
"shape": [
|
| 394 |
+
768,
|
| 395 |
+
384,
|
| 396 |
+
3,
|
| 397 |
+
3
|
| 398 |
+
]
|
| 399 |
+
},
|
| 400 |
+
"enc_blocks.4.conv1.bias": {
|
| 401 |
+
"w_norm": 0.279443621635437,
|
| 402 |
+
"g_norm": 1.3359977856453042e-05,
|
| 403 |
+
"shape": [
|
| 404 |
+
768
|
| 405 |
+
]
|
| 406 |
+
},
|
| 407 |
+
"enc_blocks.4.t_proj.weight": {
|
| 408 |
+
"w_norm": 16.025165557861328,
|
| 409 |
+
"g_norm": 1.004903060675133e-05,
|
| 410 |
+
"shape": [
|
| 411 |
+
768,
|
| 412 |
+
256
|
| 413 |
+
]
|
| 414 |
+
},
|
| 415 |
+
"enc_blocks.4.t_proj.bias": {
|
| 416 |
+
"w_norm": 0.9901605248451233,
|
| 417 |
+
"g_norm": 1.3356218914850615e-05,
|
| 418 |
+
"shape": [
|
| 419 |
+
768
|
| 420 |
+
]
|
| 421 |
+
},
|
| 422 |
+
"enc_blocks.4.norm2.weight": {
|
| 423 |
+
"w_norm": 27.63109588623047,
|
| 424 |
+
"g_norm": 1.321973559242906e-05,
|
| 425 |
+
"shape": [
|
| 426 |
+
768
|
| 427 |
+
]
|
| 428 |
+
},
|
| 429 |
+
"enc_blocks.4.norm2.bias": {
|
| 430 |
+
"w_norm": 0.07533392310142517,
|
| 431 |
+
"g_norm": 1.3589736227004323e-05,
|
| 432 |
+
"shape": [
|
| 433 |
+
768
|
| 434 |
+
]
|
| 435 |
+
},
|
| 436 |
+
"enc_blocks.4.conv2.weight": {
|
| 437 |
+
"w_norm": 16.13887596130371,
|
| 438 |
+
"g_norm": 0.0006326561560854316,
|
| 439 |
+
"shape": [
|
| 440 |
+
768,
|
| 441 |
+
768,
|
| 442 |
+
3,
|
| 443 |
+
3
|
| 444 |
+
]
|
| 445 |
+
},
|
| 446 |
+
"enc_blocks.4.conv2.bias": {
|
| 447 |
+
"w_norm": 0.19331632554531097,
|
| 448 |
+
"g_norm": 1.6220848920056596e-05,
|
| 449 |
+
"shape": [
|
| 450 |
+
768
|
| 451 |
+
]
|
| 452 |
+
},
|
| 453 |
+
"enc_blocks.4.skip.weight": {
|
| 454 |
+
"w_norm": 15.950276374816895,
|
| 455 |
+
"g_norm": 0.0005513212527148426,
|
| 456 |
+
"shape": [
|
| 457 |
+
768,
|
| 458 |
+
384,
|
| 459 |
+
1,
|
| 460 |
+
1
|
| 461 |
+
]
|
| 462 |
+
},
|
| 463 |
+
"enc_blocks.4.skip.bias": {
|
| 464 |
+
"w_norm": 0.8089227676391602,
|
| 465 |
+
"g_norm": 1.6220848920056596e-05,
|
| 466 |
+
"shape": [
|
| 467 |
+
768
|
| 468 |
+
]
|
| 469 |
+
},
|
| 470 |
+
"enc_blocks.5.norm1.weight": {
|
| 471 |
+
"w_norm": 27.711660385131836,
|
| 472 |
+
"g_norm": 7.78115281718783e-06,
|
| 473 |
+
"shape": [
|
| 474 |
+
768
|
| 475 |
+
]
|
| 476 |
+
},
|
| 477 |
+
"enc_blocks.5.norm1.bias": {
|
| 478 |
+
"w_norm": 0.04768697917461395,
|
| 479 |
+
"g_norm": 6.8741815084649716e-06,
|
| 480 |
+
"shape": [
|
| 481 |
+
768
|
| 482 |
+
]
|
| 483 |
+
},
|
| 484 |
+
"enc_blocks.5.conv1.weight": {
|
| 485 |
+
"w_norm": 16.37407684326172,
|
| 486 |
+
"g_norm": 0.00023456118651665747,
|
| 487 |
+
"shape": [
|
| 488 |
+
768,
|
| 489 |
+
768,
|
| 490 |
+
3,
|
| 491 |
+
3
|
| 492 |
+
]
|
| 493 |
+
},
|
| 494 |
+
"enc_blocks.5.conv1.bias": {
|
| 495 |
+
"w_norm": 0.2019558846950531,
|
| 496 |
+
"g_norm": 4.756613179779379e-06,
|
| 497 |
+
"shape": [
|
| 498 |
+
768
|
| 499 |
+
]
|
| 500 |
+
},
|
| 501 |
+
"enc_blocks.5.t_proj.weight": {
|
| 502 |
+
"w_norm": 15.995716094970703,
|
| 503 |
+
"g_norm": 3.4388413041597232e-06,
|
| 504 |
+
"shape": [
|
| 505 |
+
768,
|
| 506 |
+
256
|
| 507 |
+
]
|
| 508 |
+
},
|
| 509 |
+
"enc_blocks.5.t_proj.bias": {
|
| 510 |
+
"w_norm": 0.9961099028587341,
|
| 511 |
+
"g_norm": 4.755159352498595e-06,
|
| 512 |
+
"shape": [
|
| 513 |
+
768
|
| 514 |
+
]
|
| 515 |
+
},
|
| 516 |
+
"enc_blocks.5.norm2.weight": {
|
| 517 |
+
"w_norm": 27.65757942199707,
|
| 518 |
+
"g_norm": 1.1370511856512167e-05,
|
| 519 |
+
"shape": [
|
| 520 |
+
768
|
| 521 |
+
]
|
| 522 |
+
},
|
| 523 |
+
"enc_blocks.5.norm2.bias": {
|
| 524 |
+
"w_norm": 0.06127138435840607,
|
| 525 |
+
"g_norm": 9.395703273185063e-06,
|
| 526 |
+
"shape": [
|
| 527 |
+
768
|
| 528 |
+
]
|
| 529 |
+
},
|
| 530 |
+
"enc_blocks.5.conv2.weight": {
|
| 531 |
+
"w_norm": 16.235984802246094,
|
| 532 |
+
"g_norm": 0.0004213395877741277,
|
| 533 |
+
"shape": [
|
| 534 |
+
768,
|
| 535 |
+
768,
|
| 536 |
+
3,
|
| 537 |
+
3
|
| 538 |
+
]
|
| 539 |
+
},
|
| 540 |
+
"enc_blocks.5.conv2.bias": {
|
| 541 |
+
"w_norm": 0.19209671020507812,
|
| 542 |
+
"g_norm": 1.1171143341925927e-05,
|
| 543 |
+
"shape": [
|
| 544 |
+
768
|
| 545 |
+
]
|
| 546 |
+
},
|
| 547 |
+
"enc_attns.3.norm.weight": {
|
| 548 |
+
"w_norm": 19.575828552246094,
|
| 549 |
+
"g_norm": 5.652593245031312e-06,
|
| 550 |
+
"shape": [
|
| 551 |
+
384
|
| 552 |
+
]
|
| 553 |
+
},
|
| 554 |
+
"enc_attns.3.norm.bias": {
|
| 555 |
+
"w_norm": 0.02052447572350502,
|
| 556 |
+
"g_norm": 7.894850568845868e-06,
|
| 557 |
+
"shape": [
|
| 558 |
+
384
|
| 559 |
+
]
|
| 560 |
+
},
|
| 561 |
+
"enc_attns.3.qkv.weight": {
|
| 562 |
+
"w_norm": 19.62198257446289,
|
| 563 |
+
"g_norm": 0.0001925578253576532,
|
| 564 |
+
"shape": [
|
| 565 |
+
1152,
|
| 566 |
+
384,
|
| 567 |
+
1,
|
| 568 |
+
1
|
| 569 |
+
]
|
| 570 |
+
},
|
| 571 |
+
"enc_attns.3.qkv.bias": {
|
| 572 |
+
"w_norm": 1.0065840482711792,
|
| 573 |
+
"g_norm": 1.3265251254779287e-05,
|
| 574 |
+
"shape": [
|
| 575 |
+
1152
|
| 576 |
+
]
|
| 577 |
+
},
|
| 578 |
+
"enc_attns.3.proj.weight": {
|
| 579 |
+
"w_norm": 11.29690170288086,
|
| 580 |
+
"g_norm": 0.0001960561639862135,
|
| 581 |
+
"shape": [
|
| 582 |
+
384,
|
| 583 |
+
384,
|
| 584 |
+
1,
|
| 585 |
+
1
|
| 586 |
+
]
|
| 587 |
+
},
|
| 588 |
+
"enc_attns.3.proj.bias": {
|
| 589 |
+
"w_norm": 0.5664379000663757,
|
| 590 |
+
"g_norm": 2.299301377206575e-05,
|
| 591 |
+
"shape": [
|
| 592 |
+
384
|
| 593 |
+
]
|
| 594 |
+
},
|
| 595 |
+
"enc_attns.4.norm.weight": {
|
| 596 |
+
"w_norm": 27.69475555419922,
|
| 597 |
+
"g_norm": 4.17779165218235e-06,
|
| 598 |
+
"shape": [
|
| 599 |
+
768
|
| 600 |
+
]
|
| 601 |
+
},
|
| 602 |
+
"enc_attns.4.norm.bias": {
|
| 603 |
+
"w_norm": 0.02993551641702652,
|
| 604 |
+
"g_norm": 6.423387731047114e-06,
|
| 605 |
+
"shape": [
|
| 606 |
+
768
|
| 607 |
+
]
|
| 608 |
+
},
|
| 609 |
+
"enc_attns.4.qkv.weight": {
|
| 610 |
+
"w_norm": 27.85542106628418,
|
| 611 |
+
"g_norm": 0.0001815570140024647,
|
| 612 |
+
"shape": [
|
| 613 |
+
2304,
|
| 614 |
+
768,
|
| 615 |
+
1,
|
| 616 |
+
1
|
| 617 |
+
]
|
| 618 |
+
},
|
| 619 |
+
"enc_attns.4.qkv.bias": {
|
| 620 |
+
"w_norm": 1.0133144855499268,
|
| 621 |
+
"g_norm": 9.848357876762748e-06,
|
| 622 |
+
"shape": [
|
| 623 |
+
2304
|
| 624 |
+
]
|
| 625 |
+
},
|
| 626 |
+
"enc_attns.4.proj.weight": {
|
| 627 |
+
"w_norm": 15.99245548248291,
|
| 628 |
+
"g_norm": 0.00018240074859932065,
|
| 629 |
+
"shape": [
|
| 630 |
+
768,
|
| 631 |
+
768,
|
| 632 |
+
1,
|
| 633 |
+
1
|
| 634 |
+
]
|
| 635 |
+
},
|
| 636 |
+
"enc_attns.4.proj.bias": {
|
| 637 |
+
"w_norm": 0.5735777020454407,
|
| 638 |
+
"g_norm": 1.5512972822762094e-05,
|
| 639 |
+
"shape": [
|
| 640 |
+
768
|
| 641 |
+
]
|
| 642 |
+
},
|
| 643 |
+
"enc_attns.5.norm.weight": {
|
| 644 |
+
"w_norm": 27.691572189331055,
|
| 645 |
+
"g_norm": 3.4669185424718307e-06,
|
| 646 |
+
"shape": [
|
| 647 |
+
768
|
| 648 |
+
]
|
| 649 |
+
},
|
| 650 |
+
"enc_attns.5.norm.bias": {
|
| 651 |
+
"w_norm": 0.028933243826031685,
|
| 652 |
+
"g_norm": 4.15879048887291e-06,
|
| 653 |
+
"shape": [
|
| 654 |
+
768
|
| 655 |
+
]
|
| 656 |
+
},
|
| 657 |
+
"enc_attns.5.qkv.weight": {
|
| 658 |
+
"w_norm": 27.788082122802734,
|
| 659 |
+
"g_norm": 0.00015120556054171175,
|
| 660 |
+
"shape": [
|
| 661 |
+
2304,
|
| 662 |
+
768,
|
| 663 |
+
1,
|
| 664 |
+
1
|
| 665 |
+
]
|
| 666 |
+
},
|
| 667 |
+
"enc_attns.5.qkv.bias": {
|
| 668 |
+
"w_norm": 0.9982998967170715,
|
| 669 |
+
"g_norm": 6.607310751860496e-06,
|
| 670 |
+
"shape": [
|
| 671 |
+
2304
|
| 672 |
+
]
|
| 673 |
+
},
|
| 674 |
+
"enc_attns.5.proj.weight": {
|
| 675 |
+
"w_norm": 16.01085662841797,
|
| 676 |
+
"g_norm": 0.00014587878831662238,
|
| 677 |
+
"shape": [
|
| 678 |
+
768,
|
| 679 |
+
768,
|
| 680 |
+
1,
|
| 681 |
+
1
|
| 682 |
+
]
|
| 683 |
+
},
|
| 684 |
+
"enc_attns.5.proj.bias": {
|
| 685 |
+
"w_norm": 0.596544086933136,
|
| 686 |
+
"g_norm": 1.08937556433375e-05,
|
| 687 |
+
"shape": [
|
| 688 |
+
768
|
| 689 |
+
]
|
| 690 |
+
},
|
| 691 |
+
"enc_downs.0.op.weight": {
|
| 692 |
+
"w_norm": 8.090078353881836,
|
| 693 |
+
"g_norm": 0.007355941459536552,
|
| 694 |
+
"shape": [
|
| 695 |
+
192,
|
| 696 |
+
192,
|
| 697 |
+
3,
|
| 698 |
+
3
|
| 699 |
+
]
|
| 700 |
+
},
|
| 701 |
+
"enc_downs.0.op.bias": {
|
| 702 |
+
"w_norm": 0.18642111122608185,
|
| 703 |
+
"g_norm": 0.000441454176325351,
|
| 704 |
+
"shape": [
|
| 705 |
+
192
|
| 706 |
+
]
|
| 707 |
+
},
|
| 708 |
+
"enc_downs.1.op.weight": {
|
| 709 |
+
"w_norm": 8.03896713256836,
|
| 710 |
+
"g_norm": 0.0029663110617548227,
|
| 711 |
+
"shape": [
|
| 712 |
+
192,
|
| 713 |
+
192,
|
| 714 |
+
3,
|
| 715 |
+
3
|
| 716 |
+
]
|
| 717 |
+
},
|
| 718 |
+
"enc_downs.1.op.bias": {
|
| 719 |
+
"w_norm": 0.1942720115184784,
|
| 720 |
+
"g_norm": 4.368715963209979e-05,
|
| 721 |
+
"shape": [
|
| 722 |
+
192
|
| 723 |
+
]
|
| 724 |
+
},
|
| 725 |
+
"enc_downs.2.op.weight": {
|
| 726 |
+
"w_norm": 11.369796752929688,
|
| 727 |
+
"g_norm": 0.002160790842026472,
|
| 728 |
+
"shape": [
|
| 729 |
+
384,
|
| 730 |
+
384,
|
| 731 |
+
3,
|
| 732 |
+
3
|
| 733 |
+
]
|
| 734 |
+
},
|
| 735 |
+
"enc_downs.2.op.bias": {
|
| 736 |
+
"w_norm": 0.18874460458755493,
|
| 737 |
+
"g_norm": 2.3616021280759014e-05,
|
| 738 |
+
"shape": [
|
| 739 |
+
384
|
| 740 |
+
]
|
| 741 |
+
},
|
| 742 |
+
"enc_downs.3.op.weight": {
|
| 743 |
+
"w_norm": 11.359977722167969,
|
| 744 |
+
"g_norm": 0.0012149059912189841,
|
| 745 |
+
"shape": [
|
| 746 |
+
384,
|
| 747 |
+
384,
|
| 748 |
+
3,
|
| 749 |
+
3
|
| 750 |
+
]
|
| 751 |
+
},
|
| 752 |
+
"enc_downs.3.op.bias": {
|
| 753 |
+
"w_norm": 0.1921733319759369,
|
| 754 |
+
"g_norm": 1.1493058082123753e-05,
|
| 755 |
+
"shape": [
|
| 756 |
+
384
|
| 757 |
+
]
|
| 758 |
+
},
|
| 759 |
+
"enc_downs.4.op.weight": {
|
| 760 |
+
"w_norm": 16.15582275390625,
|
| 761 |
+
"g_norm": 0.0011253517586737871,
|
| 762 |
+
"shape": [
|
| 763 |
+
768,
|
| 764 |
+
768,
|
| 765 |
+
3,
|
| 766 |
+
3
|
| 767 |
+
]
|
| 768 |
+
},
|
| 769 |
+
"enc_downs.4.op.bias": {
|
| 770 |
+
"w_norm": 0.19734814763069153,
|
| 771 |
+
"g_norm": 1.0824763194250409e-05,
|
| 772 |
+
"shape": [
|
| 773 |
+
768
|
| 774 |
+
]
|
| 775 |
+
},
|
| 776 |
+
"mid1.norm1.weight": {
|
| 777 |
+
"w_norm": 27.711822509765625,
|
| 778 |
+
"g_norm": 1.0373867553425953e-05,
|
| 779 |
+
"shape": [
|
| 780 |
+
768
|
| 781 |
+
]
|
| 782 |
+
},
|
| 783 |
+
"mid1.norm1.bias": {
|
| 784 |
+
"w_norm": 0.04126482829451561,
|
| 785 |
+
"g_norm": 6.879010015836684e-06,
|
| 786 |
+
"shape": [
|
| 787 |
+
768
|
| 788 |
+
]
|
| 789 |
+
},
|
| 790 |
+
"mid1.conv1.weight": {
|
| 791 |
+
"w_norm": 16.37283706665039,
|
| 792 |
+
"g_norm": 0.0003461484448052943,
|
| 793 |
+
"shape": [
|
| 794 |
+
768,
|
| 795 |
+
768,
|
| 796 |
+
3,
|
| 797 |
+
3
|
| 798 |
+
]
|
| 799 |
+
},
|
| 800 |
+
"mid1.conv1.bias": {
|
| 801 |
+
"w_norm": 0.20190510153770447,
|
| 802 |
+
"g_norm": 6.445853614422958e-06,
|
| 803 |
+
"shape": [
|
| 804 |
+
768
|
| 805 |
+
]
|
| 806 |
+
},
|
| 807 |
+
"mid1.t_proj.weight": {
|
| 808 |
+
"w_norm": 16.00591278076172,
|
| 809 |
+
"g_norm": 4.6713685151189566e-06,
|
| 810 |
+
"shape": [
|
| 811 |
+
768,
|
| 812 |
+
256
|
| 813 |
+
]
|
| 814 |
+
},
|
| 815 |
+
"mid1.t_proj.bias": {
|
| 816 |
+
"w_norm": 1.031727910041809,
|
| 817 |
+
"g_norm": 6.443968686653534e-06,
|
| 818 |
+
"shape": [
|
| 819 |
+
768
|
| 820 |
+
]
|
| 821 |
+
},
|
| 822 |
+
"mid1.norm2.weight": {
|
| 823 |
+
"w_norm": 27.664451599121094,
|
| 824 |
+
"g_norm": 9.174211299978197e-06,
|
| 825 |
+
"shape": [
|
| 826 |
+
768
|
| 827 |
+
]
|
| 828 |
+
},
|
| 829 |
+
"mid1.norm2.bias": {
|
| 830 |
+
"w_norm": 0.061578843742609024,
|
| 831 |
+
"g_norm": 6.349394425342325e-06,
|
| 832 |
+
"shape": [
|
| 833 |
+
768
|
| 834 |
+
]
|
| 835 |
+
},
|
| 836 |
+
"mid1.conv2.weight": {
|
| 837 |
+
"w_norm": 16.23827362060547,
|
| 838 |
+
"g_norm": 0.0003607813559938222,
|
| 839 |
+
"shape": [
|
| 840 |
+
768,
|
| 841 |
+
768,
|
| 842 |
+
3,
|
| 843 |
+
3
|
| 844 |
+
]
|
| 845 |
+
},
|
| 846 |
+
"mid1.conv2.bias": {
|
| 847 |
+
"w_norm": 0.19554272294044495,
|
| 848 |
+
"g_norm": 8.114502634271048e-06,
|
| 849 |
+
"shape": [
|
| 850 |
+
768
|
| 851 |
+
]
|
| 852 |
+
},
|
| 853 |
+
"mid_attn.norm.weight": {
|
| 854 |
+
"w_norm": 27.695194244384766,
|
| 855 |
+
"g_norm": 2.296697402925929e-06,
|
| 856 |
+
"shape": [
|
| 857 |
+
768
|
| 858 |
+
]
|
| 859 |
+
},
|
| 860 |
+
"mid_attn.norm.bias": {
|
| 861 |
+
"w_norm": 0.029120877385139465,
|
| 862 |
+
"g_norm": 3.081472186750034e-06,
|
| 863 |
+
"shape": [
|
| 864 |
+
768
|
| 865 |
+
]
|
| 866 |
+
},
|
| 867 |
+
"mid_attn.qkv.weight": {
|
| 868 |
+
"w_norm": 27.785205841064453,
|
| 869 |
+
"g_norm": 0.00010322411253582686,
|
| 870 |
+
"shape": [
|
| 871 |
+
2304,
|
| 872 |
+
768,
|
| 873 |
+
1,
|
| 874 |
+
1
|
| 875 |
+
]
|
| 876 |
+
},
|
| 877 |
+
"mid_attn.qkv.bias": {
|
| 878 |
+
"w_norm": 0.9866750240325928,
|
| 879 |
+
"g_norm": 4.757572241942398e-06,
|
| 880 |
+
"shape": [
|
| 881 |
+
2304
|
| 882 |
+
]
|
| 883 |
+
},
|
| 884 |
+
"mid_attn.proj.weight": {
|
| 885 |
+
"w_norm": 16.010156631469727,
|
| 886 |
+
"g_norm": 0.00010198546078754589,
|
| 887 |
+
"shape": [
|
| 888 |
+
768,
|
| 889 |
+
768,
|
| 890 |
+
1,
|
| 891 |
+
1
|
| 892 |
+
]
|
| 893 |
+
},
|
| 894 |
+
"mid_attn.proj.bias": {
|
| 895 |
+
"w_norm": 0.5767161846160889,
|
| 896 |
+
"g_norm": 7.791844836901873e-06,
|
| 897 |
+
"shape": [
|
| 898 |
+
768
|
| 899 |
+
]
|
| 900 |
+
},
|
| 901 |
+
"mid2.norm1.weight": {
|
| 902 |
+
"w_norm": 27.712369918823242,
|
| 903 |
+
"g_norm": 1.7629792637308128e-05,
|
| 904 |
+
"shape": [
|
| 905 |
+
768
|
| 906 |
+
]
|
| 907 |
+
},
|
| 908 |
+
"mid2.norm1.bias": {
|
| 909 |
+
"w_norm": 0.041550856083631516,
|
| 910 |
+
"g_norm": 1.2748863809974864e-05,
|
| 911 |
+
"shape": [
|
| 912 |
+
768
|
| 913 |
+
]
|
| 914 |
+
},
|
| 915 |
+
"mid2.conv1.weight": {
|
| 916 |
+
"w_norm": 16.345983505249023,
|
| 917 |
+
"g_norm": 0.0003871308290399611,
|
| 918 |
+
"shape": [
|
| 919 |
+
768,
|
| 920 |
+
768,
|
| 921 |
+
3,
|
| 922 |
+
3
|
| 923 |
+
]
|
| 924 |
+
},
|
| 925 |
+
"mid2.conv1.bias": {
|
| 926 |
+
"w_norm": 0.19644778966903687,
|
| 927 |
+
"g_norm": 7.461239420081256e-06,
|
| 928 |
+
"shape": [
|
| 929 |
+
768
|
| 930 |
+
]
|
| 931 |
+
},
|
| 932 |
+
"mid2.t_proj.weight": {
|
| 933 |
+
"w_norm": 15.993643760681152,
|
| 934 |
+
"g_norm": 5.395043444877956e-06,
|
| 935 |
+
"shape": [
|
| 936 |
+
768,
|
| 937 |
+
256
|
| 938 |
+
]
|
| 939 |
+
},
|
| 940 |
+
"mid2.t_proj.bias": {
|
| 941 |
+
"w_norm": 0.9956456422805786,
|
| 942 |
+
"g_norm": 7.462194844265468e-06,
|
| 943 |
+
"shape": [
|
| 944 |
+
768
|
| 945 |
+
]
|
| 946 |
+
},
|
| 947 |
+
"mid2.norm2.weight": {
|
| 948 |
+
"w_norm": 27.654315948486328,
|
| 949 |
+
"g_norm": 9.289041372539941e-06,
|
| 950 |
+
"shape": [
|
| 951 |
+
768
|
| 952 |
+
]
|
| 953 |
+
},
|
| 954 |
+
"mid2.norm2.bias": {
|
| 955 |
+
"w_norm": 0.06868414580821991,
|
| 956 |
+
"g_norm": 7.056999493215699e-06,
|
| 957 |
+
"shape": [
|
| 958 |
+
768
|
| 959 |
+
]
|
| 960 |
+
},
|
| 961 |
+
"mid2.conv2.weight": {
|
| 962 |
+
"w_norm": 16.199613571166992,
|
| 963 |
+
"g_norm": 0.00027969831717200577,
|
| 964 |
+
"shape": [
|
| 965 |
+
768,
|
| 966 |
+
768,
|
| 967 |
+
3,
|
| 968 |
+
3
|
| 969 |
+
]
|
| 970 |
+
},
|
| 971 |
+
"mid2.conv2.bias": {
|
| 972 |
+
"w_norm": 0.19018852710723877,
|
| 973 |
+
"g_norm": 7.015074061200721e-06,
|
| 974 |
+
"shape": [
|
| 975 |
+
768
|
| 976 |
+
]
|
| 977 |
+
},
|
| 978 |
+
"dec_blocks.0.norm1.weight": {
|
| 979 |
+
"w_norm": 39.191375732421875,
|
| 980 |
+
"g_norm": 1.2209551641717553e-05,
|
| 981 |
+
"shape": [
|
| 982 |
+
1536
|
| 983 |
+
]
|
| 984 |
+
},
|
| 985 |
+
"dec_blocks.0.norm1.bias": {
|
| 986 |
+
"w_norm": 0.05582411214709282,
|
| 987 |
+
"g_norm": 8.90219962457195e-06,
|
| 988 |
+
"shape": [
|
| 989 |
+
1536
|
| 990 |
+
]
|
| 991 |
+
},
|
| 992 |
+
"dec_blocks.0.conv1.weight": {
|
| 993 |
+
"w_norm": 16.71096420288086,
|
| 994 |
+
"g_norm": 0.0003418675623834133,
|
| 995 |
+
"shape": [
|
| 996 |
+
768,
|
| 997 |
+
1536,
|
| 998 |
+
3,
|
| 999 |
+
3
|
| 1000 |
+
]
|
| 1001 |
+
},
|
| 1002 |
+
"dec_blocks.0.conv1.bias": {
|
| 1003 |
+
"w_norm": 0.1404905915260315,
|
| 1004 |
+
"g_norm": 5.017229796067113e-06,
|
| 1005 |
+
"shape": [
|
| 1006 |
+
768
|
| 1007 |
+
]
|
| 1008 |
+
},
|
| 1009 |
+
"dec_blocks.0.t_proj.weight": {
|
| 1010 |
+
"w_norm": 16.009931564331055,
|
| 1011 |
+
"g_norm": 3.6284368434280623e-06,
|
| 1012 |
+
"shape": [
|
| 1013 |
+
768,
|
| 1014 |
+
256
|
| 1015 |
+
]
|
| 1016 |
+
},
|
| 1017 |
+
"dec_blocks.0.t_proj.bias": {
|
| 1018 |
+
"w_norm": 1.0188021659851074,
|
| 1019 |
+
"g_norm": 5.017205239710165e-06,
|
| 1020 |
+
"shape": [
|
| 1021 |
+
768
|
| 1022 |
+
]
|
| 1023 |
+
},
|
| 1024 |
+
"dec_blocks.0.norm2.weight": {
|
| 1025 |
+
"w_norm": 27.624052047729492,
|
| 1026 |
+
"g_norm": 1.1853226169478148e-05,
|
| 1027 |
+
"shape": [
|
| 1028 |
+
768
|
| 1029 |
+
]
|
| 1030 |
+
},
|
| 1031 |
+
"dec_blocks.0.norm2.bias": {
|
| 1032 |
+
"w_norm": 0.08831377327442169,
|
| 1033 |
+
"g_norm": 1.0930866665148642e-05,
|
| 1034 |
+
"shape": [
|
| 1035 |
+
768
|
| 1036 |
+
]
|
| 1037 |
+
},
|
| 1038 |
+
"dec_blocks.0.conv2.weight": {
|
| 1039 |
+
"w_norm": 16.116975784301758,
|
| 1040 |
+
"g_norm": 0.0005032767076045275,
|
| 1041 |
+
"shape": [
|
| 1042 |
+
768,
|
| 1043 |
+
768,
|
| 1044 |
+
3,
|
| 1045 |
+
3
|
| 1046 |
+
]
|
| 1047 |
+
},
|
| 1048 |
+
"dec_blocks.0.conv2.bias": {
|
| 1049 |
+
"w_norm": 0.19690799713134766,
|
| 1050 |
+
"g_norm": 1.4778754120925441e-05,
|
| 1051 |
+
"shape": [
|
| 1052 |
+
768
|
| 1053 |
+
]
|
| 1054 |
+
},
|
| 1055 |
+
"dec_blocks.0.skip.weight": {
|
| 1056 |
+
"w_norm": 15.933785438537598,
|
| 1057 |
+
"g_norm": 0.0012050033546984196,
|
| 1058 |
+
"shape": [
|
| 1059 |
+
768,
|
| 1060 |
+
1536,
|
| 1061 |
+
1,
|
| 1062 |
+
1
|
| 1063 |
+
]
|
| 1064 |
+
},
|
| 1065 |
+
"dec_blocks.0.skip.bias": {
|
| 1066 |
+
"w_norm": 0.41081660985946655,
|
| 1067 |
+
"g_norm": 1.4778754120925441e-05,
|
| 1068 |
+
"shape": [
|
| 1069 |
+
768
|
| 1070 |
+
]
|
| 1071 |
+
},
|
| 1072 |
+
"dec_blocks.1.norm1.weight": {
|
| 1073 |
+
"w_norm": 39.185028076171875,
|
| 1074 |
+
"g_norm": 4.89912235934753e-06,
|
| 1075 |
+
"shape": [
|
| 1076 |
+
1536
|
| 1077 |
+
]
|
| 1078 |
+
},
|
| 1079 |
+
"dec_blocks.1.norm1.bias": {
|
| 1080 |
+
"w_norm": 0.052894700318574905,
|
| 1081 |
+
"g_norm": 4.081042334291851e-06,
|
| 1082 |
+
"shape": [
|
| 1083 |
+
1536
|
| 1084 |
+
]
|
| 1085 |
+
},
|
| 1086 |
+
"dec_blocks.1.conv1.weight": {
|
| 1087 |
+
"w_norm": 16.64948272705078,
|
| 1088 |
+
"g_norm": 0.0003468487993814051,
|
| 1089 |
+
"shape": [
|
| 1090 |
+
768,
|
| 1091 |
+
1536,
|
| 1092 |
+
3,
|
| 1093 |
+
3
|
| 1094 |
+
]
|
| 1095 |
+
},
|
| 1096 |
+
"dec_blocks.1.conv1.bias": {
|
| 1097 |
+
"w_norm": 0.13766644895076752,
|
| 1098 |
+
"g_norm": 5.367211088014301e-06,
|
| 1099 |
+
"shape": [
|
| 1100 |
+
768
|
| 1101 |
+
]
|
| 1102 |
+
},
|
| 1103 |
+
"dec_blocks.1.t_proj.weight": {
|
| 1104 |
+
"w_norm": 16.036720275878906,
|
| 1105 |
+
"g_norm": 4.022125267511001e-06,
|
| 1106 |
+
"shape": [
|
| 1107 |
+
768,
|
| 1108 |
+
256
|
| 1109 |
+
]
|
| 1110 |
+
},
|
| 1111 |
+
"dec_blocks.1.t_proj.bias": {
|
| 1112 |
+
"w_norm": 0.9849644899368286,
|
| 1113 |
+
"g_norm": 5.367090125218965e-06,
|
| 1114 |
+
"shape": [
|
| 1115 |
+
768
|
| 1116 |
+
]
|
| 1117 |
+
},
|
| 1118 |
+
"dec_blocks.1.norm2.weight": {
|
| 1119 |
+
"w_norm": 27.63217544555664,
|
| 1120 |
+
"g_norm": 1.2087589311704505e-05,
|
| 1121 |
+
"shape": [
|
| 1122 |
+
768
|
| 1123 |
+
]
|
| 1124 |
+
},
|
| 1125 |
+
"dec_blocks.1.norm2.bias": {
|
| 1126 |
+
"w_norm": 0.08338069170713425,
|
| 1127 |
+
"g_norm": 1.302398322877707e-05,
|
| 1128 |
+
"shape": [
|
| 1129 |
+
768
|
| 1130 |
+
]
|
| 1131 |
+
},
|
| 1132 |
+
"dec_blocks.1.conv2.weight": {
|
| 1133 |
+
"w_norm": 16.113073348999023,
|
| 1134 |
+
"g_norm": 0.0011885776184499264,
|
| 1135 |
+
"shape": [
|
| 1136 |
+
768,
|
| 1137 |
+
768,
|
| 1138 |
+
3,
|
| 1139 |
+
3
|
| 1140 |
+
]
|
| 1141 |
+
},
|
| 1142 |
+
"dec_blocks.1.conv2.bias": {
|
| 1143 |
+
"w_norm": 0.1954251527786255,
|
| 1144 |
+
"g_norm": 2.834928091033362e-05,
|
| 1145 |
+
"shape": [
|
| 1146 |
+
768
|
| 1147 |
+
]
|
| 1148 |
+
},
|
| 1149 |
+
"dec_blocks.1.skip.weight": {
|
| 1150 |
+
"w_norm": 15.929043769836426,
|
| 1151 |
+
"g_norm": 0.0016937535256147385,
|
| 1152 |
+
"shape": [
|
| 1153 |
+
768,
|
| 1154 |
+
1536,
|
| 1155 |
+
1,
|
| 1156 |
+
1
|
| 1157 |
+
]
|
| 1158 |
+
},
|
| 1159 |
+
"dec_blocks.1.skip.bias": {
|
| 1160 |
+
"w_norm": 0.40631330013275146,
|
| 1161 |
+
"g_norm": 2.834928091033362e-05,
|
| 1162 |
+
"shape": [
|
| 1163 |
+
768
|
| 1164 |
+
]
|
| 1165 |
+
},
|
| 1166 |
+
"dec_blocks.2.norm1.weight": {
|
| 1167 |
+
"w_norm": 33.94264602661133,
|
| 1168 |
+
"g_norm": 2.3586289898958057e-05,
|
| 1169 |
+
"shape": [
|
| 1170 |
+
1152
|
| 1171 |
+
]
|
| 1172 |
+
},
|
| 1173 |
+
"dec_blocks.2.norm1.bias": {
|
| 1174 |
+
"w_norm": 0.05327305942773819,
|
| 1175 |
+
"g_norm": 1.8042264855466783e-05,
|
| 1176 |
+
"shape": [
|
| 1177 |
+
1152
|
| 1178 |
+
]
|
| 1179 |
+
},
|
| 1180 |
+
"dec_blocks.2.conv1.weight": {
|
| 1181 |
+
"w_norm": 11.731250762939453,
|
| 1182 |
+
"g_norm": 0.0008712481940165162,
|
| 1183 |
+
"shape": [
|
| 1184 |
+
384,
|
| 1185 |
+
1152,
|
| 1186 |
+
3,
|
| 1187 |
+
3
|
| 1188 |
+
]
|
| 1189 |
+
},
|
| 1190 |
+
"dec_blocks.2.conv1.bias": {
|
| 1191 |
+
"w_norm": 0.11506623029708862,
|
| 1192 |
+
"g_norm": 1.4161601939122193e-05,
|
| 1193 |
+
"shape": [
|
| 1194 |
+
384
|
| 1195 |
+
]
|
| 1196 |
+
},
|
| 1197 |
+
"dec_blocks.2.t_proj.weight": {
|
| 1198 |
+
"w_norm": 11.318585395812988,
|
| 1199 |
+
"g_norm": 1.028281531034736e-05,
|
| 1200 |
+
"shape": [
|
| 1201 |
+
384,
|
| 1202 |
+
256
|
| 1203 |
+
]
|
| 1204 |
+
},
|
| 1205 |
+
"dec_blocks.2.t_proj.bias": {
|
| 1206 |
+
"w_norm": 0.7085891962051392,
|
| 1207 |
+
"g_norm": 1.4159787497192156e-05,
|
| 1208 |
+
"shape": [
|
| 1209 |
+
384
|
| 1210 |
+
]
|
| 1211 |
+
},
|
| 1212 |
+
"dec_blocks.2.norm2.weight": {
|
| 1213 |
+
"w_norm": 19.5445556640625,
|
| 1214 |
+
"g_norm": 4.0533999708713964e-05,
|
| 1215 |
+
"shape": [
|
| 1216 |
+
384
|
| 1217 |
+
]
|
| 1218 |
+
},
|
| 1219 |
+
"dec_blocks.2.norm2.bias": {
|
| 1220 |
+
"w_norm": 0.05140449106693268,
|
| 1221 |
+
"g_norm": 3.8357742596417665e-05,
|
| 1222 |
+
"shape": [
|
| 1223 |
+
384
|
| 1224 |
+
]
|
| 1225 |
+
},
|
| 1226 |
+
"dec_blocks.2.conv2.weight": {
|
| 1227 |
+
"w_norm": 11.374390602111816,
|
| 1228 |
+
"g_norm": 0.0014708536909893155,
|
| 1229 |
+
"shape": [
|
| 1230 |
+
384,
|
| 1231 |
+
384,
|
| 1232 |
+
3,
|
| 1233 |
+
3
|
| 1234 |
+
]
|
| 1235 |
+
},
|
| 1236 |
+
"dec_blocks.2.conv2.bias": {
|
| 1237 |
+
"w_norm": 0.18952299654483795,
|
| 1238 |
+
"g_norm": 5.938815957051702e-05,
|
| 1239 |
+
"shape": [
|
| 1240 |
+
384
|
| 1241 |
+
]
|
| 1242 |
+
},
|
| 1243 |
+
"dec_blocks.2.skip.weight": {
|
| 1244 |
+
"w_norm": 11.216829299926758,
|
| 1245 |
+
"g_norm": 0.004319904837757349,
|
| 1246 |
+
"shape": [
|
| 1247 |
+
384,
|
| 1248 |
+
1152,
|
| 1249 |
+
1,
|
| 1250 |
+
1
|
| 1251 |
+
]
|
| 1252 |
+
},
|
| 1253 |
+
"dec_blocks.2.skip.bias": {
|
| 1254 |
+
"w_norm": 0.3324746787548065,
|
| 1255 |
+
"g_norm": 5.938815957051702e-05,
|
| 1256 |
+
"shape": [
|
| 1257 |
+
384
|
| 1258 |
+
]
|
| 1259 |
+
},
|
| 1260 |
+
"dec_blocks.3.norm1.weight": {
|
| 1261 |
+
"w_norm": 27.71185874938965,
|
| 1262 |
+
"g_norm": 3.5906316043110564e-05,
|
| 1263 |
+
"shape": [
|
| 1264 |
+
768
|
| 1265 |
+
]
|
| 1266 |
+
},
|
| 1267 |
+
"dec_blocks.3.norm1.bias": {
|
| 1268 |
+
"w_norm": 0.04073373228311539,
|
| 1269 |
+
"g_norm": 3.671555532491766e-05,
|
| 1270 |
+
"shape": [
|
| 1271 |
+
768
|
| 1272 |
+
]
|
| 1273 |
+
},
|
| 1274 |
+
"dec_blocks.3.conv1.weight": {
|
| 1275 |
+
"w_norm": 11.635823249816895,
|
| 1276 |
+
"g_norm": 0.0019167193677276373,
|
| 1277 |
+
"shape": [
|
| 1278 |
+
384,
|
| 1279 |
+
768,
|
| 1280 |
+
3,
|
| 1281 |
+
3
|
| 1282 |
+
]
|
| 1283 |
+
},
|
| 1284 |
+
"dec_blocks.3.conv1.bias": {
|
| 1285 |
+
"w_norm": 0.14201989769935608,
|
| 1286 |
+
"g_norm": 4.213491774862632e-05,
|
| 1287 |
+
"shape": [
|
| 1288 |
+
384
|
| 1289 |
+
]
|
| 1290 |
+
},
|
| 1291 |
+
"dec_blocks.3.t_proj.weight": {
|
| 1292 |
+
"w_norm": 11.332667350769043,
|
| 1293 |
+
"g_norm": 3.186779940733686e-05,
|
| 1294 |
+
"shape": [
|
| 1295 |
+
384,
|
| 1296 |
+
256
|
| 1297 |
+
]
|
| 1298 |
+
},
|
| 1299 |
+
"dec_blocks.3.t_proj.bias": {
|
| 1300 |
+
"w_norm": 0.7082791328430176,
|
| 1301 |
+
"g_norm": 4.213417923892848e-05,
|
| 1302 |
+
"shape": [
|
| 1303 |
+
384
|
| 1304 |
+
]
|
| 1305 |
+
},
|
| 1306 |
+
"dec_blocks.3.norm2.weight": {
|
| 1307 |
+
"w_norm": 19.534440994262695,
|
| 1308 |
+
"g_norm": 6.097566074458882e-05,
|
| 1309 |
+
"shape": [
|
| 1310 |
+
384
|
| 1311 |
+
]
|
| 1312 |
+
},
|
| 1313 |
+
"dec_blocks.3.norm2.bias": {
|
| 1314 |
+
"w_norm": 0.048718083649873734,
|
| 1315 |
+
"g_norm": 8.37421539472416e-05,
|
| 1316 |
+
"shape": [
|
| 1317 |
+
384
|
| 1318 |
+
]
|
| 1319 |
+
},
|
| 1320 |
+
"dec_blocks.3.conv2.weight": {
|
| 1321 |
+
"w_norm": 11.369616508483887,
|
| 1322 |
+
"g_norm": 0.003669221419841051,
|
| 1323 |
+
"shape": [
|
| 1324 |
+
384,
|
| 1325 |
+
384,
|
| 1326 |
+
3,
|
| 1327 |
+
3
|
| 1328 |
+
]
|
| 1329 |
+
},
|
| 1330 |
+
"dec_blocks.3.conv2.bias": {
|
| 1331 |
+
"w_norm": 0.19281557202339172,
|
| 1332 |
+
"g_norm": 0.00017457322974223644,
|
| 1333 |
+
"shape": [
|
| 1334 |
+
384
|
| 1335 |
+
]
|
| 1336 |
+
},
|
| 1337 |
+
"dec_blocks.3.skip.weight": {
|
| 1338 |
+
"w_norm": 11.228267669677734,
|
| 1339 |
+
"g_norm": 0.0061339568346738815,
|
| 1340 |
+
"shape": [
|
| 1341 |
+
384,
|
| 1342 |
+
768,
|
| 1343 |
+
1,
|
| 1344 |
+
1
|
| 1345 |
+
]
|
| 1346 |
+
},
|
| 1347 |
+
"dec_blocks.3.skip.bias": {
|
| 1348 |
+
"w_norm": 0.4074385464191437,
|
| 1349 |
+
"g_norm": 0.00017457322974223644,
|
| 1350 |
+
"shape": [
|
| 1351 |
+
384
|
| 1352 |
+
]
|
| 1353 |
+
},
|
| 1354 |
+
"dec_blocks.4.norm1.weight": {
|
| 1355 |
+
"w_norm": 23.997482299804688,
|
| 1356 |
+
"g_norm": 0.00011013081530109048,
|
| 1357 |
+
"shape": [
|
| 1358 |
+
576
|
| 1359 |
+
]
|
| 1360 |
+
},
|
| 1361 |
+
"dec_blocks.4.norm1.bias": {
|
| 1362 |
+
"w_norm": 0.043228741735219955,
|
| 1363 |
+
"g_norm": 0.00013270195631776005,
|
| 1364 |
+
"shape": [
|
| 1365 |
+
576
|
| 1366 |
+
]
|
| 1367 |
+
},
|
| 1368 |
+
"dec_blocks.4.conv1.weight": {
|
| 1369 |
+
"w_norm": 8.184913635253906,
|
| 1370 |
+
"g_norm": 0.005579778924584389,
|
| 1371 |
+
"shape": [
|
| 1372 |
+
192,
|
| 1373 |
+
576,
|
| 1374 |
+
3,
|
| 1375 |
+
3
|
| 1376 |
+
]
|
| 1377 |
+
},
|
| 1378 |
+
"dec_blocks.4.conv1.bias": {
|
| 1379 |
+
"w_norm": 0.11560454964637756,
|
| 1380 |
+
"g_norm": 0.00015491456724703312,
|
| 1381 |
+
"shape": [
|
| 1382 |
+
192
|
| 1383 |
+
]
|
| 1384 |
+
},
|
| 1385 |
+
"dec_blocks.4.t_proj.weight": {
|
| 1386 |
+
"w_norm": 7.999128341674805,
|
| 1387 |
+
"g_norm": 0.00011237112630624324,
|
| 1388 |
+
"shape": [
|
| 1389 |
+
192,
|
| 1390 |
+
256
|
| 1391 |
+
]
|
| 1392 |
+
},
|
| 1393 |
+
"dec_blocks.4.t_proj.bias": {
|
| 1394 |
+
"w_norm": 0.4804953634738922,
|
| 1395 |
+
"g_norm": 0.0001549788867123425,
|
| 1396 |
+
"shape": [
|
| 1397 |
+
192
|
| 1398 |
+
]
|
| 1399 |
+
},
|
| 1400 |
+
"dec_blocks.4.norm2.weight": {
|
| 1401 |
+
"w_norm": 13.831470489501953,
|
| 1402 |
+
"g_norm": 0.00043151574209332466,
|
| 1403 |
+
"shape": [
|
| 1404 |
+
192
|
| 1405 |
+
]
|
| 1406 |
+
},
|
| 1407 |
+
"dec_blocks.4.norm2.bias": {
|
| 1408 |
+
"w_norm": 0.01964525133371353,
|
| 1409 |
+
"g_norm": 0.00037272763438522816,
|
| 1410 |
+
"shape": [
|
| 1411 |
+
192
|
| 1412 |
+
]
|
| 1413 |
+
},
|
| 1414 |
+
"dec_blocks.4.conv2.weight": {
|
| 1415 |
+
"w_norm": 8.019150733947754,
|
| 1416 |
+
"g_norm": 0.011501150205731392,
|
| 1417 |
+
"shape": [
|
| 1418 |
+
192,
|
| 1419 |
+
192,
|
| 1420 |
+
3,
|
| 1421 |
+
3
|
| 1422 |
+
]
|
| 1423 |
+
},
|
| 1424 |
+
"dec_blocks.4.conv2.bias": {
|
| 1425 |
+
"w_norm": 0.19350716471672058,
|
| 1426 |
+
"g_norm": 0.0006961687467992306,
|
| 1427 |
+
"shape": [
|
| 1428 |
+
192
|
| 1429 |
+
]
|
| 1430 |
+
},
|
| 1431 |
+
"dec_blocks.4.skip.weight": {
|
| 1432 |
+
"w_norm": 7.952949047088623,
|
| 1433 |
+
"g_norm": 0.01609424687922001,
|
| 1434 |
+
"shape": [
|
| 1435 |
+
192,
|
| 1436 |
+
576,
|
| 1437 |
+
1,
|
| 1438 |
+
1
|
| 1439 |
+
]
|
| 1440 |
+
},
|
| 1441 |
+
"dec_blocks.4.skip.bias": {
|
| 1442 |
+
"w_norm": 0.34671664237976074,
|
| 1443 |
+
"g_norm": 0.0006961687467992306,
|
| 1444 |
+
"shape": [
|
| 1445 |
+
192
|
| 1446 |
+
]
|
| 1447 |
+
},
|
| 1448 |
+
"dec_blocks.5.norm1.weight": {
|
| 1449 |
+
"w_norm": 19.585325241088867,
|
| 1450 |
+
"g_norm": 0.001735931378789246,
|
| 1451 |
+
"shape": [
|
| 1452 |
+
384
|
| 1453 |
+
]
|
| 1454 |
+
},
|
| 1455 |
+
"dec_blocks.5.norm1.bias": {
|
| 1456 |
+
"w_norm": 0.036612581461668015,
|
| 1457 |
+
"g_norm": 0.0026527568697929382,
|
| 1458 |
+
"shape": [
|
| 1459 |
+
384
|
| 1460 |
+
]
|
| 1461 |
+
},
|
| 1462 |
+
"dec_blocks.5.conv1.weight": {
|
| 1463 |
+
"w_norm": 8.18410873413086,
|
| 1464 |
+
"g_norm": 0.037934672087430954,
|
| 1465 |
+
"shape": [
|
| 1466 |
+
192,
|
| 1467 |
+
384,
|
| 1468 |
+
3,
|
| 1469 |
+
3
|
| 1470 |
+
]
|
| 1471 |
+
},
|
| 1472 |
+
"dec_blocks.5.conv1.bias": {
|
| 1473 |
+
"w_norm": 0.13837724924087524,
|
| 1474 |
+
"g_norm": 0.0017917185323312879,
|
| 1475 |
+
"shape": [
|
| 1476 |
+
192
|
| 1477 |
+
]
|
| 1478 |
+
},
|
| 1479 |
+
"dec_blocks.5.t_proj.weight": {
|
| 1480 |
+
"w_norm": 8.034667015075684,
|
| 1481 |
+
"g_norm": 0.0013399251038208604,
|
| 1482 |
+
"shape": [
|
| 1483 |
+
192,
|
| 1484 |
+
256
|
| 1485 |
+
]
|
| 1486 |
+
},
|
| 1487 |
+
"dec_blocks.5.t_proj.bias": {
|
| 1488 |
+
"w_norm": 0.5041279792785645,
|
| 1489 |
+
"g_norm": 0.0017916002543643117,
|
| 1490 |
+
"shape": [
|
| 1491 |
+
192
|
| 1492 |
+
]
|
| 1493 |
+
},
|
| 1494 |
+
"dec_blocks.5.norm2.weight": {
|
| 1495 |
+
"w_norm": 13.868912696838379,
|
| 1496 |
+
"g_norm": 0.0018698613857850432,
|
| 1497 |
+
"shape": [
|
| 1498 |
+
192
|
| 1499 |
+
]
|
| 1500 |
+
},
|
| 1501 |
+
"dec_blocks.5.norm2.bias": {
|
| 1502 |
+
"w_norm": 0.022572237998247147,
|
| 1503 |
+
"g_norm": 0.0022469060495495796,
|
| 1504 |
+
"shape": [
|
| 1505 |
+
192
|
| 1506 |
+
]
|
| 1507 |
+
},
|
| 1508 |
+
"dec_blocks.5.conv2.weight": {
|
| 1509 |
+
"w_norm": 8.10611343383789,
|
| 1510 |
+
"g_norm": 0.0494694784283638,
|
| 1511 |
+
"shape": [
|
| 1512 |
+
192,
|
| 1513 |
+
192,
|
| 1514 |
+
3,
|
| 1515 |
+
3
|
| 1516 |
+
]
|
| 1517 |
+
},
|
| 1518 |
+
"dec_blocks.5.conv2.bias": {
|
| 1519 |
+
"w_norm": 0.19638051092624664,
|
| 1520 |
+
"g_norm": 0.002709712367504835,
|
| 1521 |
+
"shape": [
|
| 1522 |
+
192
|
| 1523 |
+
]
|
| 1524 |
+
},
|
| 1525 |
+
"dec_blocks.5.skip.weight": {
|
| 1526 |
+
"w_norm": 8.038369178771973,
|
| 1527 |
+
"g_norm": 0.05201084911823273,
|
| 1528 |
+
"shape": [
|
| 1529 |
+
192,
|
| 1530 |
+
384,
|
| 1531 |
+
1,
|
| 1532 |
+
1
|
| 1533 |
+
]
|
| 1534 |
+
},
|
| 1535 |
+
"dec_blocks.5.skip.bias": {
|
| 1536 |
+
"w_norm": 0.41981905698776245,
|
| 1537 |
+
"g_norm": 0.002709712367504835,
|
| 1538 |
+
"shape": [
|
| 1539 |
+
192
|
| 1540 |
+
]
|
| 1541 |
+
},
|
| 1542 |
+
"dec_attns.0.norm.weight": {
|
| 1543 |
+
"w_norm": 27.69066047668457,
|
| 1544 |
+
"g_norm": 7.059255040076096e-06,
|
| 1545 |
+
"shape": [
|
| 1546 |
+
768
|
| 1547 |
+
]
|
| 1548 |
+
},
|
| 1549 |
+
"dec_attns.0.norm.bias": {
|
| 1550 |
+
"w_norm": 0.022313563153147697,
|
| 1551 |
+
"g_norm": 7.783974979247432e-06,
|
| 1552 |
+
"shape": [
|
| 1553 |
+
768
|
| 1554 |
+
]
|
| 1555 |
+
},
|
| 1556 |
+
"dec_attns.0.qkv.weight": {
|
| 1557 |
+
"w_norm": 27.7947940826416,
|
| 1558 |
+
"g_norm": 0.0002512983337510377,
|
| 1559 |
+
"shape": [
|
| 1560 |
+
2304,
|
| 1561 |
+
768,
|
| 1562 |
+
1,
|
| 1563 |
+
1
|
| 1564 |
+
]
|
| 1565 |
+
},
|
| 1566 |
+
"dec_attns.0.qkv.bias": {
|
| 1567 |
+
"w_norm": 1.0090641975402832,
|
| 1568 |
+
"g_norm": 9.96327435132116e-06,
|
| 1569 |
+
"shape": [
|
| 1570 |
+
2304
|
| 1571 |
+
]
|
| 1572 |
+
},
|
| 1573 |
+
"dec_attns.0.proj.weight": {
|
| 1574 |
+
"w_norm": 16.015512466430664,
|
| 1575 |
+
"g_norm": 0.00023575192608404905,
|
| 1576 |
+
"shape": [
|
| 1577 |
+
768,
|
| 1578 |
+
768,
|
| 1579 |
+
1,
|
| 1580 |
+
1
|
| 1581 |
+
]
|
| 1582 |
+
},
|
| 1583 |
+
"dec_attns.0.proj.bias": {
|
| 1584 |
+
"w_norm": 0.5766009092330933,
|
| 1585 |
+
"g_norm": 1.5225058632495347e-05,
|
| 1586 |
+
"shape": [
|
| 1587 |
+
768
|
| 1588 |
+
]
|
| 1589 |
+
},
|
| 1590 |
+
"dec_attns.1.norm.weight": {
|
| 1591 |
+
"w_norm": 27.70824432373047,
|
| 1592 |
+
"g_norm": 1.0122570529347286e-05,
|
| 1593 |
+
"shape": [
|
| 1594 |
+
768
|
| 1595 |
+
]
|
| 1596 |
+
},
|
| 1597 |
+
"dec_attns.1.norm.bias": {
|
| 1598 |
+
"w_norm": 0.025859301909804344,
|
| 1599 |
+
"g_norm": 1.2997406884096563e-05,
|
| 1600 |
+
"shape": [
|
| 1601 |
+
768
|
| 1602 |
+
]
|
| 1603 |
+
},
|
| 1604 |
+
"dec_attns.1.qkv.weight": {
|
| 1605 |
+
"w_norm": 27.807323455810547,
|
| 1606 |
+
"g_norm": 0.00034215542837046087,
|
| 1607 |
+
"shape": [
|
| 1608 |
+
2304,
|
| 1609 |
+
768,
|
| 1610 |
+
1,
|
| 1611 |
+
1
|
| 1612 |
+
]
|
| 1613 |
+
},
|
| 1614 |
+
"dec_attns.1.qkv.bias": {
|
| 1615 |
+
"w_norm": 1.005326271057129,
|
| 1616 |
+
"g_norm": 1.735401565383654e-05,
|
| 1617 |
+
"shape": [
|
| 1618 |
+
2304
|
| 1619 |
+
]
|
| 1620 |
+
},
|
| 1621 |
+
"dec_attns.1.proj.weight": {
|
| 1622 |
+
"w_norm": 16.04290771484375,
|
| 1623 |
+
"g_norm": 0.00034307295572943985,
|
| 1624 |
+
"shape": [
|
| 1625 |
+
768,
|
| 1626 |
+
768,
|
| 1627 |
+
1,
|
| 1628 |
+
1
|
| 1629 |
+
]
|
| 1630 |
+
},
|
| 1631 |
+
"dec_attns.1.proj.bias": {
|
| 1632 |
+
"w_norm": 0.5798501372337341,
|
| 1633 |
+
"g_norm": 2.702936762943864e-05,
|
| 1634 |
+
"shape": [
|
| 1635 |
+
768
|
| 1636 |
+
]
|
| 1637 |
+
},
|
| 1638 |
+
"dec_attns.2.norm.weight": {
|
| 1639 |
+
"w_norm": 19.579172134399414,
|
| 1640 |
+
"g_norm": 1.5314006304834038e-05,
|
| 1641 |
+
"shape": [
|
| 1642 |
+
384
|
| 1643 |
+
]
|
| 1644 |
+
},
|
| 1645 |
+
"dec_attns.2.norm.bias": {
|
| 1646 |
+
"w_norm": 0.017062561586499214,
|
| 1647 |
+
"g_norm": 2.1284606191329658e-05,
|
| 1648 |
+
"shape": [
|
| 1649 |
+
384
|
| 1650 |
+
]
|
| 1651 |
+
},
|
| 1652 |
+
"dec_attns.2.qkv.weight": {
|
| 1653 |
+
"w_norm": 19.6417293548584,
|
| 1654 |
+
"g_norm": 0.0004970437148585916,
|
| 1655 |
+
"shape": [
|
| 1656 |
+
1152,
|
| 1657 |
+
384,
|
| 1658 |
+
1,
|
| 1659 |
+
1
|
| 1660 |
+
]
|
| 1661 |
+
},
|
| 1662 |
+
"dec_attns.2.qkv.bias": {
|
| 1663 |
+
"w_norm": 0.9886179566383362,
|
| 1664 |
+
"g_norm": 3.3770891604945064e-05,
|
| 1665 |
+
"shape": [
|
| 1666 |
+
1152
|
| 1667 |
+
]
|
| 1668 |
+
},
|
| 1669 |
+
"dec_attns.2.proj.weight": {
|
| 1670 |
+
"w_norm": 11.308408737182617,
|
| 1671 |
+
"g_norm": 0.0004960380611009896,
|
| 1672 |
+
"shape": [
|
| 1673 |
+
384,
|
| 1674 |
+
384,
|
| 1675 |
+
1,
|
| 1676 |
+
1
|
| 1677 |
+
]
|
| 1678 |
+
},
|
| 1679 |
+
"dec_attns.2.proj.bias": {
|
| 1680 |
+
"w_norm": 0.5879140496253967,
|
| 1681 |
+
"g_norm": 5.74711557419505e-05,
|
| 1682 |
+
"shape": [
|
| 1683 |
+
384
|
| 1684 |
+
]
|
| 1685 |
+
},
|
| 1686 |
+
"dec_ups.0.op.weight": {
|
| 1687 |
+
"w_norm": 16.0877628326416,
|
| 1688 |
+
"g_norm": 0.001645346055738628,
|
| 1689 |
+
"shape": [
|
| 1690 |
+
768,
|
| 1691 |
+
768,
|
| 1692 |
+
3,
|
| 1693 |
+
3
|
| 1694 |
+
]
|
| 1695 |
+
},
|
| 1696 |
+
"dec_ups.0.op.bias": {
|
| 1697 |
+
"w_norm": 0.20079049468040466,
|
| 1698 |
+
"g_norm": 1.3754704013990704e-05,
|
| 1699 |
+
"shape": [
|
| 1700 |
+
768
|
| 1701 |
+
]
|
| 1702 |
+
},
|
| 1703 |
+
"dec_ups.1.op.weight": {
|
| 1704 |
+
"w_norm": 16.08888053894043,
|
| 1705 |
+
"g_norm": 0.003802034305408597,
|
| 1706 |
+
"shape": [
|
| 1707 |
+
768,
|
| 1708 |
+
768,
|
| 1709 |
+
3,
|
| 1710 |
+
3
|
| 1711 |
+
]
|
| 1712 |
+
},
|
| 1713 |
+
"dec_ups.1.op.bias": {
|
| 1714 |
+
"w_norm": 0.18735860288143158,
|
| 1715 |
+
"g_norm": 3.066011777264066e-05,
|
| 1716 |
+
"shape": [
|
| 1717 |
+
768
|
| 1718 |
+
]
|
| 1719 |
+
},
|
| 1720 |
+
"dec_ups.2.op.weight": {
|
| 1721 |
+
"w_norm": 11.317456245422363,
|
| 1722 |
+
"g_norm": 0.008105107583105564,
|
| 1723 |
+
"shape": [
|
| 1724 |
+
384,
|
| 1725 |
+
384,
|
| 1726 |
+
3,
|
| 1727 |
+
3
|
| 1728 |
+
]
|
| 1729 |
+
},
|
| 1730 |
+
"dec_ups.2.op.bias": {
|
| 1731 |
+
"w_norm": 0.19506803154945374,
|
| 1732 |
+
"g_norm": 8.368221460841596e-05,
|
| 1733 |
+
"shape": [
|
| 1734 |
+
384
|
| 1735 |
+
]
|
| 1736 |
+
},
|
| 1737 |
+
"dec_ups.3.op.weight": {
|
| 1738 |
+
"w_norm": 11.315321922302246,
|
| 1739 |
+
"g_norm": 0.01358941849321127,
|
| 1740 |
+
"shape": [
|
| 1741 |
+
384,
|
| 1742 |
+
384,
|
| 1743 |
+
3,
|
| 1744 |
+
3
|
| 1745 |
+
]
|
| 1746 |
+
},
|
| 1747 |
+
"dec_ups.3.op.bias": {
|
| 1748 |
+
"w_norm": 0.1970231831073761,
|
| 1749 |
+
"g_norm": 0.0002463943383190781,
|
| 1750 |
+
"shape": [
|
| 1751 |
+
384
|
| 1752 |
+
]
|
| 1753 |
+
},
|
| 1754 |
+
"dec_ups.4.op.weight": {
|
| 1755 |
+
"w_norm": 7.9966630935668945,
|
| 1756 |
+
"g_norm": 0.04155389592051506,
|
| 1757 |
+
"shape": [
|
| 1758 |
+
192,
|
| 1759 |
+
192,
|
| 1760 |
+
3,
|
| 1761 |
+
3
|
| 1762 |
+
]
|
| 1763 |
+
},
|
| 1764 |
+
"dec_ups.4.op.bias": {
|
| 1765 |
+
"w_norm": 0.19377826154232025,
|
| 1766 |
+
"g_norm": 0.0008116448298096657,
|
| 1767 |
+
"shape": [
|
| 1768 |
+
192
|
| 1769 |
+
]
|
| 1770 |
+
},
|
| 1771 |
+
"out_norm.weight": {
|
| 1772 |
+
"w_norm": 13.845489501953125,
|
| 1773 |
+
"g_norm": 0.008532080799341202,
|
| 1774 |
+
"shape": [
|
| 1775 |
+
192
|
| 1776 |
+
]
|
| 1777 |
+
},
|
| 1778 |
+
"out_norm.bias": {
|
| 1779 |
+
"w_norm": 0.024909643456339836,
|
| 1780 |
+
"g_norm": 0.010218916460871696,
|
| 1781 |
+
"shape": [
|
| 1782 |
+
192
|
| 1783 |
+
]
|
| 1784 |
+
},
|
| 1785 |
+
"out_conv.weight": {
|
| 1786 |
+
"w_norm": 0.09844083338975906,
|
| 1787 |
+
"g_norm": 0.9382269978523254,
|
| 1788 |
+
"shape": [
|
| 1789 |
+
3,
|
| 1790 |
+
192,
|
| 1791 |
+
3,
|
| 1792 |
+
3
|
| 1793 |
+
]
|
| 1794 |
+
},
|
| 1795 |
+
"out_conv.bias": {
|
| 1796 |
+
"w_norm": 0.005248754285275936,
|
| 1797 |
+
"g_norm": 0.058636076748371124,
|
| 1798 |
+
"shape": [
|
| 1799 |
+
3
|
| 1800 |
+
]
|
| 1801 |
+
}
|
| 1802 |
+
},
|
| 1803 |
+
"activations": {
|
| 1804 |
+
"input_conv": {
|
| 1805 |
+
"mean": 0.005270535591989756,
|
| 1806 |
+
"std": 0.4314875900745392,
|
| 1807 |
+
"abs_max": 3.140625,
|
| 1808 |
+
"frac_dead": 0.00039295354508794844,
|
| 1809 |
+
"frac_sat": 0.0
|
| 1810 |
+
},
|
| 1811 |
+
"time_mlp.1": {
|
| 1812 |
+
"mean": 0.020573442801833153,
|
| 1813 |
+
"std": 0.3014470338821411,
|
| 1814 |
+
"abs_max": 2.390625,
|
| 1815 |
+
"frac_dead": 0.0,
|
| 1816 |
+
"frac_sat": 0.0
|
| 1817 |
+
},
|
| 1818 |
+
"time_mlp.3": {
|
| 1819 |
+
"mean": 0.013320415280759335,
|
| 1820 |
+
"std": 0.1021292507648468,
|
| 1821 |
+
"abs_max": 0.46875,
|
| 1822 |
+
"frac_dead": 0.00032552084303461015,
|
| 1823 |
+
"frac_sat": 0.0
|
| 1824 |
+
},
|
| 1825 |
+
"enc_blocks.0.norm1": {
|
| 1826 |
+
"mean": -0.001979447901248932,
|
| 1827 |
+
"std": 0.9984237551689148,
|
| 1828 |
+
"abs_max": 6.403844833374023,
|
| 1829 |
+
"frac_dead": 1.238452114193933e-06,
|
| 1830 |
+
"frac_sat": 0.0
|
| 1831 |
+
},
|
| 1832 |
+
"enc_blocks.0.conv1": {
|
| 1833 |
+
"mean": -0.007452946621924639,
|
| 1834 |
+
"std": 0.5369975566864014,
|
| 1835 |
+
"abs_max": 4.4375,
|
| 1836 |
+
"frac_dead": 6.309151649475098e-05,
|
| 1837 |
+
"frac_sat": 0.0
|
| 1838 |
+
},
|
| 1839 |
+
"enc_blocks.0.t_proj": {
|
| 1840 |
+
"mean": -0.0015568019589409232,
|
| 1841 |
+
"std": 0.05406259372830391,
|
| 1842 |
+
"abs_max": 0.2451171875,
|
| 1843 |
+
"frac_dead": 0.0,
|
| 1844 |
+
"frac_sat": 0.0
|
| 1845 |
+
},
|
| 1846 |
+
"enc_blocks.0.norm2": {
|
| 1847 |
+
"mean": -0.0009606948588043451,
|
| 1848 |
+
"std": 0.9998885989189148,
|
| 1849 |
+
"abs_max": 7.327714920043945,
|
| 1850 |
+
"frac_dead": 6.192260570969665e-07,
|
| 1851 |
+
"frac_sat": 0.0
|
| 1852 |
+
},
|
| 1853 |
+
"enc_blocks.0.conv2": {
|
| 1854 |
+
"mean": -0.0457494780421257,
|
| 1855 |
+
"std": 0.7087180614471436,
|
| 1856 |
+
"abs_max": 8.5,
|
| 1857 |
+
"frac_dead": 4.7124096454354e-05,
|
| 1858 |
+
"frac_sat": 0.0
|
| 1859 |
+
},
|
| 1860 |
+
"enc_downs.0.op": {
|
| 1861 |
+
"mean": 0.1254103183746338,
|
| 1862 |
+
"std": 1.4901121854782104,
|
| 1863 |
+
"abs_max": 17.875,
|
| 1864 |
+
"frac_dead": 2.2225909560802393e-05,
|
| 1865 |
+
"frac_sat": 5.988280099700205e-05
|
| 1866 |
+
},
|
| 1867 |
+
"enc_blocks.1.norm1": {
|
| 1868 |
+
"mean": -0.000360555131919682,
|
| 1869 |
+
"std": 1.0000308752059937,
|
| 1870 |
+
"abs_max": 9.678850173950195,
|
| 1871 |
+
"frac_dead": 9.006924415189133e-07,
|
| 1872 |
+
"frac_sat": 0.0
|
| 1873 |
+
},
|
| 1874 |
+
"enc_blocks.1.conv1": {
|
| 1875 |
+
"mean": 0.07103142887353897,
|
| 1876 |
+
"std": 0.9594767093658447,
|
| 1877 |
+
"abs_max": 11.5625,
|
| 1878 |
+
"frac_dead": 3.520647805999033e-05,
|
| 1879 |
+
"frac_sat": 1.059638137235197e-07
|
| 1880 |
+
},
|
| 1881 |
+
"enc_blocks.1.t_proj": {
|
| 1882 |
+
"mean": 0.0024000913836061954,
|
| 1883 |
+
"std": 0.04997396841645241,
|
| 1884 |
+
"abs_max": 0.1552734375,
|
| 1885 |
+
"frac_dead": 0.0,
|
| 1886 |
+
"frac_sat": 0.0
|
| 1887 |
+
},
|
| 1888 |
+
"enc_blocks.1.norm2": {
|
| 1889 |
+
"mean": -0.0016624480485916138,
|
| 1890 |
+
"std": 0.9981207847595215,
|
| 1891 |
+
"abs_max": 9.288371086120605,
|
| 1892 |
+
"frac_dead": 4.5034622075945663e-07,
|
| 1893 |
+
"frac_sat": 0.0
|
| 1894 |
+
},
|
| 1895 |
+
"enc_blocks.1.conv2": {
|
| 1896 |
+
"mean": -0.0022635753266513348,
|
| 1897 |
+
"std": 0.6630864143371582,
|
| 1898 |
+
"abs_max": 9.875,
|
| 1899 |
+
"frac_dead": 6.64260660414584e-05,
|
| 1900 |
+
"frac_sat": 0.0
|
| 1901 |
+
},
|
| 1902 |
+
"enc_downs.1.op": {
|
| 1903 |
+
"mean": -0.0900968611240387,
|
| 1904 |
+
"std": 2.5945587158203125,
|
| 1905 |
+
"abs_max": 25.5,
|
| 1906 |
+
"frac_dead": 1.4781951904296875e-05,
|
| 1907 |
+
"frac_sat": 0.004522058647125959
|
| 1908 |
+
},
|
| 1909 |
+
"enc_blocks.2.norm1": {
|
| 1910 |
+
"mean": 0.0002139185817213729,
|
| 1911 |
+
"std": 1.0010061264038086,
|
| 1912 |
+
"abs_max": 9.153732299804688,
|
| 1913 |
+
"frac_dead": 5.298190899338806e-07,
|
| 1914 |
+
"frac_sat": 0.0
|
| 1915 |
+
},
|
| 1916 |
+
"enc_blocks.2.conv1": {
|
| 1917 |
+
"mean": -0.14553506672382355,
|
| 1918 |
+
"std": 1.095105767250061,
|
| 1919 |
+
"abs_max": 13.0,
|
| 1920 |
+
"frac_dead": 3.62396240234375e-05,
|
| 1921 |
+
"frac_sat": 2.5166405066556763e-06
|
| 1922 |
+
},
|
| 1923 |
+
"enc_blocks.2.t_proj": {
|
| 1924 |
+
"mean": -0.002175527159124613,
|
| 1925 |
+
"std": 0.04917168989777565,
|
| 1926 |
+
"abs_max": 0.193359375,
|
| 1927 |
+
"frac_dead": 0.0,
|
| 1928 |
+
"frac_sat": 0.0
|
| 1929 |
+
},
|
| 1930 |
+
"enc_blocks.2.norm2": {
|
| 1931 |
+
"mean": -0.0024002897553145885,
|
| 1932 |
+
"std": 0.996701180934906,
|
| 1933 |
+
"abs_max": 11.8389253616333,
|
| 1934 |
+
"frac_dead": 6.092919306865952e-07,
|
| 1935 |
+
"frac_sat": 1.721911985441693e-06
|
| 1936 |
+
},
|
| 1937 |
+
"enc_blocks.2.conv2": {
|
| 1938 |
+
"mean": 0.06730736792087555,
|
| 1939 |
+
"std": 0.6235703229904175,
|
| 1940 |
+
"abs_max": 6.96875,
|
| 1941 |
+
"frac_dead": 3.822644430329092e-05,
|
| 1942 |
+
"frac_sat": 0.0
|
| 1943 |
+
},
|
| 1944 |
+
"enc_blocks.2.skip": {
|
| 1945 |
+
"mean": 0.08201966434717178,
|
| 1946 |
+
"std": 1.3968431949615479,
|
| 1947 |
+
"abs_max": 17.25,
|
| 1948 |
+
"frac_dead": 9.748670709086582e-05,
|
| 1949 |
+
"frac_sat": 7.735358667559922e-05
|
| 1950 |
+
},
|
| 1951 |
+
"enc_downs.2.op": {
|
| 1952 |
+
"mean": 0.2103821188211441,
|
| 1953 |
+
"std": 1.7188763618469238,
|
| 1954 |
+
"abs_max": 16.25,
|
| 1955 |
+
"frac_dead": 1.5788607925060205e-05,
|
| 1956 |
+
"frac_sat": 6.283654511207715e-05
|
| 1957 |
+
},
|
| 1958 |
+
"enc_blocks.3.norm1": {
|
| 1959 |
+
"mean": 0.00011012620234396309,
|
| 1960 |
+
"std": 1.0003811120986938,
|
| 1961 |
+
"abs_max": 9.707320213317871,
|
| 1962 |
+
"frac_dead": 5.298190899338806e-07,
|
| 1963 |
+
"frac_sat": 0.0
|
| 1964 |
+
},
|
| 1965 |
+
"enc_blocks.3.conv1": {
|
| 1966 |
+
"mean": 0.12389453500509262,
|
| 1967 |
+
"std": 1.1216708421707153,
|
| 1968 |
+
"abs_max": 15.4375,
|
| 1969 |
+
"frac_dead": 2.3523965865024365e-05,
|
| 1970 |
+
"frac_sat": 6.357828624459216e-06
|
| 1971 |
+
},
|
| 1972 |
+
"enc_blocks.3.t_proj": {
|
| 1973 |
+
"mean": -0.002851994475349784,
|
| 1974 |
+
"std": 0.04747543856501579,
|
| 1975 |
+
"abs_max": 0.1826171875,
|
| 1976 |
+
"frac_dead": 0.00010850694525288418,
|
| 1977 |
+
"frac_sat": 0.0
|
| 1978 |
+
},
|
| 1979 |
+
"enc_blocks.3.norm2": {
|
| 1980 |
+
"mean": -0.0023243087343871593,
|
| 1981 |
+
"std": 0.9972028732299805,
|
| 1982 |
+
"abs_max": 12.23245620727539,
|
| 1983 |
+
"frac_dead": 6.35782896551973e-07,
|
| 1984 |
+
"frac_sat": 1.4834934063401306e-06
|
| 1985 |
+
},
|
| 1986 |
+
"enc_blocks.3.conv2": {
|
| 1987 |
+
"mean": 0.09541334211826324,
|
| 1988 |
+
"std": 0.7750728726387024,
|
| 1989 |
+
"abs_max": 11.0625,
|
| 1990 |
+
"frac_dead": 4.132588583161123e-05,
|
| 1991 |
+
"frac_sat": 8.477105097881577e-07
|
| 1992 |
+
},
|
| 1993 |
+
"enc_attns.3.norm": {
|
| 1994 |
+
"mean": 3.4679967939155176e-05,
|
| 1995 |
+
"std": 0.999211311340332,
|
| 1996 |
+
"abs_max": 10.729257583618164,
|
| 1997 |
+
"frac_dead": 4.238552548940788e-07,
|
| 1998 |
+
"frac_sat": 4.238552548940788e-07
|
| 1999 |
+
},
|
| 2000 |
+
"enc_attns.3.qkv": {
|
| 2001 |
+
"mean": -0.011044418439269066,
|
| 2002 |
+
"std": 0.6183685660362244,
|
| 2003 |
+
"abs_max": 8.0625,
|
| 2004 |
+
"frac_dead": 0.00011701937182806432,
|
| 2005 |
+
"frac_sat": 0.0
|
| 2006 |
+
},
|
| 2007 |
+
"enc_attns.3.proj": {
|
| 2008 |
+
"mean": -0.0013116929912939668,
|
| 2009 |
+
"std": 0.2898203730583191,
|
| 2010 |
+
"abs_max": 2.421875,
|
| 2011 |
+
"frac_dead": 0.00019571516895666718,
|
| 2012 |
+
"frac_sat": 0.0
|
| 2013 |
+
},
|
| 2014 |
+
"enc_downs.3.op": {
|
| 2015 |
+
"mean": -0.13180769979953766,
|
| 2016 |
+
"std": 1.869809627532959,
|
| 2017 |
+
"abs_max": 18.125,
|
| 2018 |
+
"frac_dead": 9.324815437139478e-06,
|
| 2019 |
+
"frac_sat": 0.00029754638671875
|
| 2020 |
+
},
|
| 2021 |
+
"enc_blocks.4.norm1": {
|
| 2022 |
+
"mean": -0.00013007968664169312,
|
| 2023 |
+
"std": 1.0002660751342773,
|
| 2024 |
+
"abs_max": 8.761725425720215,
|
| 2025 |
+
"frac_dead": 1.271565793103946e-06,
|
| 2026 |
+
"frac_sat": 0.0
|
| 2027 |
+
},
|
| 2028 |
+
"enc_blocks.4.conv1": {
|
| 2029 |
+
"mean": -0.24413247406482697,
|
| 2030 |
+
"std": 0.9627959728240967,
|
| 2031 |
+
"abs_max": 10.0,
|
| 2032 |
+
"frac_dead": 3.221299994038418e-05,
|
| 2033 |
+
"frac_sat": 0.0
|
| 2034 |
+
},
|
| 2035 |
+
"enc_blocks.4.t_proj": {
|
| 2036 |
+
"mean": -0.00039894311339594424,
|
| 2037 |
+
"std": 0.04767155647277832,
|
| 2038 |
+
"abs_max": 0.21875,
|
| 2039 |
+
"frac_dead": 0.00016276042151730508,
|
| 2040 |
+
"frac_sat": 0.0
|
| 2041 |
+
},
|
| 2042 |
+
"enc_blocks.4.norm2": {
|
| 2043 |
+
"mean": -0.002712855814024806,
|
| 2044 |
+
"std": 0.9977284669876099,
|
| 2045 |
+
"abs_max": 9.868551254272461,
|
| 2046 |
+
"frac_dead": 8.477105097881577e-07,
|
| 2047 |
+
"frac_sat": 0.0
|
| 2048 |
+
},
|
| 2049 |
+
"enc_blocks.4.conv2": {
|
| 2050 |
+
"mean": 0.0014793144073337317,
|
| 2051 |
+
"std": 0.7221308946609497,
|
| 2052 |
+
"abs_max": 6.0625,
|
| 2053 |
+
"frac_dead": 2.649095404194668e-05,
|
| 2054 |
+
"frac_sat": 0.0
|
| 2055 |
+
},
|
| 2056 |
+
"enc_blocks.4.skip": {
|
| 2057 |
+
"mean": 0.03583603352308273,
|
| 2058 |
+
"std": 1.0848972797393799,
|
| 2059 |
+
"abs_max": 10.6875,
|
| 2060 |
+
"frac_dead": 8.0108642578125e-05,
|
| 2061 |
+
"frac_sat": 6.35782896551973e-07
|
| 2062 |
+
},
|
| 2063 |
+
"enc_attns.4.norm": {
|
| 2064 |
+
"mean": -4.4359101593727246e-05,
|
| 2065 |
+
"std": 0.9994379878044128,
|
| 2066 |
+
"abs_max": 11.126827239990234,
|
| 2067 |
+
"frac_dead": 1.0596381798677612e-06,
|
| 2068 |
+
"frac_sat": 4.238552548940788e-07
|
| 2069 |
+
},
|
| 2070 |
+
"enc_attns.4.qkv": {
|
| 2071 |
+
"mean": 0.009110452607274055,
|
| 2072 |
+
"std": 0.6766706705093384,
|
| 2073 |
+
"abs_max": 7.71875,
|
| 2074 |
+
"frac_dead": 7.876643212512136e-05,
|
| 2075 |
+
"frac_sat": 0.0
|
| 2076 |
+
},
|
| 2077 |
+
"enc_attns.4.proj": {
|
| 2078 |
+
"mean": -0.011879508383572102,
|
| 2079 |
+
"std": 0.3317599892616272,
|
| 2080 |
+
"abs_max": 2.6875,
|
| 2081 |
+
"frac_dead": 0.00016615126514807343,
|
| 2082 |
+
"frac_sat": 0.0
|
| 2083 |
+
},
|
| 2084 |
+
"enc_downs.4.op": {
|
| 2085 |
+
"mean": 0.04674041271209717,
|
| 2086 |
+
"std": 2.207994222640991,
|
| 2087 |
+
"abs_max": 18.5,
|
| 2088 |
+
"frac_dead": 6.781684078305261e-06,
|
| 2089 |
+
"frac_sat": 0.00032552084303461015
|
| 2090 |
+
},
|
| 2091 |
+
"enc_blocks.5.norm1": {
|
| 2092 |
+
"mean": 6.525673961732537e-05,
|
| 2093 |
+
"std": 1.000154972076416,
|
| 2094 |
+
"abs_max": 6.9245285987854,
|
| 2095 |
+
"frac_dead": 8.477105097881577e-07,
|
| 2096 |
+
"frac_sat": 0.0
|
| 2097 |
+
},
|
| 2098 |
+
"enc_blocks.5.conv1": {
|
| 2099 |
+
"mean": -0.17496569454669952,
|
| 2100 |
+
"std": 1.6681394577026367,
|
| 2101 |
+
"abs_max": 11.9375,
|
| 2102 |
+
"frac_dead": 1.0172526344831567e-05,
|
| 2103 |
+
"frac_sat": 2.0345052689663135e-05
|
| 2104 |
+
},
|
| 2105 |
+
"enc_blocks.5.t_proj": {
|
| 2106 |
+
"mean": -0.0018383245915174484,
|
| 2107 |
+
"std": 0.04849156364798546,
|
| 2108 |
+
"abs_max": 0.2314453125,
|
| 2109 |
+
"frac_dead": 0.0,
|
| 2110 |
+
"frac_sat": 0.0
|
| 2111 |
+
},
|
| 2112 |
+
"enc_blocks.5.norm2": {
|
| 2113 |
+
"mean": -0.0017972373170778155,
|
| 2114 |
+
"std": 0.9986117482185364,
|
| 2115 |
+
"abs_max": 6.892906665802002,
|
| 2116 |
+
"frac_dead": 8.477105097881577e-07,
|
| 2117 |
+
"frac_sat": 0.0
|
| 2118 |
+
},
|
| 2119 |
+
"enc_blocks.5.conv2": {
|
| 2120 |
+
"mean": 0.12745817005634308,
|
| 2121 |
+
"std": 1.3628087043762207,
|
| 2122 |
+
"abs_max": 6.0625,
|
| 2123 |
+
"frac_dead": 1.695421087788418e-05,
|
| 2124 |
+
"frac_sat": 0.0
|
| 2125 |
+
},
|
| 2126 |
+
"enc_attns.5.norm": {
|
| 2127 |
+
"mean": 8.80869702086784e-05,
|
| 2128 |
+
"std": 0.9993440508842468,
|
| 2129 |
+
"abs_max": 6.990722179412842,
|
| 2130 |
+
"frac_dead": 1.6954210195763153e-06,
|
| 2131 |
+
"frac_sat": 0.0
|
| 2132 |
+
},
|
| 2133 |
+
"enc_attns.5.qkv": {
|
| 2134 |
+
"mean": 0.0036823570262640715,
|
| 2135 |
+
"std": 0.6195452809333801,
|
| 2136 |
+
"abs_max": 6.0625,
|
| 2137 |
+
"frac_dead": 8.053249621298164e-05,
|
| 2138 |
+
"frac_sat": 0.0
|
| 2139 |
+
},
|
| 2140 |
+
"enc_attns.5.proj": {
|
| 2141 |
+
"mean": 0.0019204264972358942,
|
| 2142 |
+
"std": 0.3217765986919403,
|
| 2143 |
+
"abs_max": 2.390625,
|
| 2144 |
+
"frac_dead": 0.0001313951361225918,
|
| 2145 |
+
"frac_sat": 0.0
|
| 2146 |
+
},
|
| 2147 |
+
"mid1.norm1": {
|
| 2148 |
+
"mean": 1.6457504443678772e-06,
|
| 2149 |
+
"std": 0.9999829530715942,
|
| 2150 |
+
"abs_max": 6.66753625869751,
|
| 2151 |
+
"frac_dead": 8.477105097881577e-07,
|
| 2152 |
+
"frac_sat": 0.0
|
| 2153 |
+
},
|
| 2154 |
+
"mid1.conv1": {
|
| 2155 |
+
"mean": -0.07720707356929779,
|
| 2156 |
+
"std": 0.9930728673934937,
|
| 2157 |
+
"abs_max": 9.1875,
|
| 2158 |
+
"frac_dead": 1.695421087788418e-05,
|
| 2159 |
+
"frac_sat": 0.0
|
| 2160 |
+
},
|
| 2161 |
+
"mid1.t_proj": {
|
| 2162 |
+
"mean": 0.002672894624993205,
|
| 2163 |
+
"std": 0.04993588104844093,
|
| 2164 |
+
"abs_max": 0.1875,
|
| 2165 |
+
"frac_dead": 0.0,
|
| 2166 |
+
"frac_sat": 0.0
|
| 2167 |
+
},
|
| 2168 |
+
"mid1.norm2": {
|
| 2169 |
+
"mean": -0.0021396661177277565,
|
| 2170 |
+
"std": 0.9983307123184204,
|
| 2171 |
+
"abs_max": 8.67036247253418,
|
| 2172 |
+
"frac_dead": 3.3908420391526306e-06,
|
| 2173 |
+
"frac_sat": 0.0
|
| 2174 |
+
},
|
| 2175 |
+
"mid1.conv2": {
|
| 2176 |
+
"mean": 0.028066406026482582,
|
| 2177 |
+
"std": 0.9323076605796814,
|
| 2178 |
+
"abs_max": 7.0,
|
| 2179 |
+
"frac_dead": 1.52587890625e-05,
|
| 2180 |
+
"frac_sat": 0.0
|
| 2181 |
+
},
|
| 2182 |
+
"mid_attn.norm": {
|
| 2183 |
+
"mean": -7.023000307526672e-06,
|
| 2184 |
+
"std": 0.9994739890098572,
|
| 2185 |
+
"abs_max": 6.992171287536621,
|
| 2186 |
+
"frac_dead": 0.0,
|
| 2187 |
+
"frac_sat": 0.0
|
| 2188 |
+
},
|
| 2189 |
+
"mid_attn.qkv": {
|
| 2190 |
+
"mean": 0.002624101471155882,
|
| 2191 |
+
"std": 0.6168164014816284,
|
| 2192 |
+
"abs_max": 4.84375,
|
| 2193 |
+
"frac_dead": 8.307563257403672e-05,
|
| 2194 |
+
"frac_sat": 0.0
|
| 2195 |
+
},
|
| 2196 |
+
"mid_attn.proj": {
|
| 2197 |
+
"mean": 0.0016497158212587237,
|
| 2198 |
+
"std": 0.28975075483322144,
|
| 2199 |
+
"abs_max": 1.234375,
|
| 2200 |
+
"frac_dead": 0.00014326308155432343,
|
| 2201 |
+
"frac_sat": 0.0
|
| 2202 |
+
},
|
| 2203 |
+
"mid2.norm1": {
|
| 2204 |
+
"mean": -4.525606709648855e-05,
|
| 2205 |
+
"std": 1.0000933408737183,
|
| 2206 |
+
"abs_max": 7.091708660125732,
|
| 2207 |
+
"frac_dead": 1.6954210195763153e-06,
|
| 2208 |
+
"frac_sat": 0.0
|
| 2209 |
+
},
|
| 2210 |
+
"mid2.conv1": {
|
| 2211 |
+
"mean": -0.1754794418811798,
|
| 2212 |
+
"std": 0.8759015202522278,
|
| 2213 |
+
"abs_max": 6.1875,
|
| 2214 |
+
"frac_dead": 2.5431314497836865e-05,
|
| 2215 |
+
"frac_sat": 0.0
|
| 2216 |
+
},
|
| 2217 |
+
"mid2.t_proj": {
|
| 2218 |
+
"mean": -0.00023175828391686082,
|
| 2219 |
+
"std": 0.04861638322472572,
|
| 2220 |
+
"abs_max": 0.2177734375,
|
| 2221 |
+
"frac_dead": 0.0,
|
| 2222 |
+
"frac_sat": 0.0
|
| 2223 |
+
},
|
| 2224 |
+
"mid2.norm2": {
|
| 2225 |
+
"mean": -0.002224498428404331,
|
| 2226 |
+
"std": 0.9979929327964783,
|
| 2227 |
+
"abs_max": 6.991872787475586,
|
| 2228 |
+
"frac_dead": 8.477105097881577e-07,
|
| 2229 |
+
"frac_sat": 0.0
|
| 2230 |
+
},
|
| 2231 |
+
"mid2.conv2": {
|
| 2232 |
+
"mean": 0.04473405331373215,
|
| 2233 |
+
"std": 1.2036309242248535,
|
| 2234 |
+
"abs_max": 8.0,
|
| 2235 |
+
"frac_dead": 1.8649630874278955e-05,
|
| 2236 |
+
"frac_sat": 0.0
|
| 2237 |
+
},
|
| 2238 |
+
"dec_blocks.0.norm1": {
|
| 2239 |
+
"mean": 0.00011099460971308872,
|
| 2240 |
+
"std": 1.00041663646698,
|
| 2241 |
+
"abs_max": 7.442845821380615,
|
| 2242 |
+
"frac_dead": 8.477105097881577e-07,
|
| 2243 |
+
"frac_sat": 0.0
|
| 2244 |
+
},
|
| 2245 |
+
"dec_blocks.0.conv1": {
|
| 2246 |
+
"mean": -0.870387077331543,
|
| 2247 |
+
"std": 1.9311200380325317,
|
| 2248 |
+
"abs_max": 17.25,
|
| 2249 |
+
"frac_dead": 5.086263172415784e-06,
|
| 2250 |
+
"frac_sat": 0.0007129245204851031
|
| 2251 |
+
},
|
| 2252 |
+
"dec_blocks.0.t_proj": {
|
| 2253 |
+
"mean": -0.0026858618948608637,
|
| 2254 |
+
"std": 0.04970140382647514,
|
| 2255 |
+
"abs_max": 0.201171875,
|
| 2256 |
+
"frac_dead": 0.0,
|
| 2257 |
+
"frac_sat": 0.0
|
| 2258 |
+
},
|
| 2259 |
+
"dec_blocks.0.norm2": {
|
| 2260 |
+
"mean": -0.003192740259692073,
|
| 2261 |
+
"std": 0.9973447322845459,
|
| 2262 |
+
"abs_max": 7.285301208496094,
|
| 2263 |
+
"frac_dead": 8.477105097881577e-07,
|
| 2264 |
+
"frac_sat": 0.0
|
| 2265 |
+
},
|
| 2266 |
+
"dec_blocks.0.conv2": {
|
| 2267 |
+
"mean": -0.059047628194093704,
|
| 2268 |
+
"std": 0.7799074649810791,
|
| 2269 |
+
"abs_max": 6.0625,
|
| 2270 |
+
"frac_dead": 1.7801919966586865e-05,
|
| 2271 |
+
"frac_sat": 0.0
|
| 2272 |
+
},
|
| 2273 |
+
"dec_blocks.0.skip": {
|
| 2274 |
+
"mean": -0.04623860865831375,
|
| 2275 |
+
"std": 1.4913451671600342,
|
| 2276 |
+
"abs_max": 9.75,
|
| 2277 |
+
"frac_dead": 2.1192763597355224e-05,
|
| 2278 |
+
"frac_sat": 0.0
|
| 2279 |
+
},
|
| 2280 |
+
"dec_attns.0.norm": {
|
| 2281 |
+
"mean": -5.304647856974043e-05,
|
| 2282 |
+
"std": 0.9991397857666016,
|
| 2283 |
+
"abs_max": 6.559094429016113,
|
| 2284 |
+
"frac_dead": 0.0,
|
| 2285 |
+
"frac_sat": 0.0
|
| 2286 |
+
},
|
| 2287 |
+
"dec_attns.0.qkv": {
|
| 2288 |
+
"mean": -0.005534946918487549,
|
| 2289 |
+
"std": 0.6146882176399231,
|
| 2290 |
+
"abs_max": 4.3125,
|
| 2291 |
+
"frac_dead": 8.533619256922975e-05,
|
| 2292 |
+
"frac_sat": 0.0
|
| 2293 |
+
},
|
| 2294 |
+
"dec_attns.0.proj": {
|
| 2295 |
+
"mean": 0.007717699743807316,
|
| 2296 |
+
"std": 0.37985163927078247,
|
| 2297 |
+
"abs_max": 1.40625,
|
| 2298 |
+
"frac_dead": 0.00016276042151730508,
|
| 2299 |
+
"frac_sat": 0.0
|
| 2300 |
+
},
|
| 2301 |
+
"dec_ups.0.op": {
|
| 2302 |
+
"mean": -0.019651450216770172,
|
| 2303 |
+
"std": 1.9292454719543457,
|
| 2304 |
+
"abs_max": 12.5,
|
| 2305 |
+
"frac_dead": 9.960598617908545e-06,
|
| 2306 |
+
"frac_sat": 1.1444091796875e-05
|
| 2307 |
+
},
|
| 2308 |
+
"dec_blocks.1.norm1": {
|
| 2309 |
+
"mean": -0.00014618535351473838,
|
| 2310 |
+
"std": 0.9997785091400146,
|
| 2311 |
+
"abs_max": 9.752230644226074,
|
| 2312 |
+
"frac_dead": 9.5367431640625e-07,
|
| 2313 |
+
"frac_sat": 0.0
|
| 2314 |
+
},
|
| 2315 |
+
"dec_blocks.1.conv1": {
|
| 2316 |
+
"mean": -0.8840498328208923,
|
| 2317 |
+
"std": 2.2571663856506348,
|
| 2318 |
+
"abs_max": 18.625,
|
| 2319 |
+
"frac_dead": 4.662407718569739e-06,
|
| 2320 |
+
"frac_sat": 0.0005062951240688562
|
| 2321 |
+
},
|
| 2322 |
+
"dec_blocks.1.t_proj": {
|
| 2323 |
+
"mean": -0.0025018462911248207,
|
| 2324 |
+
"std": 0.04777690768241882,
|
| 2325 |
+
"abs_max": 0.1845703125,
|
| 2326 |
+
"frac_dead": 0.0,
|
| 2327 |
+
"frac_sat": 0.0
|
| 2328 |
+
},
|
| 2329 |
+
"dec_blocks.1.norm2": {
|
| 2330 |
+
"mean": -0.002659045159816742,
|
| 2331 |
+
"std": 0.997779905796051,
|
| 2332 |
+
"abs_max": 8.14177417755127,
|
| 2333 |
+
"frac_dead": 1.6954210195763153e-06,
|
| 2334 |
+
"frac_sat": 0.0
|
| 2335 |
+
},
|
| 2336 |
+
"dec_blocks.1.conv2": {
|
| 2337 |
+
"mean": 0.0010152194881811738,
|
| 2338 |
+
"std": 1.294610619544983,
|
| 2339 |
+
"abs_max": 5.3125,
|
| 2340 |
+
"frac_dead": 1.1232164069951978e-05,
|
| 2341 |
+
"frac_sat": 0.0
|
| 2342 |
+
},
|
| 2343 |
+
"dec_blocks.1.skip": {
|
| 2344 |
+
"mean": 0.034126296639442444,
|
| 2345 |
+
"std": 1.1006033420562744,
|
| 2346 |
+
"abs_max": 7.09375,
|
| 2347 |
+
"frac_dead": 3.094143539783545e-05,
|
| 2348 |
+
"frac_sat": 0.0
|
| 2349 |
+
},
|
| 2350 |
+
"dec_attns.1.norm": {
|
| 2351 |
+
"mean": -9.974920249078423e-05,
|
| 2352 |
+
"std": 0.9996920824050903,
|
| 2353 |
+
"abs_max": 6.520663261413574,
|
| 2354 |
+
"frac_dead": 4.238552548940788e-07,
|
| 2355 |
+
"frac_sat": 0.0
|
| 2356 |
+
},
|
| 2357 |
+
"dec_attns.1.qkv": {
|
| 2358 |
+
"mean": 0.013948864303529263,
|
| 2359 |
+
"std": 0.7189094424247742,
|
| 2360 |
+
"abs_max": 5.03125,
|
| 2361 |
+
"frac_dead": 6.364892760757357e-05,
|
| 2362 |
+
"frac_sat": 0.0
|
| 2363 |
+
},
|
| 2364 |
+
"dec_attns.1.proj": {
|
| 2365 |
+
"mean": -0.011062300764024258,
|
| 2366 |
+
"std": 0.35880205035209656,
|
| 2367 |
+
"abs_max": 1.65625,
|
| 2368 |
+
"frac_dead": 0.00010829501843545586,
|
| 2369 |
+
"frac_sat": 0.0
|
| 2370 |
+
},
|
| 2371 |
+
"dec_ups.1.op": {
|
| 2372 |
+
"mean": 0.06802994757890701,
|
| 2373 |
+
"std": 2.4219303131103516,
|
| 2374 |
+
"abs_max": 11.8125,
|
| 2375 |
+
"frac_dead": 7.841322258173022e-06,
|
| 2376 |
+
"frac_sat": 1.4887916222505737e-05
|
| 2377 |
+
},
|
| 2378 |
+
"dec_blocks.2.norm1": {
|
| 2379 |
+
"mean": 0.00033097798586823046,
|
| 2380 |
+
"std": 1.0006484985351562,
|
| 2381 |
+
"abs_max": 12.5630464553833,
|
| 2382 |
+
"frac_dead": 8.477105097881577e-07,
|
| 2383 |
+
"frac_sat": 6.004615897836629e-07
|
| 2384 |
+
},
|
| 2385 |
+
"dec_blocks.2.conv1": {
|
| 2386 |
+
"mean": -0.5608227252960205,
|
| 2387 |
+
"std": 2.558612108230591,
|
| 2388 |
+
"abs_max": 32.25,
|
| 2389 |
+
"frac_dead": 8.47710543894209e-06,
|
| 2390 |
+
"frac_sat": 0.006476614158600569
|
| 2391 |
+
},
|
| 2392 |
+
"dec_blocks.2.t_proj": {
|
| 2393 |
+
"mean": -0.005891525652259588,
|
| 2394 |
+
"std": 0.05158890783786774,
|
| 2395 |
+
"abs_max": 0.2060546875,
|
| 2396 |
+
"frac_dead": 0.0,
|
| 2397 |
+
"frac_sat": 0.0
|
| 2398 |
+
},
|
| 2399 |
+
"dec_blocks.2.norm2": {
|
| 2400 |
+
"mean": -0.003271313151344657,
|
| 2401 |
+
"std": 0.9989966750144958,
|
| 2402 |
+
"abs_max": 12.778143882751465,
|
| 2403 |
+
"frac_dead": 7.417467031700653e-07,
|
| 2404 |
+
"frac_sat": 9.5367431640625e-06
|
| 2405 |
+
},
|
| 2406 |
+
"dec_blocks.2.conv2": {
|
| 2407 |
+
"mean": -0.0038905921392142773,
|
| 2408 |
+
"std": 0.559281587600708,
|
| 2409 |
+
"abs_max": 6.3125,
|
| 2410 |
+
"frac_dead": 3.804100924753584e-05,
|
| 2411 |
+
"frac_sat": 0.0
|
| 2412 |
+
},
|
| 2413 |
+
"dec_blocks.2.skip": {
|
| 2414 |
+
"mean": 0.01863963156938553,
|
| 2415 |
+
"std": 1.6659446954727173,
|
| 2416 |
+
"abs_max": 11.5,
|
| 2417 |
+
"frac_dead": 2.5325351089122705e-05,
|
| 2418 |
+
"frac_sat": 1.0596381798677612e-06
|
| 2419 |
+
},
|
| 2420 |
+
"dec_attns.2.norm": {
|
| 2421 |
+
"mean": -6.945100176380947e-05,
|
| 2422 |
+
"std": 0.9992446899414062,
|
| 2423 |
+
"abs_max": 12.212302207946777,
|
| 2424 |
+
"frac_dead": 4.238552548940788e-07,
|
| 2425 |
+
"frac_sat": 6.35782896551973e-07
|
| 2426 |
+
},
|
| 2427 |
+
"dec_attns.2.qkv": {
|
| 2428 |
+
"mean": 0.013296275399625301,
|
| 2429 |
+
"std": 0.6340527534484863,
|
| 2430 |
+
"abs_max": 6.875,
|
| 2431 |
+
"frac_dead": 9.907616913551465e-05,
|
| 2432 |
+
"frac_sat": 0.0
|
| 2433 |
+
},
|
| 2434 |
+
"dec_attns.2.proj": {
|
| 2435 |
+
"mean": 0.01610799878835678,
|
| 2436 |
+
"std": 0.27409058809280396,
|
| 2437 |
+
"abs_max": 1.265625,
|
| 2438 |
+
"frac_dead": 0.0002217822620877996,
|
| 2439 |
+
"frac_sat": 0.0
|
| 2440 |
+
},
|
| 2441 |
+
"dec_ups.2.op": {
|
| 2442 |
+
"mean": 0.1586993783712387,
|
| 2443 |
+
"std": 1.4212331771850586,
|
| 2444 |
+
"abs_max": 13.0,
|
| 2445 |
+
"frac_dead": 1.5285280824173242e-05,
|
| 2446 |
+
"frac_sat": 9.112887710216455e-06
|
| 2447 |
+
},
|
| 2448 |
+
"dec_blocks.3.norm1": {
|
| 2449 |
+
"mean": 0.0002693136630114168,
|
| 2450 |
+
"std": 1.0004836320877075,
|
| 2451 |
+
"abs_max": 13.641996383666992,
|
| 2452 |
+
"frac_dead": 6.887647714393097e-07,
|
| 2453 |
+
"frac_sat": 2.370940364926355e-06
|
| 2454 |
+
},
|
| 2455 |
+
"dec_blocks.3.conv1": {
|
| 2456 |
+
"mean": -0.7613422274589539,
|
| 2457 |
+
"std": 1.847955346107483,
|
| 2458 |
+
"abs_max": 36.0,
|
| 2459 |
+
"frac_dead": 1.3536877304431982e-05,
|
| 2460 |
+
"frac_sat": 0.0020664269104599953
|
| 2461 |
+
},
|
| 2462 |
+
"dec_blocks.3.t_proj": {
|
| 2463 |
+
"mean": -0.0013175251660868526,
|
| 2464 |
+
"std": 0.048924800008535385,
|
| 2465 |
+
"abs_max": 0.1806640625,
|
| 2466 |
+
"frac_dead": 0.0,
|
| 2467 |
+
"frac_sat": 0.0
|
| 2468 |
+
},
|
| 2469 |
+
"dec_blocks.3.norm2": {
|
| 2470 |
+
"mean": -0.002513655461370945,
|
| 2471 |
+
"std": 0.997054934501648,
|
| 2472 |
+
"abs_max": 15.864608764648438,
|
| 2473 |
+
"frac_dead": 4.76837158203125e-07,
|
| 2474 |
+
"frac_sat": 2.1272235244396143e-05
|
| 2475 |
+
},
|
| 2476 |
+
"dec_blocks.3.conv2": {
|
| 2477 |
+
"mean": 0.00954698771238327,
|
| 2478 |
+
"std": 0.4573476016521454,
|
| 2479 |
+
"abs_max": 3.921875,
|
| 2480 |
+
"frac_dead": 4.7021443606354296e-05,
|
| 2481 |
+
"frac_sat": 0.0
|
| 2482 |
+
},
|
| 2483 |
+
"dec_blocks.3.skip": {
|
| 2484 |
+
"mean": 0.020214399322867393,
|
| 2485 |
+
"std": 0.9816720485687256,
|
| 2486 |
+
"abs_max": 11.0,
|
| 2487 |
+
"frac_dead": 5.141894143889658e-05,
|
| 2488 |
+
"frac_sat": 1.8543667579251633e-07
|
| 2489 |
+
},
|
| 2490 |
+
"dec_ups.3.op": {
|
| 2491 |
+
"mean": 0.07220441848039627,
|
| 2492 |
+
"std": 1.1501033306121826,
|
| 2493 |
+
"abs_max": 13.6875,
|
| 2494 |
+
"frac_dead": 2.4133258193614893e-05,
|
| 2495 |
+
"frac_sat": 2.4636585749249207e-06
|
| 2496 |
+
},
|
| 2497 |
+
"dec_blocks.4.norm1": {
|
| 2498 |
+
"mean": 0.0004646189045161009,
|
| 2499 |
+
"std": 1.0008416175842285,
|
| 2500 |
+
"abs_max": 11.96432876586914,
|
| 2501 |
+
"frac_dead": 8.035589189603343e-07,
|
| 2502 |
+
"frac_sat": 5.298190899338806e-07
|
| 2503 |
+
},
|
| 2504 |
+
"dec_blocks.4.conv1": {
|
| 2505 |
+
"mean": -0.20519103109836578,
|
| 2506 |
+
"std": 1.7457730770111084,
|
| 2507 |
+
"abs_max": 21.375,
|
| 2508 |
+
"frac_dead": 1.2106365829822607e-05,
|
| 2509 |
+
"frac_sat": 0.0002633333206176758
|
| 2510 |
+
},
|
| 2511 |
+
"dec_blocks.4.t_proj": {
|
| 2512 |
+
"mean": -0.001211022725328803,
|
| 2513 |
+
"std": 0.051456209272146225,
|
| 2514 |
+
"abs_max": 0.1806640625,
|
| 2515 |
+
"frac_dead": 0.0,
|
| 2516 |
+
"frac_sat": 0.0
|
| 2517 |
+
},
|
| 2518 |
+
"dec_blocks.4.norm2": {
|
| 2519 |
+
"mean": -0.001078862464055419,
|
| 2520 |
+
"std": 0.9987243413925171,
|
| 2521 |
+
"abs_max": 11.200509071350098,
|
| 2522 |
+
"frac_dead": 1.1788474694185425e-06,
|
| 2523 |
+
"frac_sat": 7.81483151968132e-07
|
| 2524 |
+
},
|
| 2525 |
+
"dec_blocks.4.conv2": {
|
| 2526 |
+
"mean": -0.05559216067194939,
|
| 2527 |
+
"std": 0.6134659051895142,
|
| 2528 |
+
"abs_max": 9.4375,
|
| 2529 |
+
"frac_dead": 7.282363367266953e-05,
|
| 2530 |
+
"frac_sat": 0.0
|
| 2531 |
+
},
|
| 2532 |
+
"dec_blocks.4.skip": {
|
| 2533 |
+
"mean": 0.0002240973262814805,
|
| 2534 |
+
"std": 1.1718063354492188,
|
| 2535 |
+
"abs_max": 14.0,
|
| 2536 |
+
"frac_dead": 6.145900988485664e-05,
|
| 2537 |
+
"frac_sat": 3.4570693969726562e-06
|
| 2538 |
+
},
|
| 2539 |
+
"dec_ups.4.op": {
|
| 2540 |
+
"mean": -0.21435530483722687,
|
| 2541 |
+
"std": 1.741746187210083,
|
| 2542 |
+
"abs_max": 21.875,
|
| 2543 |
+
"frac_dead": 2.5100178390857764e-05,
|
| 2544 |
+
"frac_sat": 0.00037919150781817734
|
| 2545 |
+
},
|
| 2546 |
+
"dec_blocks.5.norm1": {
|
| 2547 |
+
"mean": -0.0008769453270360827,
|
| 2548 |
+
"std": 0.9995813965797424,
|
| 2549 |
+
"abs_max": 13.412727355957031,
|
| 2550 |
+
"frac_dead": 1.6573402490394074e-06,
|
| 2551 |
+
"frac_sat": 2.880891088352655e-06
|
| 2552 |
+
},
|
| 2553 |
+
"dec_blocks.5.conv1": {
|
| 2554 |
+
"mean": -0.09107916802167892,
|
| 2555 |
+
"std": 1.1201961040496826,
|
| 2556 |
+
"abs_max": 14.0625,
|
| 2557 |
+
"frac_dead": 2.0802021026611328e-05,
|
| 2558 |
+
"frac_sat": 4.119343429920264e-06
|
| 2559 |
+
},
|
| 2560 |
+
"dec_blocks.5.t_proj": {
|
| 2561 |
+
"mean": -0.0012433892115950584,
|
| 2562 |
+
"std": 0.05638977140188217,
|
| 2563 |
+
"abs_max": 0.1982421875,
|
| 2564 |
+
"frac_dead": 0.0,
|
| 2565 |
+
"frac_sat": 0.0
|
| 2566 |
+
},
|
| 2567 |
+
"dec_blocks.5.norm2": {
|
| 2568 |
+
"mean": 0.0013370562810450792,
|
| 2569 |
+
"std": 1.0027819871902466,
|
| 2570 |
+
"abs_max": 15.821227073669434,
|
| 2571 |
+
"frac_dead": 6.655852189396683e-07,
|
| 2572 |
+
"frac_sat": 1.7881393432617188e-07
|
| 2573 |
+
},
|
| 2574 |
+
"dec_blocks.5.conv2": {
|
| 2575 |
+
"mean": -0.7922682166099548,
|
| 2576 |
+
"std": 1.2607128620147705,
|
| 2577 |
+
"abs_max": 19.375,
|
| 2578 |
+
"frac_dead": 3.741847103810869e-05,
|
| 2579 |
+
"frac_sat": 7.07341605448164e-05
|
| 2580 |
+
},
|
| 2581 |
+
"dec_blocks.5.skip": {
|
| 2582 |
+
"mean": -0.12430792301893234,
|
| 2583 |
+
"std": 1.1362963914871216,
|
| 2584 |
+
"abs_max": 11.875,
|
| 2585 |
+
"frac_dead": 7.243951404234394e-05,
|
| 2586 |
+
"frac_sat": 2.914004824106087e-07
|
| 2587 |
+
},
|
| 2588 |
+
"out_norm": {
|
| 2589 |
+
"mean": -0.0004818137385882437,
|
| 2590 |
+
"std": 1.0001599788665771,
|
| 2591 |
+
"abs_max": 12.347586631774902,
|
| 2592 |
+
"frac_dead": 1.0463926400916534e-06,
|
| 2593 |
+
"frac_sat": 1.0000335350923706e-06
|
| 2594 |
+
},
|
| 2595 |
+
"out_conv": {
|
| 2596 |
+
"mean": 0.4717104732990265,
|
| 2597 |
+
"std": 0.23990732431411743,
|
| 2598 |
+
"abs_max": 1.234375,
|
| 2599 |
+
"frac_dead": 1.0596381798677612e-06,
|
| 2600 |
+
"frac_sat": 0.0
|
| 2601 |
+
}
|
| 2602 |
+
}
|
| 2603 |
+
}
|
diagnostics/per_t_loss_iter1000.json
ADDED
|
@@ -0,0 +1,42 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"iter": 1000,
|
| 3 |
+
"n_bins": 10,
|
| 4 |
+
"T": 1000,
|
| 5 |
+
"bin_edges": [
|
| 6 |
+
0,
|
| 7 |
+
100,
|
| 8 |
+
200,
|
| 9 |
+
300,
|
| 10 |
+
400,
|
| 11 |
+
500,
|
| 12 |
+
600,
|
| 13 |
+
700,
|
| 14 |
+
800,
|
| 15 |
+
900,
|
| 16 |
+
1000
|
| 17 |
+
],
|
| 18 |
+
"mean_loss_per_bin": [
|
| 19 |
+
0.013784047216176987,
|
| 20 |
+
0.016449427232146263,
|
| 21 |
+
0.0210866816341877,
|
| 22 |
+
0.026539748534560204,
|
| 23 |
+
0.03410493955016136,
|
| 24 |
+
0.046971529722213745,
|
| 25 |
+
0.061805617064237595,
|
| 26 |
+
0.07447069883346558,
|
| 27 |
+
0.08510672301054001,
|
| 28 |
+
0.08638645708560944
|
| 29 |
+
],
|
| 30 |
+
"count_per_bin": [
|
| 31 |
+
1251.0,
|
| 32 |
+
1180.0,
|
| 33 |
+
1207.0,
|
| 34 |
+
1205.0,
|
| 35 |
+
1223.0,
|
| 36 |
+
1179.0,
|
| 37 |
+
1209.0,
|
| 38 |
+
1250.0,
|
| 39 |
+
1178.0,
|
| 40 |
+
1118.0
|
| 41 |
+
]
|
| 42 |
+
}
|
diagnostics/per_t_loss_iter1500.json
ADDED
|
@@ -0,0 +1,42 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"iter": 1500,
|
| 3 |
+
"n_bins": 10,
|
| 4 |
+
"T": 1000,
|
| 5 |
+
"bin_edges": [
|
| 6 |
+
0,
|
| 7 |
+
100,
|
| 8 |
+
200,
|
| 9 |
+
300,
|
| 10 |
+
400,
|
| 11 |
+
500,
|
| 12 |
+
600,
|
| 13 |
+
700,
|
| 14 |
+
800,
|
| 15 |
+
900,
|
| 16 |
+
1000
|
| 17 |
+
],
|
| 18 |
+
"mean_loss_per_bin": [
|
| 19 |
+
0.010882952250540257,
|
| 20 |
+
0.013080555014312267,
|
| 21 |
+
0.01784704625606537,
|
| 22 |
+
0.02446296066045761,
|
| 23 |
+
0.03163571283221245,
|
| 24 |
+
0.039810243993997574,
|
| 25 |
+
0.05174117907881737,
|
| 26 |
+
0.06388191133737564,
|
| 27 |
+
0.07731848955154419,
|
| 28 |
+
0.08887525647878647
|
| 29 |
+
],
|
| 30 |
+
"count_per_bin": [
|
| 31 |
+
1184.0,
|
| 32 |
+
1229.0,
|
| 33 |
+
1261.0,
|
| 34 |
+
1184.0,
|
| 35 |
+
1171.0,
|
| 36 |
+
1203.0,
|
| 37 |
+
1265.0,
|
| 38 |
+
1195.0,
|
| 39 |
+
1178.0,
|
| 40 |
+
1130.0
|
| 41 |
+
]
|
| 42 |
+
}
|
diagnostics/per_t_loss_iter2000.json
ADDED
|
@@ -0,0 +1,42 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"iter": 2000,
|
| 3 |
+
"n_bins": 10,
|
| 4 |
+
"T": 1000,
|
| 5 |
+
"bin_edges": [
|
| 6 |
+
0,
|
| 7 |
+
100,
|
| 8 |
+
200,
|
| 9 |
+
300,
|
| 10 |
+
400,
|
| 11 |
+
500,
|
| 12 |
+
600,
|
| 13 |
+
700,
|
| 14 |
+
800,
|
| 15 |
+
900,
|
| 16 |
+
1000
|
| 17 |
+
],
|
| 18 |
+
"mean_loss_per_bin": [
|
| 19 |
+
0.007996727712452412,
|
| 20 |
+
0.010598840191960335,
|
| 21 |
+
0.01449462678283453,
|
| 22 |
+
0.019310589879751205,
|
| 23 |
+
0.025772448629140854,
|
| 24 |
+
0.032493047416210175,
|
| 25 |
+
0.04158454388380051,
|
| 26 |
+
0.05454123765230179,
|
| 27 |
+
0.07048270851373672,
|
| 28 |
+
0.0811087042093277
|
| 29 |
+
],
|
| 30 |
+
"count_per_bin": [
|
| 31 |
+
1270.0,
|
| 32 |
+
1186.0,
|
| 33 |
+
1204.0,
|
| 34 |
+
1187.0,
|
| 35 |
+
1172.0,
|
| 36 |
+
1162.0,
|
| 37 |
+
1186.0,
|
| 38 |
+
1199.0,
|
| 39 |
+
1196.0,
|
| 40 |
+
1238.0
|
| 41 |
+
]
|
| 42 |
+
}
|
diagnostics/per_t_loss_iter2500.json
ADDED
|
@@ -0,0 +1,42 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"iter": 2500,
|
| 3 |
+
"n_bins": 10,
|
| 4 |
+
"T": 1000,
|
| 5 |
+
"bin_edges": [
|
| 6 |
+
0,
|
| 7 |
+
100,
|
| 8 |
+
200,
|
| 9 |
+
300,
|
| 10 |
+
400,
|
| 11 |
+
500,
|
| 12 |
+
600,
|
| 13 |
+
700,
|
| 14 |
+
800,
|
| 15 |
+
900,
|
| 16 |
+
1000
|
| 17 |
+
],
|
| 18 |
+
"mean_loss_per_bin": [
|
| 19 |
+
0.006924212444573641,
|
| 20 |
+
0.009298368357121944,
|
| 21 |
+
0.013597586192190647,
|
| 22 |
+
0.018246997147798538,
|
| 23 |
+
0.024278221651911736,
|
| 24 |
+
0.03068898618221283,
|
| 25 |
+
0.038973864167928696,
|
| 26 |
+
0.05130283907055855,
|
| 27 |
+
0.06653974950313568,
|
| 28 |
+
0.08158992975950241
|
| 29 |
+
],
|
| 30 |
+
"count_per_bin": [
|
| 31 |
+
1153.0,
|
| 32 |
+
1208.0,
|
| 33 |
+
1214.0,
|
| 34 |
+
1148.0,
|
| 35 |
+
1231.0,
|
| 36 |
+
1202.0,
|
| 37 |
+
1174.0,
|
| 38 |
+
1198.0,
|
| 39 |
+
1268.0,
|
| 40 |
+
1204.0
|
| 41 |
+
]
|
| 42 |
+
}
|
diagnostics/per_t_loss_iter3000.json
ADDED
|
@@ -0,0 +1,42 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"iter": 3000,
|
| 3 |
+
"n_bins": 10,
|
| 4 |
+
"T": 1000,
|
| 5 |
+
"bin_edges": [
|
| 6 |
+
0,
|
| 7 |
+
100,
|
| 8 |
+
200,
|
| 9 |
+
300,
|
| 10 |
+
400,
|
| 11 |
+
500,
|
| 12 |
+
600,
|
| 13 |
+
700,
|
| 14 |
+
800,
|
| 15 |
+
900,
|
| 16 |
+
1000
|
| 17 |
+
],
|
| 18 |
+
"mean_loss_per_bin": [
|
| 19 |
+
0.006272789556533098,
|
| 20 |
+
0.008974757976830006,
|
| 21 |
+
0.012472765520215034,
|
| 22 |
+
0.018289972096681595,
|
| 23 |
+
0.023113682866096497,
|
| 24 |
+
0.028571533039212227,
|
| 25 |
+
0.034495629370212555,
|
| 26 |
+
0.04728328064084053,
|
| 27 |
+
0.062270186841487885,
|
| 28 |
+
0.08114871382713318
|
| 29 |
+
],
|
| 30 |
+
"count_per_bin": [
|
| 31 |
+
1200.0,
|
| 32 |
+
1202.0,
|
| 33 |
+
1193.0,
|
| 34 |
+
1150.0,
|
| 35 |
+
1190.0,
|
| 36 |
+
1193.0,
|
| 37 |
+
1232.0,
|
| 38 |
+
1197.0,
|
| 39 |
+
1157.0,
|
| 40 |
+
1286.0
|
| 41 |
+
]
|
| 42 |
+
}
|
diagnostics/per_t_loss_iter500.json
ADDED
|
@@ -0,0 +1,42 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"iter": 500,
|
| 3 |
+
"n_bins": 10,
|
| 4 |
+
"T": 1000,
|
| 5 |
+
"bin_edges": [
|
| 6 |
+
0,
|
| 7 |
+
100,
|
| 8 |
+
200,
|
| 9 |
+
300,
|
| 10 |
+
400,
|
| 11 |
+
500,
|
| 12 |
+
600,
|
| 13 |
+
700,
|
| 14 |
+
800,
|
| 15 |
+
900,
|
| 16 |
+
1000
|
| 17 |
+
],
|
| 18 |
+
"mean_loss_per_bin": [
|
| 19 |
+
0.026563743129372597,
|
| 20 |
+
0.030675718560814857,
|
| 21 |
+
0.03587967902421951,
|
| 22 |
+
0.04313715547323227,
|
| 23 |
+
0.05277048051357269,
|
| 24 |
+
0.06482600420713425,
|
| 25 |
+
0.07861171662807465,
|
| 26 |
+
0.09098884463310242,
|
| 27 |
+
0.1006721556186676,
|
| 28 |
+
0.09052539616823196
|
| 29 |
+
],
|
| 30 |
+
"count_per_bin": [
|
| 31 |
+
1195.0,
|
| 32 |
+
1271.0,
|
| 33 |
+
1212.0,
|
| 34 |
+
1203.0,
|
| 35 |
+
1128.0,
|
| 36 |
+
1191.0,
|
| 37 |
+
1200.0,
|
| 38 |
+
1204.0,
|
| 39 |
+
1207.0,
|
| 40 |
+
1189.0
|
| 41 |
+
]
|
| 42 |
+
}
|
diagnostics/samples_iter1000.png
ADDED
|
Git LFS Details
|
diagnostics/samples_iter1500.png
ADDED
|
Git LFS Details
|
diagnostics/samples_iter2000.png
ADDED
|
Git LFS Details
|
diagnostics/samples_iter2500.png
ADDED
|
Git LFS Details
|
diagnostics/samples_iter3000.png
ADDED
|
Git LFS Details
|
diagnostics/samples_iter500.png
ADDED
|
Git LFS Details
|
diagnostics/scalars.csv
ADDED
|
@@ -0,0 +1,152 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
iter,loss,L_he,L_ihc,frac_he,grad_norm_he_in,grad_norm_ihc_in,grad_norm_ratio,weight_norm_he_in,weight_norm_ihc_in,global_grad_norm,global_weight_norm,wall_min
|
| 2 |
+
1,0.324302,0.000000,0.324302,0.000,0.000000e+00,0.000000e+00,0.0000,5.583580e+00,5.638622e+00,7.548024e+00,1.940006e+02,0.12
|
| 3 |
+
20,0.089554,0.081879,0.244727,0.632,2.766188e-02,0.000000e+00,0.0000,5.584023e+00,5.640231e+00,3.419123e+00,1.940468e+02,0.58
|
| 4 |
+
40,0.066283,0.071500,0.145946,0.564,4.849450e-02,0.000000e+00,0.0000,5.581692e+00,5.648373e+00,1.805875e+00,1.940822e+02,1.05
|
| 5 |
+
60,0.064310,0.069549,0.111944,0.576,9.486559e-03,1.092996e-02,1.1522,5.578803e+00,5.653113e+00,1.594219e+00,1.941521e+02,1.53
|
| 6 |
+
80,0.058560,0.063121,0.099887,0.557,1.695267e-01,0.000000e+00,0.0000,5.574450e+00,5.660866e+00,2.832275e+00,1.942087e+02,2.03
|
| 7 |
+
100,0.043342,0.065942,0.081857,0.535,6.481587e-02,0.000000e+00,0.0000,5.571587e+00,5.664681e+00,1.141574e+00,1.942350e+02,2.51
|
| 8 |
+
120,0.061614,0.083823,0.066899,0.400,1.632406e-02,2.006876e-02,1.2294,5.569613e+00,5.667080e+00,6.737235e-01,1.942438e+02,2.99
|
| 9 |
+
140,0.048947,0.062868,0.060563,0.525,3.446998e-02,0.000000e+00,0.0000,5.569869e+00,5.668146e+00,1.040985e+00,1.942578e+02,3.49
|
| 10 |
+
160,0.049611,0.063149,0.056116,0.550,5.976567e-02,2.827867e-02,0.4732,5.570674e+00,5.666029e+00,5.637211e-01,1.942699e+02,3.98
|
| 11 |
+
180,0.102211,0.058272,0.064615,0.575,8.403253e-02,9.264228e-02,1.1025,5.570328e+00,5.662053e+00,1.827627e+00,1.942747e+02,4.52
|
| 12 |
+
200,0.097045,0.054950,0.072664,0.580,1.209880e-01,7.290323e-02,0.6026,5.570700e+00,5.662363e+00,3.526256e+00,1.943031e+02,5.02
|
| 13 |
+
220,0.033361,0.054821,0.065552,0.550,1.923540e-02,1.956470e-02,1.0171,5.570779e+00,5.661845e+00,1.287460e+00,1.943287e+02,5.52
|
| 14 |
+
240,0.080996,0.054572,0.054693,0.400,4.239826e-02,5.615345e-02,1.3244,5.571740e+00,5.661584e+00,1.122531e+00,1.943435e+02,6.01
|
| 15 |
+
260,0.063296,0.052931,0.051666,0.483,9.337264e-02,0.000000e+00,0.0000,5.571977e+00,5.661757e+00,2.647556e+00,1.943480e+02,6.51
|
| 16 |
+
280,0.044879,0.058842,0.046476,0.487,9.683609e-02,1.059916e-01,1.0945,5.570150e+00,5.663661e+00,1.611818e+00,1.943532e+02,7.01
|
| 17 |
+
300,0.026623,0.055695,0.043483,0.510,5.903780e-02,5.889061e-02,0.9975,5.569783e+00,5.663438e+00,9.914533e-01,1.943600e+02,7.51
|
| 18 |
+
320,0.047551,0.051881,0.050056,0.500,1.334103e-01,7.245490e-02,0.5431,5.569125e+00,5.663356e+00,1.822765e+00,1.943776e+02,8.00
|
| 19 |
+
340,0.037491,0.053044,0.043878,0.350,1.224490e-01,4.781137e-02,0.3905,5.569310e+00,5.663685e+00,6.291920e-01,1.943802e+02,8.53
|
| 20 |
+
360,0.045606,0.050659,0.055919,0.417,9.152754e-02,0.000000e+00,0.0000,5.568435e+00,5.663766e+00,1.633592e+00,1.943848e+02,9.03
|
| 21 |
+
380,0.059510,0.055742,0.055933,0.400,9.560531e-02,8.094897e-02,0.8467,5.568558e+00,5.663101e+00,1.242207e+00,1.943912e+02,9.53
|
| 22 |
+
400,0.051844,0.051799,0.047156,0.400,2.020893e-01,1.885759e-01,0.9331,5.568215e+00,5.662859e+00,2.524159e+00,1.944012e+02,10.03
|
| 23 |
+
420,0.071549,0.052409,0.042461,0.650,5.727115e-02,0.000000e+00,0.0000,5.566864e+00,5.663506e+00,1.425454e+00,1.944113e+02,10.53
|
| 24 |
+
440,0.034839,0.050828,0.045298,0.500,9.655500e-02,5.825986e-02,0.6034,5.565866e+00,5.664511e+00,9.088429e-01,1.944141e+02,11.02
|
| 25 |
+
460,0.045731,0.051757,0.047090,0.500,1.329883e-01,3.960652e-02,0.2978,5.565227e+00,5.666553e+00,8.658687e-01,1.944144e+02,11.51
|
| 26 |
+
480,0.048368,0.060934,0.054261,0.475,5.729677e-02,0.000000e+00,0.0000,5.562338e+00,5.668830e+00,1.320445e+00,1.944179e+02,12.02
|
| 27 |
+
500,0.047274,0.052398,0.053334,0.520,5.300744e-02,5.737199e-02,1.0823,5.562580e+00,5.667985e+00,9.767188e-01,1.944252e+02,12.55
|
| 28 |
+
520,0.072722,0.052426,0.047749,0.600,1.831990e-01,0.000000e+00,0.0000,5.561999e+00,5.668086e+00,3.654111e+00,1.944359e+02,13.28
|
| 29 |
+
540,0.055554,0.055720,0.053440,0.525,3.209858e-02,6.382165e-02,1.9883,5.561582e+00,5.665567e+00,3.070851e-01,1.944423e+02,13.78
|
| 30 |
+
560,0.034488,0.050804,0.044285,0.533,1.854259e-01,1.345614e-01,0.7257,5.561594e+00,5.665299e+00,1.196357e+00,1.944471e+02,14.28
|
| 31 |
+
580,0.039879,0.047124,0.040886,0.550,5.992791e-02,0.000000e+00,0.0000,5.561020e+00,5.666300e+00,8.843369e-01,1.944544e+02,14.78
|
| 32 |
+
600,0.039756,0.045176,0.039605,0.530,5.762917e-02,0.000000e+00,0.0000,5.559803e+00,5.666136e+00,9.891162e-01,1.944563e+02,15.27
|
| 33 |
+
620,0.030025,0.051738,0.043717,0.550,5.840681e-02,4.439857e-02,0.7602,5.558890e+00,5.667346e+00,5.061632e-01,1.944599e+02,15.77
|
| 34 |
+
640,0.027296,0.047171,0.037159,0.500,8.537418e-02,3.776766e-02,0.4424,5.559151e+00,5.667233e+00,7.796339e-01,1.944667e+02,16.26
|
| 35 |
+
660,0.047204,0.049115,0.036882,0.517,9.682799e-02,0.000000e+00,0.0000,5.558652e+00,5.667965e+00,7.320146e-01,1.944746e+02,16.82
|
| 36 |
+
680,0.023718,0.055043,0.032491,0.512,1.610207e-01,7.466858e-02,0.4637,5.556867e+00,5.669338e+00,9.509891e-01,1.944761e+02,17.35
|
| 37 |
+
700,0.040362,0.047732,0.035761,0.510,8.112706e-02,0.000000e+00,0.0000,5.556290e+00,5.671142e+00,1.107794e+00,1.944794e+02,17.89
|
| 38 |
+
720,0.047241,0.045719,0.036350,0.400,1.615395e-02,0.000000e+00,0.0000,5.555193e+00,5.672166e+00,6.203774e-01,1.944808e+02,18.42
|
| 39 |
+
740,0.053635,0.053625,0.036141,0.475,5.036171e-02,3.867149e-02,0.7679,5.554117e+00,5.672397e+00,4.827808e-01,1.944874e+02,18.96
|
| 40 |
+
760,0.029041,0.052613,0.033061,0.433,1.281855e-01,1.245965e-01,0.9720,5.553406e+00,5.673182e+00,9.581065e-01,1.944897e+02,19.50
|
| 41 |
+
780,0.069766,0.059018,0.034347,0.463,1.865866e-02,0.000000e+00,0.0000,5.553125e+00,5.673737e+00,4.002537e-01,1.944946e+02,20.04
|
| 42 |
+
800,0.049542,0.056743,0.041526,0.480,1.448918e-01,0.000000e+00,0.0000,5.552910e+00,5.674387e+00,1.459499e+00,1.945031e+02,20.55
|
| 43 |
+
820,0.057185,0.062902,0.054434,0.550,4.916273e-02,5.786880e-02,1.1771,5.552092e+00,5.672155e+00,1.628715e+00,1.944977e+02,21.08
|
| 44 |
+
840,0.032963,0.052129,0.047537,0.575,1.070931e-01,9.371609e-02,0.8751,5.553123e+00,5.670137e+00,1.139313e+00,1.945024e+02,21.57
|
| 45 |
+
860,0.053212,0.050836,0.042579,0.533,1.508684e-02,0.000000e+00,0.0000,5.552309e+00,5.671721e+00,3.820408e-01,1.945045e+02,22.08
|
| 46 |
+
880,0.036197,0.044644,0.035957,0.500,3.739530e-02,4.333575e-02,1.1589,5.552338e+00,5.671531e+00,3.969577e-01,1.945076e+02,22.57
|
| 47 |
+
900,0.030575,0.045782,0.032057,0.500,1.993808e-01,1.166504e-01,0.5851,5.550829e+00,5.674816e+00,1.744140e+00,1.945163e+02,23.07
|
| 48 |
+
920,0.048108,0.048038,0.035179,0.400,1.735038e-01,9.684280e-02,0.5582,5.550590e+00,5.675790e+00,1.438468e+00,1.945157e+02,23.57
|
| 49 |
+
940,0.054144,0.049635,0.029078,0.450,1.099340e-01,0.000000e+00,0.0000,5.550489e+00,5.676662e+00,1.127040e+00,1.945251e+02,24.07
|
| 50 |
+
960,0.048444,0.050640,0.036340,0.467,1.843586e-01,1.655948e-01,0.8982,5.549499e+00,5.677701e+00,1.449629e+00,1.945288e+02,24.57
|
| 51 |
+
980,0.122711,0.088276,0.049724,0.475,2.262226e-01,0.000000e+00,0.0000,5.549997e+00,5.676556e+00,5.142752e+00,1.945322e+02,25.10
|
| 52 |
+
1000,0.042467,0.061811,0.046647,0.480,6.969271e-02,0.000000e+00,0.0000,5.550200e+00,5.674387e+00,1.414025e+00,1.945425e+02,25.60
|
| 53 |
+
1020,0.039346,0.050864,0.041000,0.650,3.308066e-02,0.000000e+00,0.0000,5.550016e+00,5.674968e+00,5.275340e-01,1.945527e+02,26.44
|
| 54 |
+
1040,0.054894,0.046731,0.036354,0.575,2.280999e-01,0.000000e+00,0.0000,5.548873e+00,5.677880e+00,2.509141e+00,1.945561e+02,26.96
|
| 55 |
+
1060,0.035178,0.050622,0.032894,0.550,6.499536e-02,0.000000e+00,0.0000,5.548066e+00,5.678804e+00,8.733478e-01,1.945620e+02,27.49
|
| 56 |
+
1080,0.021282,0.044066,0.034804,0.575,2.265670e-02,4.638132e-02,2.0471,5.547132e+00,5.681318e+00,5.455552e-01,1.945771e+02,28.01
|
| 57 |
+
1100,0.034238,0.041705,0.038459,0.540,1.906997e-02,0.000000e+00,0.0000,5.546820e+00,5.680777e+00,3.974923e-01,1.945912e+02,28.54
|
| 58 |
+
1120,0.012649,0.040392,0.030906,0.500,2.311216e-02,5.687568e-02,2.4609,5.546716e+00,5.681136e+00,2.906373e-01,1.945973e+02,29.07
|
| 59 |
+
1140,0.027933,0.040245,0.032691,0.575,9.464227e-02,9.191974e-02,0.9712,5.545743e+00,5.683726e+00,8.857016e-01,1.946002e+02,29.61
|
| 60 |
+
1160,0.027641,0.047474,0.035061,0.567,6.040934e-02,4.746462e-02,0.7857,5.546196e+00,5.680885e+00,6.108094e-01,1.946039e+02,30.14
|
| 61 |
+
1180,0.042171,0.048496,0.035034,0.600,4.149661e-02,0.000000e+00,0.0000,5.546381e+00,5.680843e+00,5.676679e-01,1.946015e+02,30.72
|
| 62 |
+
1200,0.018491,0.049659,0.042841,0.550,4.627964e-02,4.814796e-02,1.0404,5.546086e+00,5.680806e+00,5.894015e-01,1.946019e+02,31.28
|
| 63 |
+
1220,0.033199,0.044426,0.044717,0.700,2.022071e-02,0.000000e+00,0.0000,5.544712e+00,5.683592e+00,5.516331e-01,1.946058e+02,31.82
|
| 64 |
+
1240,0.057997,0.045481,0.037749,0.575,1.568177e-02,0.000000e+00,0.0000,5.544625e+00,5.683716e+00,4.256030e-01,1.946113e+02,32.34
|
| 65 |
+
1260,0.043509,0.044120,0.037564,0.583,1.083245e-01,0.000000e+00,0.0000,5.543569e+00,5.685157e+00,1.200596e+00,1.946112e+02,32.86
|
| 66 |
+
1280,0.050910,0.044558,0.040452,0.600,4.556788e-02,0.000000e+00,0.0000,5.543593e+00,5.683908e+00,4.236941e-01,1.946181e+02,33.37
|
| 67 |
+
1300,0.021324,0.046538,0.038818,0.570,7.591476e-02,3.317392e-02,0.4370,5.543845e+00,5.684491e+00,9.917816e-01,1.946209e+02,33.92
|
| 68 |
+
1320,0.013786,0.042765,0.035769,0.600,5.248239e-02,5.492505e-02,1.0465,5.543582e+00,5.685115e+00,4.774507e-01,1.946365e+02,34.42
|
| 69 |
+
1340,0.046074,0.044198,0.032406,0.600,8.377425e-02,0.000000e+00,0.0000,5.542824e+00,5.686985e+00,7.204953e-01,1.946430e+02,34.92
|
| 70 |
+
1360,0.052405,0.042716,0.027410,0.533,2.257293e-01,0.000000e+00,0.0000,5.542475e+00,5.687851e+00,1.633654e+00,1.946517e+02,35.42
|
| 71 |
+
1380,0.037036,0.041430,0.027357,0.537,3.882533e-02,0.000000e+00,0.0000,5.541284e+00,5.690038e+00,5.205108e-01,1.946570e+02,35.92
|
| 72 |
+
1400,0.049506,0.043883,0.039027,0.540,1.066708e-01,0.000000e+00,0.0000,5.541039e+00,5.690053e+00,1.282204e+00,1.946588e+02,36.42
|
| 73 |
+
1420,0.032331,0.042209,0.040357,0.400,1.077098e-01,7.954398e-02,0.7385,5.540939e+00,5.689275e+00,9.635142e-01,1.946600e+02,36.94
|
| 74 |
+
1440,0.067204,0.053284,0.042821,0.425,1.060799e-01,0.000000e+00,0.0000,5.540968e+00,5.689755e+00,1.597346e+00,1.946660e+02,37.43
|
| 75 |
+
1460,0.046470,0.045193,0.036346,0.500,6.375637e-02,0.000000e+00,0.0000,5.540559e+00,5.689218e+00,6.803362e-01,1.946665e+02,37.96
|
| 76 |
+
1480,0.027071,0.038279,0.038793,0.550,9.144689e-03,0.000000e+00,0.0000,5.540367e+00,5.690754e+00,5.281061e-01,1.946700e+02,38.46
|
| 77 |
+
1500,0.034629,0.041203,0.039800,0.580,7.319029e-02,0.000000e+00,0.0000,5.539738e+00,5.690425e+00,1.133185e+00,1.946750e+02,38.96
|
| 78 |
+
1520,0.067367,0.045813,0.033882,0.450,2.660393e-01,0.000000e+00,0.0000,5.540263e+00,5.690777e+00,1.933010e+00,1.946752e+02,39.70
|
| 79 |
+
1540,0.041980,0.042484,0.035466,0.425,4.374797e-02,5.043596e-02,1.1529,5.539608e+00,5.690997e+00,4.208666e-01,1.946810e+02,40.20
|
| 80 |
+
1560,0.033896,0.041791,0.031637,0.450,1.128763e-01,1.015497e-01,0.8997,5.538445e+00,5.693890e+00,7.631294e-01,1.946858e+02,40.73
|
| 81 |
+
1580,0.041068,0.040949,0.027690,0.475,1.045281e-01,0.000000e+00,0.0000,5.538965e+00,5.693044e+00,6.166756e-01,1.946885e+02,41.23
|
| 82 |
+
1600,0.027662,0.038047,0.027223,0.490,9.405930e-02,0.000000e+00,0.0000,5.538282e+00,5.694565e+00,4.379833e-01,1.946942e+02,41.74
|
| 83 |
+
1620,0.068109,0.039601,0.029563,0.450,3.191025e-01,0.000000e+00,0.0000,5.538123e+00,5.694763e+00,2.038394e+00,1.946973e+02,42.24
|
| 84 |
+
1640,0.048585,0.048424,0.027421,0.475,1.145634e-01,0.000000e+00,0.0000,5.537518e+00,5.695827e+00,1.244316e+00,1.946989e+02,42.77
|
| 85 |
+
1660,0.026821,0.042670,0.027332,0.400,4.244092e-02,2.434769e-02,0.5737,5.537140e+00,5.696678e+00,2.106124e-01,1.947040e+02,43.27
|
| 86 |
+
1680,0.027608,0.040122,0.025330,0.438,6.940031e-02,3.320777e-02,0.4785,5.537290e+00,5.696623e+00,5.453376e-01,1.947043e+02,43.76
|
| 87 |
+
1700,0.040613,0.049979,0.026589,0.460,3.250907e-02,0.000000e+00,0.0000,5.537097e+00,5.696033e+00,3.000476e-01,1.947068e+02,44.25
|
| 88 |
+
1720,0.040012,0.046941,0.032164,0.500,5.535909e-02,0.000000e+00,0.0000,5.537088e+00,5.696715e+00,8.963549e-01,1.947103e+02,44.73
|
| 89 |
+
1740,0.027709,0.039612,0.029663,0.575,1.906851e-02,0.000000e+00,0.0000,5.537459e+00,5.695901e+00,1.864649e-01,1.947145e+02,45.21
|
| 90 |
+
1760,0.025691,0.040833,0.027968,0.583,5.865173e-02,0.000000e+00,0.0000,5.536853e+00,5.697044e+00,6.697341e-01,1.947152e+02,45.69
|
| 91 |
+
1780,0.057605,0.041501,0.031269,0.562,2.818498e-01,0.000000e+00,0.0000,5.535778e+00,5.698503e+00,1.885291e+00,1.947235e+02,46.17
|
| 92 |
+
1800,0.017908,0.042587,0.031309,0.550,2.288410e-02,2.754798e-02,1.2038,5.535971e+00,5.698851e+00,2.667125e-01,1.947271e+02,46.67
|
| 93 |
+
1820,0.087921,0.048942,0.037609,0.400,5.441241e-01,0.000000e+00,0.0000,5.535568e+00,5.699698e+00,3.289169e+00,1.947329e+02,47.18
|
| 94 |
+
1840,0.026403,0.043830,0.047176,0.400,6.284043e-02,3.013557e-02,0.4796,5.534711e+00,5.699631e+00,1.475459e+00,1.947396e+02,47.69
|
| 95 |
+
1860,0.028723,0.042275,0.037263,0.450,4.909779e-02,4.314172e-02,0.8787,5.535226e+00,5.698647e+00,3.969506e-01,1.947814e+02,48.19
|
| 96 |
+
1880,0.062291,0.042702,0.035530,0.512,1.254650e-01,9.826966e-02,0.7832,5.535111e+00,5.699285e+00,8.574362e-01,1.948803e+02,48.71
|
| 97 |
+
1900,0.027759,0.039529,0.030349,0.490,4.966358e-02,0.000000e+00,0.0000,5.535405e+00,5.697826e+00,4.551176e-01,1.949534e+02,49.21
|
| 98 |
+
1920,0.023615,0.036719,0.025705,0.550,3.259157e-02,0.000000e+00,0.0000,5.535572e+00,5.697465e+00,4.351305e-01,1.949911e+02,49.73
|
| 99 |
+
1940,0.015398,0.037778,0.022409,0.525,5.442769e-02,1.150551e-02,0.2114,5.535234e+00,5.698595e+00,3.670271e-01,1.950269e+02,50.26
|
| 100 |
+
1960,0.022194,0.032503,0.024996,0.567,6.702087e-02,4.902003e-02,0.7314,5.535077e+00,5.700035e+00,3.598176e-01,1.950639e+02,50.79
|
| 101 |
+
1980,0.021655,0.035160,0.026033,0.562,3.060283e-02,2.383552e-02,0.7789,5.534276e+00,5.701390e+00,2.679132e-01,1.950762e+02,51.30
|
| 102 |
+
2000,0.012190,0.033719,0.022770,0.560,5.516376e-02,3.470086e-02,0.6291,5.533987e+00,5.701233e+00,4.319662e-01,1.950824e+02,51.82
|
| 103 |
+
2020,0.053668,0.039132,0.025279,0.450,1.672840e-01,0.000000e+00,0.0000,5.533281e+00,5.702682e+00,1.036174e+00,1.950837e+02,52.65
|
| 104 |
+
2040,0.039757,0.044809,0.025929,0.575,1.191412e-01,0.000000e+00,0.0000,5.532632e+00,5.703778e+00,6.945384e-01,1.950842e+02,53.16
|
| 105 |
+
2060,0.032820,0.042971,0.026034,0.550,1.974268e-01,1.347190e-01,0.6824,5.532865e+00,5.702568e+00,1.467161e+00,1.950864e+02,53.65
|
| 106 |
+
2080,0.066178,0.048310,0.029676,0.550,2.937919e-01,2.532831e-01,0.8621,5.532256e+00,5.704848e+00,3.305574e+00,1.950891e+02,54.15
|
| 107 |
+
2100,0.036526,0.045941,0.034931,0.530,4.971378e-02,0.000000e+00,0.0000,5.532545e+00,5.703135e+00,7.438536e-01,1.950904e+02,54.65
|
| 108 |
+
2120,0.033855,0.045313,0.027949,0.350,5.133863e-02,2.395115e-02,0.4665,5.532421e+00,5.705148e+00,3.093131e-01,1.950963e+02,55.18
|
| 109 |
+
2140,0.029976,0.038979,0.027678,0.475,1.435284e-01,0.000000e+00,0.0000,5.532067e+00,5.706230e+00,9.761593e-01,1.951007e+02,55.67
|
| 110 |
+
2160,0.023054,0.040270,0.027492,0.500,2.062072e-01,1.556067e-01,0.7546,5.531598e+00,5.706337e+00,1.503422e+00,1.951032e+02,56.17
|
| 111 |
+
2180,0.034572,0.037756,0.031363,0.500,1.418750e-01,0.000000e+00,0.0000,5.531488e+00,5.706691e+00,9.728462e-01,1.951085e+02,56.66
|
| 112 |
+
2200,0.019214,0.040566,0.027344,0.520,3.052374e-02,3.194327e-02,1.0465,5.531553e+00,5.706663e+00,3.734846e-01,1.951109e+02,57.16
|
| 113 |
+
2220,0.052481,0.041808,0.026860,0.350,2.503049e-01,0.000000e+00,0.0000,5.531326e+00,5.706203e+00,1.548072e+00,1.951292e+02,57.67
|
| 114 |
+
2240,0.026918,0.039601,0.026128,0.425,6.144042e-02,0.000000e+00,0.0000,5.530539e+00,5.708461e+00,5.563463e-01,1.951498e+02,58.17
|
| 115 |
+
2260,0.030097,0.037250,0.028137,0.433,7.947051e-02,9.466857e-02,1.1912,5.530687e+00,5.709204e+00,5.590177e-01,1.951719e+02,58.67
|
| 116 |
+
2280,0.024773,0.039931,0.033231,0.412,1.253409e-01,1.445064e-01,1.1529,5.530683e+00,5.709189e+00,1.508361e+00,1.951791e+02,59.20
|
| 117 |
+
2300,0.039643,0.038911,0.028586,0.440,4.422435e-02,0.000000e+00,0.0000,5.529804e+00,5.710015e+00,6.169448e-01,1.951959e+02,59.71
|
| 118 |
+
2320,0.043022,0.040254,0.025798,0.550,2.210120e-02,0.000000e+00,0.0000,5.529785e+00,5.710114e+00,2.675370e-01,1.952071e+02,60.22
|
| 119 |
+
2340,0.033206,0.035401,0.028339,0.500,4.200990e-02,0.000000e+00,0.0000,5.529185e+00,5.711895e+00,4.950977e-01,1.952226e+02,60.72
|
| 120 |
+
2360,0.051155,0.039664,0.024521,0.483,1.718423e-01,0.000000e+00,0.0000,5.528528e+00,5.713365e+00,9.128702e-01,1.952298e+02,61.21
|
| 121 |
+
2380,0.051099,0.039818,0.025438,0.450,1.166677e-01,0.000000e+00,0.0000,5.528824e+00,5.713803e+00,6.380069e-01,1.952337e+02,61.72
|
| 122 |
+
2400,0.023114,0.035536,0.025342,0.480,8.651908e-02,8.112321e-02,0.9376,5.528874e+00,5.712612e+00,5.602957e-01,1.952408e+02,62.27
|
| 123 |
+
2420,0.027762,0.039908,0.025564,0.800,6.123512e-02,0.000000e+00,0.0000,5.528845e+00,5.713005e+00,6.065254e-01,1.952467e+02,62.77
|
| 124 |
+
2440,0.024739,0.047856,0.029175,0.600,2.923617e-02,3.074437e-02,1.0516,5.528222e+00,5.713144e+00,5.559472e-01,1.952526e+02,63.29
|
| 125 |
+
2460,0.014144,0.046433,0.026852,0.483,3.800712e-02,3.542037e-02,0.9319,5.527998e+00,5.714625e+00,2.053619e-01,1.952570e+02,63.80
|
| 126 |
+
2480,0.042576,0.038321,0.025243,0.537,1.445996e-01,4.728479e-02,0.3270,5.527900e+00,5.714583e+00,6.218336e-01,1.952598e+02,64.30
|
| 127 |
+
2500,0.044968,0.038114,0.025877,0.540,6.437147e-02,0.000000e+00,0.0000,5.527482e+00,5.715956e+00,6.068603e-01,1.952577e+02,64.81
|
| 128 |
+
2520,0.022765,0.033883,0.025136,0.500,3.764430e-02,1.300157e-02,0.3454,5.527965e+00,5.715024e+00,1.998836e-01,1.952606e+02,65.54
|
| 129 |
+
2540,0.035248,0.036961,0.027320,0.525,2.202167e-01,0.000000e+00,0.0000,5.527516e+00,5.716482e+00,1.172769e+00,1.952620e+02,66.03
|
| 130 |
+
2560,0.026551,0.040164,0.024997,0.517,5.409054e-02,4.468304e-02,0.8261,5.527327e+00,5.715868e+00,7.524672e-01,1.952655e+02,66.53
|
| 131 |
+
2580,0.043754,0.046120,0.022073,0.450,9.371877e-02,0.000000e+00,0.0000,5.526151e+00,5.717966e+00,1.024397e+00,1.952676e+02,67.02
|
| 132 |
+
2600,0.037182,0.044849,0.028349,0.450,6.932639e-02,6.990625e-02,1.0084,5.526687e+00,5.717985e+00,4.684196e-01,1.952709e+02,67.56
|
| 133 |
+
2620,0.017043,0.040249,0.020343,0.400,6.303381e-02,6.454483e-02,1.0240,5.526752e+00,5.717994e+00,3.335332e-01,1.952738e+02,68.06
|
| 134 |
+
2640,0.026466,0.036984,0.022768,0.500,1.526175e-02,0.000000e+00,0.0000,5.526723e+00,5.717186e+00,2.342612e-01,1.952776e+02,68.55
|
| 135 |
+
2660,0.031365,0.037492,0.024923,0.517,2.391365e-01,1.762657e-01,0.7371,5.526470e+00,5.717839e+00,1.823048e+00,1.952771e+02,69.06
|
| 136 |
+
2680,0.032438,0.037450,0.028325,0.487,3.368880e-02,3.372874e-02,1.0012,5.525617e+00,5.718029e+00,4.676573e-01,1.952804e+02,69.57
|
| 137 |
+
2700,0.035828,0.036466,0.029197,0.510,5.747565e-02,0.000000e+00,0.0000,5.526024e+00,5.717881e+00,6.089337e-01,1.952845e+02,70.08
|
| 138 |
+
2720,0.037801,0.035960,0.030564,0.650,2.579338e-01,0.000000e+00,0.0000,5.525787e+00,5.717782e+00,1.409335e+00,1.952878e+02,70.61
|
| 139 |
+
2740,0.019110,0.038256,0.027985,0.725,1.732983e-01,1.313457e-01,0.7579,5.525673e+00,5.718334e+00,1.251163e+00,1.952901e+02,71.12
|
| 140 |
+
2760,0.029690,0.033837,0.023806,0.667,5.298081e-02,0.000000e+00,0.0000,5.525640e+00,5.719467e+00,3.434825e-01,1.952947e+02,71.67
|
| 141 |
+
2780,0.042759,0.053436,0.025772,0.637,1.254243e-01,0.000000e+00,0.0000,5.524930e+00,5.721020e+00,7.920596e-01,1.952959e+02,72.18
|
| 142 |
+
2800,0.049414,0.045787,0.030251,0.640,1.855478e-02,0.000000e+00,0.0000,5.524045e+00,5.721817e+00,2.690353e-01,1.953081e+02,72.69
|
| 143 |
+
2820,0.050984,0.041147,0.028339,0.550,3.648949e-02,2.234597e-02,0.6124,5.524422e+00,5.721487e+00,4.319374e-01,1.953437e+02,73.19
|
| 144 |
+
2840,0.040554,0.036563,0.026669,0.525,1.824847e-01,1.946003e-01,1.0664,5.524665e+00,5.722227e+00,1.347533e+00,1.953574e+02,73.71
|
| 145 |
+
2860,0.026881,0.037020,0.025541,0.517,1.466156e-01,1.376464e-01,0.9388,5.524162e+00,5.723609e+00,6.214639e-01,1.953635e+02,74.21
|
| 146 |
+
2880,0.023069,0.036872,0.029192,0.487,4.845770e-02,2.488118e-02,0.5135,5.523201e+00,5.724657e+00,2.708839e-01,1.953718e+02,74.71
|
| 147 |
+
2900,0.021199,0.032433,0.024000,0.480,3.668863e-02,0.000000e+00,0.0000,5.523520e+00,5.724818e+00,3.000468e-01,1.953821e+02,75.21
|
| 148 |
+
2920,0.032186,0.033642,0.022519,0.450,9.800445e-02,8.224854e-02,0.8392,5.523409e+00,5.725048e+00,4.028916e-01,1.954001e+02,75.74
|
| 149 |
+
2940,0.047098,0.037725,0.032542,0.450,2.131317e-01,0.000000e+00,0.0000,5.523479e+00,5.725803e+00,1.094337e+00,1.954088e+02,76.24
|
| 150 |
+
2960,0.033531,0.040156,0.026629,0.433,1.662405e-01,0.000000e+00,0.0000,5.522806e+00,5.725940e+00,6.949836e-01,1.954487e+02,76.77
|
| 151 |
+
2980,0.046077,0.038951,0.024089,0.438,3.408943e-01,0.000000e+00,0.0000,5.522840e+00,5.727234e+00,1.650528e+00,1.955783e+02,77.28
|
| 152 |
+
3000,0.033280,0.038864,0.027389,0.460,1.318695e-01,1.540455e-01,1.1682,5.522182e+00,5.728519e+00,1.039811e+00,1.956214e+02,77.80
|
diagnostics/tensorboard/events.out.tfevents.1781963285.instance-2asqjmcx-main.1034.0
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:4ac864c4d53af6c0bcf0d70670cdaaa235ced5374e6382cf610d9b0b7f8d2ff8
|
| 3 |
+
size 10780779
|