Add adapter_phase1/trainer_state.json (ckpt-506 final phase 1)
Browse files- adapter_phase1/trainer_state.json +3576 -0
adapter_phase1/trainer_state.json
ADDED
|
@@ -0,0 +1,3576 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_global_step": null,
|
| 3 |
+
"best_metric": null,
|
| 4 |
+
"best_model_checkpoint": null,
|
| 5 |
+
"epoch": 2.0,
|
| 6 |
+
"eval_steps": 500,
|
| 7 |
+
"global_step": 506,
|
| 8 |
+
"is_hyper_param_search": false,
|
| 9 |
+
"is_local_process_zero": true,
|
| 10 |
+
"is_world_process_zero": true,
|
| 11 |
+
"log_history": [
|
| 12 |
+
{
|
| 13 |
+
"epoch": 0.003954522985664854,
|
| 14 |
+
"grad_norm": 0.2544356882572174,
|
| 15 |
+
"learning_rate": 0.0,
|
| 16 |
+
"loss": 1.180017113685608,
|
| 17 |
+
"step": 1
|
| 18 |
+
},
|
| 19 |
+
{
|
| 20 |
+
"epoch": 0.007909045971329708,
|
| 21 |
+
"grad_norm": 0.2469092607498169,
|
| 22 |
+
"learning_rate": 4e-05,
|
| 23 |
+
"loss": 1.1813130378723145,
|
| 24 |
+
"step": 2
|
| 25 |
+
},
|
| 26 |
+
{
|
| 27 |
+
"epoch": 0.011863568956994563,
|
| 28 |
+
"grad_norm": 0.20873534679412842,
|
| 29 |
+
"learning_rate": 8e-05,
|
| 30 |
+
"loss": 1.1847681999206543,
|
| 31 |
+
"step": 3
|
| 32 |
+
},
|
| 33 |
+
{
|
| 34 |
+
"epoch": 0.015818091942659415,
|
| 35 |
+
"grad_norm": 0.22851118445396423,
|
| 36 |
+
"learning_rate": 0.00012,
|
| 37 |
+
"loss": 1.1241787672042847,
|
| 38 |
+
"step": 4
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 0.019772614928324272,
|
| 42 |
+
"grad_norm": 0.22521217167377472,
|
| 43 |
+
"learning_rate": 0.00016,
|
| 44 |
+
"loss": 1.1387571096420288,
|
| 45 |
+
"step": 5
|
| 46 |
+
},
|
| 47 |
+
{
|
| 48 |
+
"epoch": 0.023727137913989126,
|
| 49 |
+
"grad_norm": 0.21741724014282227,
|
| 50 |
+
"learning_rate": 0.0002,
|
| 51 |
+
"loss": 1.103634238243103,
|
| 52 |
+
"step": 6
|
| 53 |
+
},
|
| 54 |
+
{
|
| 55 |
+
"epoch": 0.02768166089965398,
|
| 56 |
+
"grad_norm": 0.1713864952325821,
|
| 57 |
+
"learning_rate": 0.0001996007984031936,
|
| 58 |
+
"loss": 1.1193780899047852,
|
| 59 |
+
"step": 7
|
| 60 |
+
},
|
| 61 |
+
{
|
| 62 |
+
"epoch": 0.03163618388531883,
|
| 63 |
+
"grad_norm": 0.1551392674446106,
|
| 64 |
+
"learning_rate": 0.00019920159680638725,
|
| 65 |
+
"loss": 1.209862470626831,
|
| 66 |
+
"step": 8
|
| 67 |
+
},
|
| 68 |
+
{
|
| 69 |
+
"epoch": 0.03559070687098369,
|
| 70 |
+
"grad_norm": 0.17870232462882996,
|
| 71 |
+
"learning_rate": 0.00019880239520958084,
|
| 72 |
+
"loss": 1.1225727796554565,
|
| 73 |
+
"step": 9
|
| 74 |
+
},
|
| 75 |
+
{
|
| 76 |
+
"epoch": 0.039545229856648545,
|
| 77 |
+
"grad_norm": 0.20667865872383118,
|
| 78 |
+
"learning_rate": 0.00019840319361277446,
|
| 79 |
+
"loss": 1.0327069759368896,
|
| 80 |
+
"step": 10
|
| 81 |
+
},
|
| 82 |
+
{
|
| 83 |
+
"epoch": 0.043499752842313395,
|
| 84 |
+
"grad_norm": 0.19965171813964844,
|
| 85 |
+
"learning_rate": 0.00019800399201596808,
|
| 86 |
+
"loss": 0.9479315280914307,
|
| 87 |
+
"step": 11
|
| 88 |
+
},
|
| 89 |
+
{
|
| 90 |
+
"epoch": 0.04745427582797825,
|
| 91 |
+
"grad_norm": 0.21041055023670197,
|
| 92 |
+
"learning_rate": 0.0001976047904191617,
|
| 93 |
+
"loss": 1.1442006826400757,
|
| 94 |
+
"step": 12
|
| 95 |
+
},
|
| 96 |
+
{
|
| 97 |
+
"epoch": 0.0514087988136431,
|
| 98 |
+
"grad_norm": 0.16018487513065338,
|
| 99 |
+
"learning_rate": 0.0001972055888223553,
|
| 100 |
+
"loss": 1.0416808128356934,
|
| 101 |
+
"step": 13
|
| 102 |
+
},
|
| 103 |
+
{
|
| 104 |
+
"epoch": 0.05536332179930796,
|
| 105 |
+
"grad_norm": 0.141277477145195,
|
| 106 |
+
"learning_rate": 0.0001968063872255489,
|
| 107 |
+
"loss": 1.1698994636535645,
|
| 108 |
+
"step": 14
|
| 109 |
+
},
|
| 110 |
+
{
|
| 111 |
+
"epoch": 0.05931784478497281,
|
| 112 |
+
"grad_norm": 0.12079285830259323,
|
| 113 |
+
"learning_rate": 0.00019640718562874253,
|
| 114 |
+
"loss": 0.9101760983467102,
|
| 115 |
+
"step": 15
|
| 116 |
+
},
|
| 117 |
+
{
|
| 118 |
+
"epoch": 0.06327236777063766,
|
| 119 |
+
"grad_norm": 0.13878197968006134,
|
| 120 |
+
"learning_rate": 0.00019600798403193614,
|
| 121 |
+
"loss": 1.1655550003051758,
|
| 122 |
+
"step": 16
|
| 123 |
+
},
|
| 124 |
+
{
|
| 125 |
+
"epoch": 0.06722689075630252,
|
| 126 |
+
"grad_norm": 0.16896755993366241,
|
| 127 |
+
"learning_rate": 0.00019560878243512974,
|
| 128 |
+
"loss": 1.0796542167663574,
|
| 129 |
+
"step": 17
|
| 130 |
+
},
|
| 131 |
+
{
|
| 132 |
+
"epoch": 0.07118141374196738,
|
| 133 |
+
"grad_norm": 0.12621314823627472,
|
| 134 |
+
"learning_rate": 0.00019520958083832338,
|
| 135 |
+
"loss": 1.1336424350738525,
|
| 136 |
+
"step": 18
|
| 137 |
+
},
|
| 138 |
+
{
|
| 139 |
+
"epoch": 0.07513593672763223,
|
| 140 |
+
"grad_norm": 0.1260470598936081,
|
| 141 |
+
"learning_rate": 0.00019481037924151697,
|
| 142 |
+
"loss": 1.02823805809021,
|
| 143 |
+
"step": 19
|
| 144 |
+
},
|
| 145 |
+
{
|
| 146 |
+
"epoch": 0.07909045971329709,
|
| 147 |
+
"grad_norm": 0.1230599507689476,
|
| 148 |
+
"learning_rate": 0.0001944111776447106,
|
| 149 |
+
"loss": 0.9280879497528076,
|
| 150 |
+
"step": 20
|
| 151 |
+
},
|
| 152 |
+
{
|
| 153 |
+
"epoch": 0.08304498269896193,
|
| 154 |
+
"grad_norm": 0.11235114187002182,
|
| 155 |
+
"learning_rate": 0.00019401197604790419,
|
| 156 |
+
"loss": 1.0234721899032593,
|
| 157 |
+
"step": 21
|
| 158 |
+
},
|
| 159 |
+
{
|
| 160 |
+
"epoch": 0.08699950568462679,
|
| 161 |
+
"grad_norm": 0.1244259923696518,
|
| 162 |
+
"learning_rate": 0.00019361277445109783,
|
| 163 |
+
"loss": 0.9619145393371582,
|
| 164 |
+
"step": 22
|
| 165 |
+
},
|
| 166 |
+
{
|
| 167 |
+
"epoch": 0.09095402867029165,
|
| 168 |
+
"grad_norm": 0.10888625681400299,
|
| 169 |
+
"learning_rate": 0.00019321357285429142,
|
| 170 |
+
"loss": 1.0350663661956787,
|
| 171 |
+
"step": 23
|
| 172 |
+
},
|
| 173 |
+
{
|
| 174 |
+
"epoch": 0.0949085516559565,
|
| 175 |
+
"grad_norm": 0.12405306845903397,
|
| 176 |
+
"learning_rate": 0.00019281437125748504,
|
| 177 |
+
"loss": 1.1137639284133911,
|
| 178 |
+
"step": 24
|
| 179 |
+
},
|
| 180 |
+
{
|
| 181 |
+
"epoch": 0.09886307464162135,
|
| 182 |
+
"grad_norm": 0.1365104615688324,
|
| 183 |
+
"learning_rate": 0.00019241516966067866,
|
| 184 |
+
"loss": 1.082958459854126,
|
| 185 |
+
"step": 25
|
| 186 |
+
},
|
| 187 |
+
{
|
| 188 |
+
"epoch": 0.1028175976272862,
|
| 189 |
+
"grad_norm": 0.12047652900218964,
|
| 190 |
+
"learning_rate": 0.00019201596806387228,
|
| 191 |
+
"loss": 1.031674861907959,
|
| 192 |
+
"step": 26
|
| 193 |
+
},
|
| 194 |
+
{
|
| 195 |
+
"epoch": 0.10677212061295106,
|
| 196 |
+
"grad_norm": 0.10935146361589432,
|
| 197 |
+
"learning_rate": 0.00019161676646706587,
|
| 198 |
+
"loss": 0.9122903347015381,
|
| 199 |
+
"step": 27
|
| 200 |
+
},
|
| 201 |
+
{
|
| 202 |
+
"epoch": 0.11072664359861592,
|
| 203 |
+
"grad_norm": 0.1395711451768875,
|
| 204 |
+
"learning_rate": 0.0001912175648702595,
|
| 205 |
+
"loss": 0.9628266096115112,
|
| 206 |
+
"step": 28
|
| 207 |
+
},
|
| 208 |
+
{
|
| 209 |
+
"epoch": 0.11468116658428078,
|
| 210 |
+
"grad_norm": 0.13380995392799377,
|
| 211 |
+
"learning_rate": 0.0001908183632734531,
|
| 212 |
+
"loss": 0.962996244430542,
|
| 213 |
+
"step": 29
|
| 214 |
+
},
|
| 215 |
+
{
|
| 216 |
+
"epoch": 0.11863568956994562,
|
| 217 |
+
"grad_norm": 0.12701313197612762,
|
| 218 |
+
"learning_rate": 0.0001904191616766467,
|
| 219 |
+
"loss": 0.9316821694374084,
|
| 220 |
+
"step": 30
|
| 221 |
+
},
|
| 222 |
+
{
|
| 223 |
+
"epoch": 0.12259021255561048,
|
| 224 |
+
"grad_norm": 0.12657427787780762,
|
| 225 |
+
"learning_rate": 0.00019001996007984032,
|
| 226 |
+
"loss": 1.0292439460754395,
|
| 227 |
+
"step": 31
|
| 228 |
+
},
|
| 229 |
+
{
|
| 230 |
+
"epoch": 0.12654473554127532,
|
| 231 |
+
"grad_norm": 0.13929884135723114,
|
| 232 |
+
"learning_rate": 0.00018962075848303394,
|
| 233 |
+
"loss": 1.0495136976242065,
|
| 234 |
+
"step": 32
|
| 235 |
+
},
|
| 236 |
+
{
|
| 237 |
+
"epoch": 0.1304992585269402,
|
| 238 |
+
"grad_norm": 0.1292039155960083,
|
| 239 |
+
"learning_rate": 0.00018922155688622756,
|
| 240 |
+
"loss": 1.0562491416931152,
|
| 241 |
+
"step": 33
|
| 242 |
+
},
|
| 243 |
+
{
|
| 244 |
+
"epoch": 0.13445378151260504,
|
| 245 |
+
"grad_norm": 0.1383693516254425,
|
| 246 |
+
"learning_rate": 0.00018882235528942115,
|
| 247 |
+
"loss": 1.012336015701294,
|
| 248 |
+
"step": 34
|
| 249 |
+
},
|
| 250 |
+
{
|
| 251 |
+
"epoch": 0.1384083044982699,
|
| 252 |
+
"grad_norm": 0.11728740483522415,
|
| 253 |
+
"learning_rate": 0.0001884231536926148,
|
| 254 |
+
"loss": 0.903199315071106,
|
| 255 |
+
"step": 35
|
| 256 |
+
},
|
| 257 |
+
{
|
| 258 |
+
"epoch": 0.14236282748393475,
|
| 259 |
+
"grad_norm": 0.12921994924545288,
|
| 260 |
+
"learning_rate": 0.0001880239520958084,
|
| 261 |
+
"loss": 1.0892961025238037,
|
| 262 |
+
"step": 36
|
| 263 |
+
},
|
| 264 |
+
{
|
| 265 |
+
"epoch": 0.1463173504695996,
|
| 266 |
+
"grad_norm": 0.1262505203485489,
|
| 267 |
+
"learning_rate": 0.000187624750499002,
|
| 268 |
+
"loss": 1.1606205701828003,
|
| 269 |
+
"step": 37
|
| 270 |
+
},
|
| 271 |
+
{
|
| 272 |
+
"epoch": 0.15027187345526447,
|
| 273 |
+
"grad_norm": 0.12508349120616913,
|
| 274 |
+
"learning_rate": 0.00018722554890219563,
|
| 275 |
+
"loss": 1.0895495414733887,
|
| 276 |
+
"step": 38
|
| 277 |
+
},
|
| 278 |
+
{
|
| 279 |
+
"epoch": 0.1542263964409293,
|
| 280 |
+
"grad_norm": 0.12130790203809738,
|
| 281 |
+
"learning_rate": 0.00018682634730538924,
|
| 282 |
+
"loss": 1.0368725061416626,
|
| 283 |
+
"step": 39
|
| 284 |
+
},
|
| 285 |
+
{
|
| 286 |
+
"epoch": 0.15818091942659418,
|
| 287 |
+
"grad_norm": 0.12927809357643127,
|
| 288 |
+
"learning_rate": 0.00018642714570858284,
|
| 289 |
+
"loss": 1.0340652465820312,
|
| 290 |
+
"step": 40
|
| 291 |
+
},
|
| 292 |
+
{
|
| 293 |
+
"epoch": 0.16213544241225902,
|
| 294 |
+
"grad_norm": 0.12635737657546997,
|
| 295 |
+
"learning_rate": 0.00018602794411177646,
|
| 296 |
+
"loss": 1.1098253726959229,
|
| 297 |
+
"step": 41
|
| 298 |
+
},
|
| 299 |
+
{
|
| 300 |
+
"epoch": 0.16608996539792387,
|
| 301 |
+
"grad_norm": 0.1303265392780304,
|
| 302 |
+
"learning_rate": 0.00018562874251497007,
|
| 303 |
+
"loss": 1.0297881364822388,
|
| 304 |
+
"step": 42
|
| 305 |
+
},
|
| 306 |
+
{
|
| 307 |
+
"epoch": 0.17004448838358874,
|
| 308 |
+
"grad_norm": 0.12825307250022888,
|
| 309 |
+
"learning_rate": 0.0001852295409181637,
|
| 310 |
+
"loss": 0.9129877090454102,
|
| 311 |
+
"step": 43
|
| 312 |
+
},
|
| 313 |
+
{
|
| 314 |
+
"epoch": 0.17399901136925358,
|
| 315 |
+
"grad_norm": 0.12521876394748688,
|
| 316 |
+
"learning_rate": 0.00018483033932135728,
|
| 317 |
+
"loss": 0.97171550989151,
|
| 318 |
+
"step": 44
|
| 319 |
+
},
|
| 320 |
+
{
|
| 321 |
+
"epoch": 0.17795353435491842,
|
| 322 |
+
"grad_norm": 0.13904021680355072,
|
| 323 |
+
"learning_rate": 0.00018443113772455093,
|
| 324 |
+
"loss": 1.097721815109253,
|
| 325 |
+
"step": 45
|
| 326 |
+
},
|
| 327 |
+
{
|
| 328 |
+
"epoch": 0.1819080573405833,
|
| 329 |
+
"grad_norm": 0.14009957015514374,
|
| 330 |
+
"learning_rate": 0.00018403193612774452,
|
| 331 |
+
"loss": 0.9984307885169983,
|
| 332 |
+
"step": 46
|
| 333 |
+
},
|
| 334 |
+
{
|
| 335 |
+
"epoch": 0.18586258032624814,
|
| 336 |
+
"grad_norm": 0.12277879565954208,
|
| 337 |
+
"learning_rate": 0.00018363273453093811,
|
| 338 |
+
"loss": 0.9881635904312134,
|
| 339 |
+
"step": 47
|
| 340 |
+
},
|
| 341 |
+
{
|
| 342 |
+
"epoch": 0.189817103311913,
|
| 343 |
+
"grad_norm": 0.13878905773162842,
|
| 344 |
+
"learning_rate": 0.00018323353293413173,
|
| 345 |
+
"loss": 0.9900161027908325,
|
| 346 |
+
"step": 48
|
| 347 |
+
},
|
| 348 |
+
{
|
| 349 |
+
"epoch": 0.19377162629757785,
|
| 350 |
+
"grad_norm": 0.1394307166337967,
|
| 351 |
+
"learning_rate": 0.00018283433133732535,
|
| 352 |
+
"loss": 0.8905137777328491,
|
| 353 |
+
"step": 49
|
| 354 |
+
},
|
| 355 |
+
{
|
| 356 |
+
"epoch": 0.1977261492832427,
|
| 357 |
+
"grad_norm": 0.1348118931055069,
|
| 358 |
+
"learning_rate": 0.00018243512974051897,
|
| 359 |
+
"loss": 1.001299500465393,
|
| 360 |
+
"step": 50
|
| 361 |
+
},
|
| 362 |
+
{
|
| 363 |
+
"epoch": 0.20168067226890757,
|
| 364 |
+
"grad_norm": 0.14841432869434357,
|
| 365 |
+
"learning_rate": 0.00018203592814371256,
|
| 366 |
+
"loss": 1.114479899406433,
|
| 367 |
+
"step": 51
|
| 368 |
+
},
|
| 369 |
+
{
|
| 370 |
+
"epoch": 0.2056351952545724,
|
| 371 |
+
"grad_norm": 0.12918154895305634,
|
| 372 |
+
"learning_rate": 0.0001816367265469062,
|
| 373 |
+
"loss": 1.075540542602539,
|
| 374 |
+
"step": 52
|
| 375 |
+
},
|
| 376 |
+
{
|
| 377 |
+
"epoch": 0.20958971824023728,
|
| 378 |
+
"grad_norm": 0.137839674949646,
|
| 379 |
+
"learning_rate": 0.0001812375249500998,
|
| 380 |
+
"loss": 0.9629291892051697,
|
| 381 |
+
"step": 53
|
| 382 |
+
},
|
| 383 |
+
{
|
| 384 |
+
"epoch": 0.21354424122590213,
|
| 385 |
+
"grad_norm": 0.12743251025676727,
|
| 386 |
+
"learning_rate": 0.00018083832335329342,
|
| 387 |
+
"loss": 0.8884077668190002,
|
| 388 |
+
"step": 54
|
| 389 |
+
},
|
| 390 |
+
{
|
| 391 |
+
"epoch": 0.21749876421156697,
|
| 392 |
+
"grad_norm": 0.14579665660858154,
|
| 393 |
+
"learning_rate": 0.00018043912175648704,
|
| 394 |
+
"loss": 1.0494593381881714,
|
| 395 |
+
"step": 55
|
| 396 |
+
},
|
| 397 |
+
{
|
| 398 |
+
"epoch": 0.22145328719723184,
|
| 399 |
+
"grad_norm": 0.1444905400276184,
|
| 400 |
+
"learning_rate": 0.00018003992015968066,
|
| 401 |
+
"loss": 0.8838294744491577,
|
| 402 |
+
"step": 56
|
| 403 |
+
},
|
| 404 |
+
{
|
| 405 |
+
"epoch": 0.22540781018289668,
|
| 406 |
+
"grad_norm": 0.13476018607616425,
|
| 407 |
+
"learning_rate": 0.00017964071856287425,
|
| 408 |
+
"loss": 0.9571032524108887,
|
| 409 |
+
"step": 57
|
| 410 |
+
},
|
| 411 |
+
{
|
| 412 |
+
"epoch": 0.22936233316856156,
|
| 413 |
+
"grad_norm": 0.1387203335762024,
|
| 414 |
+
"learning_rate": 0.00017924151696606787,
|
| 415 |
+
"loss": 1.0317763090133667,
|
| 416 |
+
"step": 58
|
| 417 |
+
},
|
| 418 |
+
{
|
| 419 |
+
"epoch": 0.2333168561542264,
|
| 420 |
+
"grad_norm": 0.14424876868724823,
|
| 421 |
+
"learning_rate": 0.0001788423153692615,
|
| 422 |
+
"loss": 1.0047612190246582,
|
| 423 |
+
"step": 59
|
| 424 |
+
},
|
| 425 |
+
{
|
| 426 |
+
"epoch": 0.23727137913989124,
|
| 427 |
+
"grad_norm": 0.13895335793495178,
|
| 428 |
+
"learning_rate": 0.0001784431137724551,
|
| 429 |
+
"loss": 1.0508701801300049,
|
| 430 |
+
"step": 60
|
| 431 |
+
},
|
| 432 |
+
{
|
| 433 |
+
"epoch": 0.2412259021255561,
|
| 434 |
+
"grad_norm": 0.1239890530705452,
|
| 435 |
+
"learning_rate": 0.0001780439121756487,
|
| 436 |
+
"loss": 1.003294587135315,
|
| 437 |
+
"step": 61
|
| 438 |
+
},
|
| 439 |
+
{
|
| 440 |
+
"epoch": 0.24518042511122096,
|
| 441 |
+
"grad_norm": 0.13500665128231049,
|
| 442 |
+
"learning_rate": 0.00017764471057884234,
|
| 443 |
+
"loss": 0.9486382007598877,
|
| 444 |
+
"step": 62
|
| 445 |
+
},
|
| 446 |
+
{
|
| 447 |
+
"epoch": 0.2491349480968858,
|
| 448 |
+
"grad_norm": 0.14559301733970642,
|
| 449 |
+
"learning_rate": 0.00017724550898203594,
|
| 450 |
+
"loss": 1.0008798837661743,
|
| 451 |
+
"step": 63
|
| 452 |
+
},
|
| 453 |
+
{
|
| 454 |
+
"epoch": 0.25308947108255064,
|
| 455 |
+
"grad_norm": 0.15551623702049255,
|
| 456 |
+
"learning_rate": 0.00017684630738522955,
|
| 457 |
+
"loss": 0.9119284749031067,
|
| 458 |
+
"step": 64
|
| 459 |
+
},
|
| 460 |
+
{
|
| 461 |
+
"epoch": 0.25704399406821554,
|
| 462 |
+
"grad_norm": 0.12912505865097046,
|
| 463 |
+
"learning_rate": 0.00017644710578842317,
|
| 464 |
+
"loss": 1.041481614112854,
|
| 465 |
+
"step": 65
|
| 466 |
+
},
|
| 467 |
+
{
|
| 468 |
+
"epoch": 0.2609985170538804,
|
| 469 |
+
"grad_norm": 0.14489996433258057,
|
| 470 |
+
"learning_rate": 0.0001760479041916168,
|
| 471 |
+
"loss": 0.8673788905143738,
|
| 472 |
+
"step": 66
|
| 473 |
+
},
|
| 474 |
+
{
|
| 475 |
+
"epoch": 0.26495304003954523,
|
| 476 |
+
"grad_norm": 0.14056646823883057,
|
| 477 |
+
"learning_rate": 0.00017564870259481038,
|
| 478 |
+
"loss": 0.9839107394218445,
|
| 479 |
+
"step": 67
|
| 480 |
+
},
|
| 481 |
+
{
|
| 482 |
+
"epoch": 0.2689075630252101,
|
| 483 |
+
"grad_norm": 0.16830863058567047,
|
| 484 |
+
"learning_rate": 0.000175249500998004,
|
| 485 |
+
"loss": 1.0539538860321045,
|
| 486 |
+
"step": 68
|
| 487 |
+
},
|
| 488 |
+
{
|
| 489 |
+
"epoch": 0.2728620860108749,
|
| 490 |
+
"grad_norm": 0.14673052728176117,
|
| 491 |
+
"learning_rate": 0.00017485029940119762,
|
| 492 |
+
"loss": 0.9210308790206909,
|
| 493 |
+
"step": 69
|
| 494 |
+
},
|
| 495 |
+
{
|
| 496 |
+
"epoch": 0.2768166089965398,
|
| 497 |
+
"grad_norm": 0.1400182992219925,
|
| 498 |
+
"learning_rate": 0.00017445109780439121,
|
| 499 |
+
"loss": 0.9093458652496338,
|
| 500 |
+
"step": 70
|
| 501 |
+
},
|
| 502 |
+
{
|
| 503 |
+
"epoch": 0.28077113198220466,
|
| 504 |
+
"grad_norm": 0.14567220211029053,
|
| 505 |
+
"learning_rate": 0.00017405189620758483,
|
| 506 |
+
"loss": 0.9801154136657715,
|
| 507 |
+
"step": 71
|
| 508 |
+
},
|
| 509 |
+
{
|
| 510 |
+
"epoch": 0.2847256549678695,
|
| 511 |
+
"grad_norm": 0.15140396356582642,
|
| 512 |
+
"learning_rate": 0.00017365269461077845,
|
| 513 |
+
"loss": 1.0192142724990845,
|
| 514 |
+
"step": 72
|
| 515 |
+
},
|
| 516 |
+
{
|
| 517 |
+
"epoch": 0.28868017795353434,
|
| 518 |
+
"grad_norm": 0.15245863795280457,
|
| 519 |
+
"learning_rate": 0.00017325349301397207,
|
| 520 |
+
"loss": 1.0319201946258545,
|
| 521 |
+
"step": 73
|
| 522 |
+
},
|
| 523 |
+
{
|
| 524 |
+
"epoch": 0.2926347009391992,
|
| 525 |
+
"grad_norm": 0.15047654509544373,
|
| 526 |
+
"learning_rate": 0.00017285429141716566,
|
| 527 |
+
"loss": 1.0096640586853027,
|
| 528 |
+
"step": 74
|
| 529 |
+
},
|
| 530 |
+
{
|
| 531 |
+
"epoch": 0.2965892239248641,
|
| 532 |
+
"grad_norm": 0.14069028198719025,
|
| 533 |
+
"learning_rate": 0.00017245508982035928,
|
| 534 |
+
"loss": 0.9506325125694275,
|
| 535 |
+
"step": 75
|
| 536 |
+
},
|
| 537 |
+
{
|
| 538 |
+
"epoch": 0.30054374691052893,
|
| 539 |
+
"grad_norm": 0.13782000541687012,
|
| 540 |
+
"learning_rate": 0.0001720558882235529,
|
| 541 |
+
"loss": 0.8749170303344727,
|
| 542 |
+
"step": 76
|
| 543 |
+
},
|
| 544 |
+
{
|
| 545 |
+
"epoch": 0.3044982698961938,
|
| 546 |
+
"grad_norm": 0.1548694223165512,
|
| 547 |
+
"learning_rate": 0.00017165668662674652,
|
| 548 |
+
"loss": 0.9927857518196106,
|
| 549 |
+
"step": 77
|
| 550 |
+
},
|
| 551 |
+
{
|
| 552 |
+
"epoch": 0.3084527928818586,
|
| 553 |
+
"grad_norm": 0.14958986639976501,
|
| 554 |
+
"learning_rate": 0.0001712574850299401,
|
| 555 |
+
"loss": 0.7992951273918152,
|
| 556 |
+
"step": 78
|
| 557 |
+
},
|
| 558 |
+
{
|
| 559 |
+
"epoch": 0.31240731586752346,
|
| 560 |
+
"grad_norm": 0.14522351324558258,
|
| 561 |
+
"learning_rate": 0.00017085828343313376,
|
| 562 |
+
"loss": 0.9930732846260071,
|
| 563 |
+
"step": 79
|
| 564 |
+
},
|
| 565 |
+
{
|
| 566 |
+
"epoch": 0.31636183885318836,
|
| 567 |
+
"grad_norm": 0.14936009049415588,
|
| 568 |
+
"learning_rate": 0.00017045908183632735,
|
| 569 |
+
"loss": 1.0836726427078247,
|
| 570 |
+
"step": 80
|
| 571 |
+
},
|
| 572 |
+
{
|
| 573 |
+
"epoch": 0.3203163618388532,
|
| 574 |
+
"grad_norm": 0.13863688707351685,
|
| 575 |
+
"learning_rate": 0.00017005988023952097,
|
| 576 |
+
"loss": 0.9540154933929443,
|
| 577 |
+
"step": 81
|
| 578 |
+
},
|
| 579 |
+
{
|
| 580 |
+
"epoch": 0.32427088482451805,
|
| 581 |
+
"grad_norm": 0.14704303443431854,
|
| 582 |
+
"learning_rate": 0.0001696606786427146,
|
| 583 |
+
"loss": 1.03548002243042,
|
| 584 |
+
"step": 82
|
| 585 |
+
},
|
| 586 |
+
{
|
| 587 |
+
"epoch": 0.3282254078101829,
|
| 588 |
+
"grad_norm": 0.13630537688732147,
|
| 589 |
+
"learning_rate": 0.0001692614770459082,
|
| 590 |
+
"loss": 0.9100798964500427,
|
| 591 |
+
"step": 83
|
| 592 |
+
},
|
| 593 |
+
{
|
| 594 |
+
"epoch": 0.33217993079584773,
|
| 595 |
+
"grad_norm": 0.14166879653930664,
|
| 596 |
+
"learning_rate": 0.0001688622754491018,
|
| 597 |
+
"loss": 1.0137436389923096,
|
| 598 |
+
"step": 84
|
| 599 |
+
},
|
| 600 |
+
{
|
| 601 |
+
"epoch": 0.33613445378151263,
|
| 602 |
+
"grad_norm": 0.14558711647987366,
|
| 603 |
+
"learning_rate": 0.00016846307385229542,
|
| 604 |
+
"loss": 1.0042855739593506,
|
| 605 |
+
"step": 85
|
| 606 |
+
},
|
| 607 |
+
{
|
| 608 |
+
"epoch": 0.3400889767671775,
|
| 609 |
+
"grad_norm": 0.13049034774303436,
|
| 610 |
+
"learning_rate": 0.00016806387225548904,
|
| 611 |
+
"loss": 0.9128093123435974,
|
| 612 |
+
"step": 86
|
| 613 |
+
},
|
| 614 |
+
{
|
| 615 |
+
"epoch": 0.3440434997528423,
|
| 616 |
+
"grad_norm": 0.1458376795053482,
|
| 617 |
+
"learning_rate": 0.00016766467065868263,
|
| 618 |
+
"loss": 0.9017834663391113,
|
| 619 |
+
"step": 87
|
| 620 |
+
},
|
| 621 |
+
{
|
| 622 |
+
"epoch": 0.34799802273850716,
|
| 623 |
+
"grad_norm": 0.13847292959690094,
|
| 624 |
+
"learning_rate": 0.00016726546906187625,
|
| 625 |
+
"loss": 0.9473759531974792,
|
| 626 |
+
"step": 88
|
| 627 |
+
},
|
| 628 |
+
{
|
| 629 |
+
"epoch": 0.351952545724172,
|
| 630 |
+
"grad_norm": 0.15093441307544708,
|
| 631 |
+
"learning_rate": 0.00016686626746506987,
|
| 632 |
+
"loss": 0.9916189312934875,
|
| 633 |
+
"step": 89
|
| 634 |
+
},
|
| 635 |
+
{
|
| 636 |
+
"epoch": 0.35590706870983685,
|
| 637 |
+
"grad_norm": 0.1472613364458084,
|
| 638 |
+
"learning_rate": 0.00016646706586826348,
|
| 639 |
+
"loss": 0.9674598574638367,
|
| 640 |
+
"step": 90
|
| 641 |
+
},
|
| 642 |
+
{
|
| 643 |
+
"epoch": 0.35986159169550175,
|
| 644 |
+
"grad_norm": 0.13850581645965576,
|
| 645 |
+
"learning_rate": 0.00016606786427145708,
|
| 646 |
+
"loss": 0.972917377948761,
|
| 647 |
+
"step": 91
|
| 648 |
+
},
|
| 649 |
+
{
|
| 650 |
+
"epoch": 0.3638161146811666,
|
| 651 |
+
"grad_norm": 0.15554893016815186,
|
| 652 |
+
"learning_rate": 0.00016566866267465072,
|
| 653 |
+
"loss": 0.8598817586898804,
|
| 654 |
+
"step": 92
|
| 655 |
+
},
|
| 656 |
+
{
|
| 657 |
+
"epoch": 0.36777063766683143,
|
| 658 |
+
"grad_norm": 0.15589317679405212,
|
| 659 |
+
"learning_rate": 0.00016526946107784431,
|
| 660 |
+
"loss": 1.0263655185699463,
|
| 661 |
+
"step": 93
|
| 662 |
+
},
|
| 663 |
+
{
|
| 664 |
+
"epoch": 0.3717251606524963,
|
| 665 |
+
"grad_norm": 0.14788870513439178,
|
| 666 |
+
"learning_rate": 0.00016487025948103793,
|
| 667 |
+
"loss": 1.0757544040679932,
|
| 668 |
+
"step": 94
|
| 669 |
+
},
|
| 670 |
+
{
|
| 671 |
+
"epoch": 0.3756796836381611,
|
| 672 |
+
"grad_norm": 0.15305693447589874,
|
| 673 |
+
"learning_rate": 0.00016447105788423155,
|
| 674 |
+
"loss": 0.9259313344955444,
|
| 675 |
+
"step": 95
|
| 676 |
+
},
|
| 677 |
+
{
|
| 678 |
+
"epoch": 0.379634206623826,
|
| 679 |
+
"grad_norm": 0.14788882434368134,
|
| 680 |
+
"learning_rate": 0.00016407185628742517,
|
| 681 |
+
"loss": 1.0758525133132935,
|
| 682 |
+
"step": 96
|
| 683 |
+
},
|
| 684 |
+
{
|
| 685 |
+
"epoch": 0.38358872960949086,
|
| 686 |
+
"grad_norm": 0.15263086557388306,
|
| 687 |
+
"learning_rate": 0.00016367265469061876,
|
| 688 |
+
"loss": 0.9790467023849487,
|
| 689 |
+
"step": 97
|
| 690 |
+
},
|
| 691 |
+
{
|
| 692 |
+
"epoch": 0.3875432525951557,
|
| 693 |
+
"grad_norm": 0.16988050937652588,
|
| 694 |
+
"learning_rate": 0.00016327345309381238,
|
| 695 |
+
"loss": 1.180403470993042,
|
| 696 |
+
"step": 98
|
| 697 |
+
},
|
| 698 |
+
{
|
| 699 |
+
"epoch": 0.39149777558082055,
|
| 700 |
+
"grad_norm": 0.142135351896286,
|
| 701 |
+
"learning_rate": 0.000162874251497006,
|
| 702 |
+
"loss": 1.0949641466140747,
|
| 703 |
+
"step": 99
|
| 704 |
+
},
|
| 705 |
+
{
|
| 706 |
+
"epoch": 0.3954522985664854,
|
| 707 |
+
"grad_norm": 0.1378023624420166,
|
| 708 |
+
"learning_rate": 0.00016247504990019962,
|
| 709 |
+
"loss": 1.0425848960876465,
|
| 710 |
+
"step": 100
|
| 711 |
+
},
|
| 712 |
+
{
|
| 713 |
+
"epoch": 0.3994068215521503,
|
| 714 |
+
"grad_norm": 0.1591092050075531,
|
| 715 |
+
"learning_rate": 0.0001620758483033932,
|
| 716 |
+
"loss": 0.9338038563728333,
|
| 717 |
+
"step": 101
|
| 718 |
+
},
|
| 719 |
+
{
|
| 720 |
+
"epoch": 0.40336134453781514,
|
| 721 |
+
"grad_norm": 0.1520695835351944,
|
| 722 |
+
"learning_rate": 0.00016167664670658683,
|
| 723 |
+
"loss": 0.8926813006401062,
|
| 724 |
+
"step": 102
|
| 725 |
+
},
|
| 726 |
+
{
|
| 727 |
+
"epoch": 0.40731586752348,
|
| 728 |
+
"grad_norm": 0.14733289182186127,
|
| 729 |
+
"learning_rate": 0.00016127744510978045,
|
| 730 |
+
"loss": 1.082033395767212,
|
| 731 |
+
"step": 103
|
| 732 |
+
},
|
| 733 |
+
{
|
| 734 |
+
"epoch": 0.4112703905091448,
|
| 735 |
+
"grad_norm": 0.14460161328315735,
|
| 736 |
+
"learning_rate": 0.00016087824351297407,
|
| 737 |
+
"loss": 0.9306143522262573,
|
| 738 |
+
"step": 104
|
| 739 |
+
},
|
| 740 |
+
{
|
| 741 |
+
"epoch": 0.41522491349480967,
|
| 742 |
+
"grad_norm": 0.16207540035247803,
|
| 743 |
+
"learning_rate": 0.00016047904191616766,
|
| 744 |
+
"loss": 1.0030015707015991,
|
| 745 |
+
"step": 105
|
| 746 |
+
},
|
| 747 |
+
{
|
| 748 |
+
"epoch": 0.41917943648047457,
|
| 749 |
+
"grad_norm": 0.1540522575378418,
|
| 750 |
+
"learning_rate": 0.0001600798403193613,
|
| 751 |
+
"loss": 1.1492794752120972,
|
| 752 |
+
"step": 106
|
| 753 |
+
},
|
| 754 |
+
{
|
| 755 |
+
"epoch": 0.4231339594661394,
|
| 756 |
+
"grad_norm": 0.13939417898654938,
|
| 757 |
+
"learning_rate": 0.0001596806387225549,
|
| 758 |
+
"loss": 0.9942459464073181,
|
| 759 |
+
"step": 107
|
| 760 |
+
},
|
| 761 |
+
{
|
| 762 |
+
"epoch": 0.42708848245180425,
|
| 763 |
+
"grad_norm": 0.17469069361686707,
|
| 764 |
+
"learning_rate": 0.0001592814371257485,
|
| 765 |
+
"loss": 0.9437049031257629,
|
| 766 |
+
"step": 108
|
| 767 |
+
},
|
| 768 |
+
{
|
| 769 |
+
"epoch": 0.4310430054374691,
|
| 770 |
+
"grad_norm": 0.15532726049423218,
|
| 771 |
+
"learning_rate": 0.00015888223552894214,
|
| 772 |
+
"loss": 0.8414091467857361,
|
| 773 |
+
"step": 109
|
| 774 |
+
},
|
| 775 |
+
{
|
| 776 |
+
"epoch": 0.43499752842313394,
|
| 777 |
+
"grad_norm": 0.1431339681148529,
|
| 778 |
+
"learning_rate": 0.00015848303393213573,
|
| 779 |
+
"loss": 0.909614086151123,
|
| 780 |
+
"step": 110
|
| 781 |
+
},
|
| 782 |
+
{
|
| 783 |
+
"epoch": 0.43895205140879884,
|
| 784 |
+
"grad_norm": 0.1456097513437271,
|
| 785 |
+
"learning_rate": 0.00015808383233532935,
|
| 786 |
+
"loss": 1.0974111557006836,
|
| 787 |
+
"step": 111
|
| 788 |
+
},
|
| 789 |
+
{
|
| 790 |
+
"epoch": 0.4429065743944637,
|
| 791 |
+
"grad_norm": 0.14371022582054138,
|
| 792 |
+
"learning_rate": 0.00015768463073852296,
|
| 793 |
+
"loss": 1.0514923334121704,
|
| 794 |
+
"step": 112
|
| 795 |
+
},
|
| 796 |
+
{
|
| 797 |
+
"epoch": 0.4468610973801285,
|
| 798 |
+
"grad_norm": 0.14185507595539093,
|
| 799 |
+
"learning_rate": 0.00015728542914171658,
|
| 800 |
+
"loss": 0.856088399887085,
|
| 801 |
+
"step": 113
|
| 802 |
+
},
|
| 803 |
+
{
|
| 804 |
+
"epoch": 0.45081562036579337,
|
| 805 |
+
"grad_norm": 0.14548474550247192,
|
| 806 |
+
"learning_rate": 0.00015688622754491018,
|
| 807 |
+
"loss": 0.9919300079345703,
|
| 808 |
+
"step": 114
|
| 809 |
+
},
|
| 810 |
+
{
|
| 811 |
+
"epoch": 0.4547701433514582,
|
| 812 |
+
"grad_norm": 0.1524973064661026,
|
| 813 |
+
"learning_rate": 0.0001564870259481038,
|
| 814 |
+
"loss": 0.8925392031669617,
|
| 815 |
+
"step": 115
|
| 816 |
+
},
|
| 817 |
+
{
|
| 818 |
+
"epoch": 0.4587246663371231,
|
| 819 |
+
"grad_norm": 0.14767196774482727,
|
| 820 |
+
"learning_rate": 0.0001560878243512974,
|
| 821 |
+
"loss": 0.8464489579200745,
|
| 822 |
+
"step": 116
|
| 823 |
+
},
|
| 824 |
+
{
|
| 825 |
+
"epoch": 0.46267918932278795,
|
| 826 |
+
"grad_norm": 0.15779703855514526,
|
| 827 |
+
"learning_rate": 0.00015568862275449103,
|
| 828 |
+
"loss": 1.0465806722640991,
|
| 829 |
+
"step": 117
|
| 830 |
+
},
|
| 831 |
+
{
|
| 832 |
+
"epoch": 0.4666337123084528,
|
| 833 |
+
"grad_norm": 0.15219292044639587,
|
| 834 |
+
"learning_rate": 0.00015528942115768462,
|
| 835 |
+
"loss": 0.9196197986602783,
|
| 836 |
+
"step": 118
|
| 837 |
+
},
|
| 838 |
+
{
|
| 839 |
+
"epoch": 0.47058823529411764,
|
| 840 |
+
"grad_norm": 0.15147675573825836,
|
| 841 |
+
"learning_rate": 0.00015489021956087827,
|
| 842 |
+
"loss": 0.9819110631942749,
|
| 843 |
+
"step": 119
|
| 844 |
+
},
|
| 845 |
+
{
|
| 846 |
+
"epoch": 0.4745427582797825,
|
| 847 |
+
"grad_norm": 0.1430458277463913,
|
| 848 |
+
"learning_rate": 0.00015449101796407186,
|
| 849 |
+
"loss": 0.9948022365570068,
|
| 850 |
+
"step": 120
|
| 851 |
+
},
|
| 852 |
+
{
|
| 853 |
+
"epoch": 0.4784972812654474,
|
| 854 |
+
"grad_norm": 0.15081898868083954,
|
| 855 |
+
"learning_rate": 0.00015409181636726548,
|
| 856 |
+
"loss": 0.9605839252471924,
|
| 857 |
+
"step": 121
|
| 858 |
+
},
|
| 859 |
+
{
|
| 860 |
+
"epoch": 0.4824518042511122,
|
| 861 |
+
"grad_norm": 0.1469513326883316,
|
| 862 |
+
"learning_rate": 0.0001536926147704591,
|
| 863 |
+
"loss": 0.876885175704956,
|
| 864 |
+
"step": 122
|
| 865 |
+
},
|
| 866 |
+
{
|
| 867 |
+
"epoch": 0.48640632723677707,
|
| 868 |
+
"grad_norm": 0.1487644910812378,
|
| 869 |
+
"learning_rate": 0.00015329341317365272,
|
| 870 |
+
"loss": 0.8893944025039673,
|
| 871 |
+
"step": 123
|
| 872 |
+
},
|
| 873 |
+
{
|
| 874 |
+
"epoch": 0.4903608502224419,
|
| 875 |
+
"grad_norm": 0.1546425074338913,
|
| 876 |
+
"learning_rate": 0.0001528942115768463,
|
| 877 |
+
"loss": 0.8940907120704651,
|
| 878 |
+
"step": 124
|
| 879 |
+
},
|
| 880 |
+
{
|
| 881 |
+
"epoch": 0.49431537320810676,
|
| 882 |
+
"grad_norm": 0.15078657865524292,
|
| 883 |
+
"learning_rate": 0.00015249500998003993,
|
| 884 |
+
"loss": 1.0783774852752686,
|
| 885 |
+
"step": 125
|
| 886 |
+
},
|
| 887 |
+
{
|
| 888 |
+
"epoch": 0.4982698961937716,
|
| 889 |
+
"grad_norm": 0.1505594104528427,
|
| 890 |
+
"learning_rate": 0.00015209580838323355,
|
| 891 |
+
"loss": 0.9070993661880493,
|
| 892 |
+
"step": 126
|
| 893 |
+
},
|
| 894 |
+
{
|
| 895 |
+
"epoch": 0.5022244191794365,
|
| 896 |
+
"grad_norm": 0.15631473064422607,
|
| 897 |
+
"learning_rate": 0.00015169660678642714,
|
| 898 |
+
"loss": 0.9393852353096008,
|
| 899 |
+
"step": 127
|
| 900 |
+
},
|
| 901 |
+
{
|
| 902 |
+
"epoch": 0.5061789421651013,
|
| 903 |
+
"grad_norm": 0.1533495932817459,
|
| 904 |
+
"learning_rate": 0.00015129740518962076,
|
| 905 |
+
"loss": 1.0298182964324951,
|
| 906 |
+
"step": 128
|
| 907 |
+
},
|
| 908 |
+
{
|
| 909 |
+
"epoch": 0.5101334651507662,
|
| 910 |
+
"grad_norm": 0.15687382221221924,
|
| 911 |
+
"learning_rate": 0.00015089820359281438,
|
| 912 |
+
"loss": 0.9548342227935791,
|
| 913 |
+
"step": 129
|
| 914 |
+
},
|
| 915 |
+
{
|
| 916 |
+
"epoch": 0.5140879881364311,
|
| 917 |
+
"grad_norm": 0.13609541952610016,
|
| 918 |
+
"learning_rate": 0.000150499001996008,
|
| 919 |
+
"loss": 0.9333449602127075,
|
| 920 |
+
"step": 130
|
| 921 |
+
},
|
| 922 |
+
{
|
| 923 |
+
"epoch": 0.5180425111220959,
|
| 924 |
+
"grad_norm": 0.1493511199951172,
|
| 925 |
+
"learning_rate": 0.0001500998003992016,
|
| 926 |
+
"loss": 0.9676602482795715,
|
| 927 |
+
"step": 131
|
| 928 |
+
},
|
| 929 |
+
{
|
| 930 |
+
"epoch": 0.5219970341077608,
|
| 931 |
+
"grad_norm": 0.1547534167766571,
|
| 932 |
+
"learning_rate": 0.0001497005988023952,
|
| 933 |
+
"loss": 1.0927131175994873,
|
| 934 |
+
"step": 132
|
| 935 |
+
},
|
| 936 |
+
{
|
| 937 |
+
"epoch": 0.5259515570934256,
|
| 938 |
+
"grad_norm": 0.155633345246315,
|
| 939 |
+
"learning_rate": 0.00014930139720558883,
|
| 940 |
+
"loss": 0.762495219707489,
|
| 941 |
+
"step": 133
|
| 942 |
+
},
|
| 943 |
+
{
|
| 944 |
+
"epoch": 0.5299060800790905,
|
| 945 |
+
"grad_norm": 0.1558597981929779,
|
| 946 |
+
"learning_rate": 0.00014890219560878245,
|
| 947 |
+
"loss": 0.957042396068573,
|
| 948 |
+
"step": 134
|
| 949 |
+
},
|
| 950 |
+
{
|
| 951 |
+
"epoch": 0.5338606030647554,
|
| 952 |
+
"grad_norm": 0.1418825089931488,
|
| 953 |
+
"learning_rate": 0.00014850299401197604,
|
| 954 |
+
"loss": 1.0327900648117065,
|
| 955 |
+
"step": 135
|
| 956 |
+
},
|
| 957 |
+
{
|
| 958 |
+
"epoch": 0.5378151260504201,
|
| 959 |
+
"grad_norm": 0.1431950032711029,
|
| 960 |
+
"learning_rate": 0.00014810379241516968,
|
| 961 |
+
"loss": 0.9374012351036072,
|
| 962 |
+
"step": 136
|
| 963 |
+
},
|
| 964 |
+
{
|
| 965 |
+
"epoch": 0.541769649036085,
|
| 966 |
+
"grad_norm": 0.16276349127292633,
|
| 967 |
+
"learning_rate": 0.00014770459081836328,
|
| 968 |
+
"loss": 0.9466843605041504,
|
| 969 |
+
"step": 137
|
| 970 |
+
},
|
| 971 |
+
{
|
| 972 |
+
"epoch": 0.5457241720217498,
|
| 973 |
+
"grad_norm": 0.14857441186904907,
|
| 974 |
+
"learning_rate": 0.0001473053892215569,
|
| 975 |
+
"loss": 0.9261118769645691,
|
| 976 |
+
"step": 138
|
| 977 |
+
},
|
| 978 |
+
{
|
| 979 |
+
"epoch": 0.5496786950074147,
|
| 980 |
+
"grad_norm": 0.16782596707344055,
|
| 981 |
+
"learning_rate": 0.0001469061876247505,
|
| 982 |
+
"loss": 1.0146270990371704,
|
| 983 |
+
"step": 139
|
| 984 |
+
},
|
| 985 |
+
{
|
| 986 |
+
"epoch": 0.5536332179930796,
|
| 987 |
+
"grad_norm": 0.15024535357952118,
|
| 988 |
+
"learning_rate": 0.00014650698602794413,
|
| 989 |
+
"loss": 0.8856394290924072,
|
| 990 |
+
"step": 140
|
| 991 |
+
},
|
| 992 |
+
{
|
| 993 |
+
"epoch": 0.5575877409787444,
|
| 994 |
+
"grad_norm": 0.1513880342245102,
|
| 995 |
+
"learning_rate": 0.00014610778443113772,
|
| 996 |
+
"loss": 1.0763099193572998,
|
| 997 |
+
"step": 141
|
| 998 |
+
},
|
| 999 |
+
{
|
| 1000 |
+
"epoch": 0.5615422639644093,
|
| 1001 |
+
"grad_norm": 0.14922404289245605,
|
| 1002 |
+
"learning_rate": 0.00014570858283433134,
|
| 1003 |
+
"loss": 1.0585473775863647,
|
| 1004 |
+
"step": 142
|
| 1005 |
+
},
|
| 1006 |
+
{
|
| 1007 |
+
"epoch": 0.5654967869500741,
|
| 1008 |
+
"grad_norm": 0.14560917019844055,
|
| 1009 |
+
"learning_rate": 0.00014530938123752496,
|
| 1010 |
+
"loss": 0.8189446330070496,
|
| 1011 |
+
"step": 143
|
| 1012 |
+
},
|
| 1013 |
+
{
|
| 1014 |
+
"epoch": 0.569451309935739,
|
| 1015 |
+
"grad_norm": 0.1498928666114807,
|
| 1016 |
+
"learning_rate": 0.00014491017964071858,
|
| 1017 |
+
"loss": 0.9366933703422546,
|
| 1018 |
+
"step": 144
|
| 1019 |
+
},
|
| 1020 |
+
{
|
| 1021 |
+
"epoch": 0.5734058329214039,
|
| 1022 |
+
"grad_norm": 0.14731937646865845,
|
| 1023 |
+
"learning_rate": 0.00014451097804391217,
|
| 1024 |
+
"loss": 0.978458821773529,
|
| 1025 |
+
"step": 145
|
| 1026 |
+
},
|
| 1027 |
+
{
|
| 1028 |
+
"epoch": 0.5773603559070687,
|
| 1029 |
+
"grad_norm": 0.14156509935855865,
|
| 1030 |
+
"learning_rate": 0.00014411177644710582,
|
| 1031 |
+
"loss": 0.9639441967010498,
|
| 1032 |
+
"step": 146
|
| 1033 |
+
},
|
| 1034 |
+
{
|
| 1035 |
+
"epoch": 0.5813148788927336,
|
| 1036 |
+
"grad_norm": 0.17466521263122559,
|
| 1037 |
+
"learning_rate": 0.0001437125748502994,
|
| 1038 |
+
"loss": 0.9246646165847778,
|
| 1039 |
+
"step": 147
|
| 1040 |
+
},
|
| 1041 |
+
{
|
| 1042 |
+
"epoch": 0.5852694018783984,
|
| 1043 |
+
"grad_norm": 0.15222910046577454,
|
| 1044 |
+
"learning_rate": 0.000143313373253493,
|
| 1045 |
+
"loss": 0.9557418823242188,
|
| 1046 |
+
"step": 148
|
| 1047 |
+
},
|
| 1048 |
+
{
|
| 1049 |
+
"epoch": 0.5892239248640633,
|
| 1050 |
+
"grad_norm": 0.14972400665283203,
|
| 1051 |
+
"learning_rate": 0.00014291417165668665,
|
| 1052 |
+
"loss": 0.8998520970344543,
|
| 1053 |
+
"step": 149
|
| 1054 |
+
},
|
| 1055 |
+
{
|
| 1056 |
+
"epoch": 0.5931784478497282,
|
| 1057 |
+
"grad_norm": 0.16542503237724304,
|
| 1058 |
+
"learning_rate": 0.00014251497005988024,
|
| 1059 |
+
"loss": 0.8584086298942566,
|
| 1060 |
+
"step": 150
|
| 1061 |
+
},
|
| 1062 |
+
{
|
| 1063 |
+
"epoch": 0.597132970835393,
|
| 1064 |
+
"grad_norm": 0.1549108326435089,
|
| 1065 |
+
"learning_rate": 0.00014211576846307386,
|
| 1066 |
+
"loss": 0.9696409702301025,
|
| 1067 |
+
"step": 151
|
| 1068 |
+
},
|
| 1069 |
+
{
|
| 1070 |
+
"epoch": 0.6010874938210579,
|
| 1071 |
+
"grad_norm": 0.14827242493629456,
|
| 1072 |
+
"learning_rate": 0.00014171656686626745,
|
| 1073 |
+
"loss": 0.8613938093185425,
|
| 1074 |
+
"step": 152
|
| 1075 |
+
},
|
| 1076 |
+
{
|
| 1077 |
+
"epoch": 0.6050420168067226,
|
| 1078 |
+
"grad_norm": 0.15346549451351166,
|
| 1079 |
+
"learning_rate": 0.0001413173652694611,
|
| 1080 |
+
"loss": 0.7989118695259094,
|
| 1081 |
+
"step": 153
|
| 1082 |
+
},
|
| 1083 |
+
{
|
| 1084 |
+
"epoch": 0.6089965397923875,
|
| 1085 |
+
"grad_norm": 0.15291966497898102,
|
| 1086 |
+
"learning_rate": 0.0001409181636726547,
|
| 1087 |
+
"loss": 1.0761021375656128,
|
| 1088 |
+
"step": 154
|
| 1089 |
+
},
|
| 1090 |
+
{
|
| 1091 |
+
"epoch": 0.6129510627780524,
|
| 1092 |
+
"grad_norm": 0.13826234638690948,
|
| 1093 |
+
"learning_rate": 0.0001405189620758483,
|
| 1094 |
+
"loss": 0.9109746217727661,
|
| 1095 |
+
"step": 155
|
| 1096 |
+
},
|
| 1097 |
+
{
|
| 1098 |
+
"epoch": 0.6169055857637172,
|
| 1099 |
+
"grad_norm": 0.14913445711135864,
|
| 1100 |
+
"learning_rate": 0.00014011976047904193,
|
| 1101 |
+
"loss": 1.0017129182815552,
|
| 1102 |
+
"step": 156
|
| 1103 |
+
},
|
| 1104 |
+
{
|
| 1105 |
+
"epoch": 0.6208601087493821,
|
| 1106 |
+
"grad_norm": 0.1526130735874176,
|
| 1107 |
+
"learning_rate": 0.00013972055888223555,
|
| 1108 |
+
"loss": 0.8955498933792114,
|
| 1109 |
+
"step": 157
|
| 1110 |
+
},
|
| 1111 |
+
{
|
| 1112 |
+
"epoch": 0.6248146317350469,
|
| 1113 |
+
"grad_norm": 0.15382827818393707,
|
| 1114 |
+
"learning_rate": 0.00013932135728542914,
|
| 1115 |
+
"loss": 1.1069719791412354,
|
| 1116 |
+
"step": 158
|
| 1117 |
+
},
|
| 1118 |
+
{
|
| 1119 |
+
"epoch": 0.6287691547207118,
|
| 1120 |
+
"grad_norm": 0.14784258604049683,
|
| 1121 |
+
"learning_rate": 0.00013892215568862276,
|
| 1122 |
+
"loss": 0.9784612655639648,
|
| 1123 |
+
"step": 159
|
| 1124 |
+
},
|
| 1125 |
+
{
|
| 1126 |
+
"epoch": 0.6327236777063767,
|
| 1127 |
+
"grad_norm": 0.15217405557632446,
|
| 1128 |
+
"learning_rate": 0.00013852295409181638,
|
| 1129 |
+
"loss": 0.9924787878990173,
|
| 1130 |
+
"step": 160
|
| 1131 |
+
},
|
| 1132 |
+
{
|
| 1133 |
+
"epoch": 0.6366782006920415,
|
| 1134 |
+
"grad_norm": 0.17497238516807556,
|
| 1135 |
+
"learning_rate": 0.00013812375249501,
|
| 1136 |
+
"loss": 0.8755843043327332,
|
| 1137 |
+
"step": 161
|
| 1138 |
+
},
|
| 1139 |
+
{
|
| 1140 |
+
"epoch": 0.6406327236777064,
|
| 1141 |
+
"grad_norm": 0.15077446401119232,
|
| 1142 |
+
"learning_rate": 0.00013772455089820359,
|
| 1143 |
+
"loss": 0.9884225726127625,
|
| 1144 |
+
"step": 162
|
| 1145 |
+
},
|
| 1146 |
+
{
|
| 1147 |
+
"epoch": 0.6445872466633712,
|
| 1148 |
+
"grad_norm": 0.1486707627773285,
|
| 1149 |
+
"learning_rate": 0.00013732534930139723,
|
| 1150 |
+
"loss": 1.0523698329925537,
|
| 1151 |
+
"step": 163
|
| 1152 |
+
},
|
| 1153 |
+
{
|
| 1154 |
+
"epoch": 0.6485417696490361,
|
| 1155 |
+
"grad_norm": 0.13988643884658813,
|
| 1156 |
+
"learning_rate": 0.00013692614770459082,
|
| 1157 |
+
"loss": 0.9002912640571594,
|
| 1158 |
+
"step": 164
|
| 1159 |
+
},
|
| 1160 |
+
{
|
| 1161 |
+
"epoch": 0.652496292634701,
|
| 1162 |
+
"grad_norm": 0.15143455564975739,
|
| 1163 |
+
"learning_rate": 0.00013652694610778444,
|
| 1164 |
+
"loss": 0.9852843284606934,
|
| 1165 |
+
"step": 165
|
| 1166 |
+
},
|
| 1167 |
+
{
|
| 1168 |
+
"epoch": 0.6564508156203658,
|
| 1169 |
+
"grad_norm": 0.15765251219272614,
|
| 1170 |
+
"learning_rate": 0.00013612774451097806,
|
| 1171 |
+
"loss": 0.9757329225540161,
|
| 1172 |
+
"step": 166
|
| 1173 |
+
},
|
| 1174 |
+
{
|
| 1175 |
+
"epoch": 0.6604053386060307,
|
| 1176 |
+
"grad_norm": 0.15192811191082,
|
| 1177 |
+
"learning_rate": 0.00013572854291417165,
|
| 1178 |
+
"loss": 0.9793543815612793,
|
| 1179 |
+
"step": 167
|
| 1180 |
+
},
|
| 1181 |
+
{
|
| 1182 |
+
"epoch": 0.6643598615916955,
|
| 1183 |
+
"grad_norm": 0.1370113044977188,
|
| 1184 |
+
"learning_rate": 0.00013532934131736527,
|
| 1185 |
+
"loss": 0.8899397253990173,
|
| 1186 |
+
"step": 168
|
| 1187 |
+
},
|
| 1188 |
+
{
|
| 1189 |
+
"epoch": 0.6683143845773604,
|
| 1190 |
+
"grad_norm": 0.16300129890441895,
|
| 1191 |
+
"learning_rate": 0.0001349301397205589,
|
| 1192 |
+
"loss": 0.9691711664199829,
|
| 1193 |
+
"step": 169
|
| 1194 |
+
},
|
| 1195 |
+
{
|
| 1196 |
+
"epoch": 0.6722689075630253,
|
| 1197 |
+
"grad_norm": 0.1447598934173584,
|
| 1198 |
+
"learning_rate": 0.0001345309381237525,
|
| 1199 |
+
"loss": 1.0262728929519653,
|
| 1200 |
+
"step": 170
|
| 1201 |
+
},
|
| 1202 |
+
{
|
| 1203 |
+
"epoch": 0.67622343054869,
|
| 1204 |
+
"grad_norm": 0.1426248848438263,
|
| 1205 |
+
"learning_rate": 0.0001341317365269461,
|
| 1206 |
+
"loss": 0.8865811824798584,
|
| 1207 |
+
"step": 171
|
| 1208 |
+
},
|
| 1209 |
+
{
|
| 1210 |
+
"epoch": 0.680177953534355,
|
| 1211 |
+
"grad_norm": 0.1596144288778305,
|
| 1212 |
+
"learning_rate": 0.00013373253493013972,
|
| 1213 |
+
"loss": 1.041471242904663,
|
| 1214 |
+
"step": 172
|
| 1215 |
+
},
|
| 1216 |
+
{
|
| 1217 |
+
"epoch": 0.6841324765200197,
|
| 1218 |
+
"grad_norm": 0.16748711466789246,
|
| 1219 |
+
"learning_rate": 0.00013333333333333334,
|
| 1220 |
+
"loss": 1.060260534286499,
|
| 1221 |
+
"step": 173
|
| 1222 |
+
},
|
| 1223 |
+
{
|
| 1224 |
+
"epoch": 0.6880869995056846,
|
| 1225 |
+
"grad_norm": 0.14779865741729736,
|
| 1226 |
+
"learning_rate": 0.00013293413173652696,
|
| 1227 |
+
"loss": 0.9902845025062561,
|
| 1228 |
+
"step": 174
|
| 1229 |
+
},
|
| 1230 |
+
{
|
| 1231 |
+
"epoch": 0.6920415224913494,
|
| 1232 |
+
"grad_norm": 0.1514303982257843,
|
| 1233 |
+
"learning_rate": 0.00013253493013972055,
|
| 1234 |
+
"loss": 0.9630454778671265,
|
| 1235 |
+
"step": 175
|
| 1236 |
+
},
|
| 1237 |
+
{
|
| 1238 |
+
"epoch": 0.6959960454770143,
|
| 1239 |
+
"grad_norm": 0.16021154820919037,
|
| 1240 |
+
"learning_rate": 0.0001321357285429142,
|
| 1241 |
+
"loss": 0.8938897848129272,
|
| 1242 |
+
"step": 176
|
| 1243 |
+
},
|
| 1244 |
+
{
|
| 1245 |
+
"epoch": 0.6999505684626792,
|
| 1246 |
+
"grad_norm": 0.1604602336883545,
|
| 1247 |
+
"learning_rate": 0.0001317365269461078,
|
| 1248 |
+
"loss": 0.7806445956230164,
|
| 1249 |
+
"step": 177
|
| 1250 |
+
},
|
| 1251 |
+
{
|
| 1252 |
+
"epoch": 0.703905091448344,
|
| 1253 |
+
"grad_norm": 0.14814768731594086,
|
| 1254 |
+
"learning_rate": 0.0001313373253493014,
|
| 1255 |
+
"loss": 0.7236217856407166,
|
| 1256 |
+
"step": 178
|
| 1257 |
+
},
|
| 1258 |
+
{
|
| 1259 |
+
"epoch": 0.7078596144340089,
|
| 1260 |
+
"grad_norm": 0.13816480338573456,
|
| 1261 |
+
"learning_rate": 0.000130938123752495,
|
| 1262 |
+
"loss": 1.059952974319458,
|
| 1263 |
+
"step": 179
|
| 1264 |
+
},
|
| 1265 |
+
{
|
| 1266 |
+
"epoch": 0.7118141374196737,
|
| 1267 |
+
"grad_norm": 0.15320385992527008,
|
| 1268 |
+
"learning_rate": 0.00013053892215568865,
|
| 1269 |
+
"loss": 1.037667989730835,
|
| 1270 |
+
"step": 180
|
| 1271 |
+
},
|
| 1272 |
+
{
|
| 1273 |
+
"epoch": 0.7157686604053386,
|
| 1274 |
+
"grad_norm": 0.14675435423851013,
|
| 1275 |
+
"learning_rate": 0.00013013972055888224,
|
| 1276 |
+
"loss": 1.0143626928329468,
|
| 1277 |
+
"step": 181
|
| 1278 |
+
},
|
| 1279 |
+
{
|
| 1280 |
+
"epoch": 0.7197231833910035,
|
| 1281 |
+
"grad_norm": 0.13777698576450348,
|
| 1282 |
+
"learning_rate": 0.00012974051896207586,
|
| 1283 |
+
"loss": 0.8841183185577393,
|
| 1284 |
+
"step": 182
|
| 1285 |
+
},
|
| 1286 |
+
{
|
| 1287 |
+
"epoch": 0.7236777063766683,
|
| 1288 |
+
"grad_norm": 0.14622290432453156,
|
| 1289 |
+
"learning_rate": 0.00012934131736526947,
|
| 1290 |
+
"loss": 0.9912742376327515,
|
| 1291 |
+
"step": 183
|
| 1292 |
+
},
|
| 1293 |
+
{
|
| 1294 |
+
"epoch": 0.7276322293623332,
|
| 1295 |
+
"grad_norm": 0.14383630454540253,
|
| 1296 |
+
"learning_rate": 0.0001289421157684631,
|
| 1297 |
+
"loss": 1.031080722808838,
|
| 1298 |
+
"step": 184
|
| 1299 |
+
},
|
| 1300 |
+
{
|
| 1301 |
+
"epoch": 0.731586752347998,
|
| 1302 |
+
"grad_norm": 0.14537449181079865,
|
| 1303 |
+
"learning_rate": 0.00012854291417165669,
|
| 1304 |
+
"loss": 0.9928056001663208,
|
| 1305 |
+
"step": 185
|
| 1306 |
+
},
|
| 1307 |
+
{
|
| 1308 |
+
"epoch": 0.7355412753336629,
|
| 1309 |
+
"grad_norm": 0.1553298830986023,
|
| 1310 |
+
"learning_rate": 0.0001281437125748503,
|
| 1311 |
+
"loss": 1.0729012489318848,
|
| 1312 |
+
"step": 186
|
| 1313 |
+
},
|
| 1314 |
+
{
|
| 1315 |
+
"epoch": 0.7394957983193278,
|
| 1316 |
+
"grad_norm": 0.15434440970420837,
|
| 1317 |
+
"learning_rate": 0.00012774451097804392,
|
| 1318 |
+
"loss": 0.9299134016036987,
|
| 1319 |
+
"step": 187
|
| 1320 |
+
},
|
| 1321 |
+
{
|
| 1322 |
+
"epoch": 0.7434503213049926,
|
| 1323 |
+
"grad_norm": 0.1517597734928131,
|
| 1324 |
+
"learning_rate": 0.00012734530938123752,
|
| 1325 |
+
"loss": 0.8981556296348572,
|
| 1326 |
+
"step": 188
|
| 1327 |
+
},
|
| 1328 |
+
{
|
| 1329 |
+
"epoch": 0.7474048442906575,
|
| 1330 |
+
"grad_norm": 0.14408698678016663,
|
| 1331 |
+
"learning_rate": 0.00012694610778443113,
|
| 1332 |
+
"loss": 0.7834718227386475,
|
| 1333 |
+
"step": 189
|
| 1334 |
+
},
|
| 1335 |
+
{
|
| 1336 |
+
"epoch": 0.7513593672763222,
|
| 1337 |
+
"grad_norm": 0.1524432897567749,
|
| 1338 |
+
"learning_rate": 0.00012654690618762475,
|
| 1339 |
+
"loss": 0.9502565264701843,
|
| 1340 |
+
"step": 190
|
| 1341 |
+
},
|
| 1342 |
+
{
|
| 1343 |
+
"epoch": 0.7553138902619871,
|
| 1344 |
+
"grad_norm": 0.16056875884532928,
|
| 1345 |
+
"learning_rate": 0.00012614770459081837,
|
| 1346 |
+
"loss": 1.0030090808868408,
|
| 1347 |
+
"step": 191
|
| 1348 |
+
},
|
| 1349 |
+
{
|
| 1350 |
+
"epoch": 0.759268413247652,
|
| 1351 |
+
"grad_norm": 0.14580710232257843,
|
| 1352 |
+
"learning_rate": 0.00012574850299401196,
|
| 1353 |
+
"loss": 0.937939465045929,
|
| 1354 |
+
"step": 192
|
| 1355 |
+
},
|
| 1356 |
+
{
|
| 1357 |
+
"epoch": 0.7632229362333168,
|
| 1358 |
+
"grad_norm": 0.14517466723918915,
|
| 1359 |
+
"learning_rate": 0.0001253493013972056,
|
| 1360 |
+
"loss": 1.1117490530014038,
|
| 1361 |
+
"step": 193
|
| 1362 |
+
},
|
| 1363 |
+
{
|
| 1364 |
+
"epoch": 0.7671774592189817,
|
| 1365 |
+
"grad_norm": 0.15558141469955444,
|
| 1366 |
+
"learning_rate": 0.0001249500998003992,
|
| 1367 |
+
"loss": 0.8213934898376465,
|
| 1368 |
+
"step": 194
|
| 1369 |
+
},
|
| 1370 |
+
{
|
| 1371 |
+
"epoch": 0.7711319822046465,
|
| 1372 |
+
"grad_norm": 0.14642053842544556,
|
| 1373 |
+
"learning_rate": 0.00012455089820359282,
|
| 1374 |
+
"loss": 0.8714714646339417,
|
| 1375 |
+
"step": 195
|
| 1376 |
+
},
|
| 1377 |
+
{
|
| 1378 |
+
"epoch": 0.7750865051903114,
|
| 1379 |
+
"grad_norm": 0.15025141835212708,
|
| 1380 |
+
"learning_rate": 0.00012415169660678644,
|
| 1381 |
+
"loss": 1.0590744018554688,
|
| 1382 |
+
"step": 196
|
| 1383 |
+
},
|
| 1384 |
+
{
|
| 1385 |
+
"epoch": 0.7790410281759763,
|
| 1386 |
+
"grad_norm": 0.1296977996826172,
|
| 1387 |
+
"learning_rate": 0.00012375249500998006,
|
| 1388 |
+
"loss": 1.0535149574279785,
|
| 1389 |
+
"step": 197
|
| 1390 |
+
},
|
| 1391 |
+
{
|
| 1392 |
+
"epoch": 0.7829955511616411,
|
| 1393 |
+
"grad_norm": 0.15777179598808289,
|
| 1394 |
+
"learning_rate": 0.00012335329341317365,
|
| 1395 |
+
"loss": 1.0486925840377808,
|
| 1396 |
+
"step": 198
|
| 1397 |
+
},
|
| 1398 |
+
{
|
| 1399 |
+
"epoch": 0.786950074147306,
|
| 1400 |
+
"grad_norm": 0.13769429922103882,
|
| 1401 |
+
"learning_rate": 0.00012295409181636727,
|
| 1402 |
+
"loss": 0.9360702633857727,
|
| 1403 |
+
"step": 199
|
| 1404 |
+
},
|
| 1405 |
+
{
|
| 1406 |
+
"epoch": 0.7909045971329708,
|
| 1407 |
+
"grad_norm": 0.15268363058567047,
|
| 1408 |
+
"learning_rate": 0.0001225548902195609,
|
| 1409 |
+
"loss": 0.8887878656387329,
|
| 1410 |
+
"step": 200
|
| 1411 |
+
},
|
| 1412 |
+
{
|
| 1413 |
+
"epoch": 0.7948591201186357,
|
| 1414 |
+
"grad_norm": 0.13850852847099304,
|
| 1415 |
+
"learning_rate": 0.0001221556886227545,
|
| 1416 |
+
"loss": 1.083563208580017,
|
| 1417 |
+
"step": 201
|
| 1418 |
+
},
|
| 1419 |
+
{
|
| 1420 |
+
"epoch": 0.7988136431043006,
|
| 1421 |
+
"grad_norm": 0.16249485313892365,
|
| 1422 |
+
"learning_rate": 0.00012175648702594811,
|
| 1423 |
+
"loss": 1.0185102224349976,
|
| 1424 |
+
"step": 202
|
| 1425 |
+
},
|
| 1426 |
+
{
|
| 1427 |
+
"epoch": 0.8027681660899654,
|
| 1428 |
+
"grad_norm": 0.14542728662490845,
|
| 1429 |
+
"learning_rate": 0.00012135728542914173,
|
| 1430 |
+
"loss": 0.985219419002533,
|
| 1431 |
+
"step": 203
|
| 1432 |
+
},
|
| 1433 |
+
{
|
| 1434 |
+
"epoch": 0.8067226890756303,
|
| 1435 |
+
"grad_norm": 0.1401057094335556,
|
| 1436 |
+
"learning_rate": 0.00012095808383233534,
|
| 1437 |
+
"loss": 0.9981472492218018,
|
| 1438 |
+
"step": 204
|
| 1439 |
+
},
|
| 1440 |
+
{
|
| 1441 |
+
"epoch": 0.8106772120612951,
|
| 1442 |
+
"grad_norm": 0.16163145005702972,
|
| 1443 |
+
"learning_rate": 0.00012055888223552896,
|
| 1444 |
+
"loss": 0.9624532461166382,
|
| 1445 |
+
"step": 205
|
| 1446 |
+
},
|
| 1447 |
+
{
|
| 1448 |
+
"epoch": 0.81463173504696,
|
| 1449 |
+
"grad_norm": 0.1544468253850937,
|
| 1450 |
+
"learning_rate": 0.00012015968063872256,
|
| 1451 |
+
"loss": 0.9114558696746826,
|
| 1452 |
+
"step": 206
|
| 1453 |
+
},
|
| 1454 |
+
{
|
| 1455 |
+
"epoch": 0.8185862580326249,
|
| 1456 |
+
"grad_norm": 0.15567342936992645,
|
| 1457 |
+
"learning_rate": 0.00011976047904191617,
|
| 1458 |
+
"loss": 0.9480348825454712,
|
| 1459 |
+
"step": 207
|
| 1460 |
+
},
|
| 1461 |
+
{
|
| 1462 |
+
"epoch": 0.8225407810182896,
|
| 1463 |
+
"grad_norm": 0.15945857763290405,
|
| 1464 |
+
"learning_rate": 0.00011936127744510979,
|
| 1465 |
+
"loss": 0.9799912571907043,
|
| 1466 |
+
"step": 208
|
| 1467 |
+
},
|
| 1468 |
+
{
|
| 1469 |
+
"epoch": 0.8264953040039545,
|
| 1470 |
+
"grad_norm": 0.15980277955532074,
|
| 1471 |
+
"learning_rate": 0.00011896207584830339,
|
| 1472 |
+
"loss": 1.0797210931777954,
|
| 1473 |
+
"step": 209
|
| 1474 |
+
},
|
| 1475 |
+
{
|
| 1476 |
+
"epoch": 0.8304498269896193,
|
| 1477 |
+
"grad_norm": 0.14261196553707123,
|
| 1478 |
+
"learning_rate": 0.00011856287425149701,
|
| 1479 |
+
"loss": 1.0038321018218994,
|
| 1480 |
+
"step": 210
|
| 1481 |
+
},
|
| 1482 |
+
{
|
| 1483 |
+
"epoch": 0.8344043499752842,
|
| 1484 |
+
"grad_norm": 0.15023842453956604,
|
| 1485 |
+
"learning_rate": 0.00011816367265469061,
|
| 1486 |
+
"loss": 1.0017417669296265,
|
| 1487 |
+
"step": 211
|
| 1488 |
+
},
|
| 1489 |
+
{
|
| 1490 |
+
"epoch": 0.8383588729609491,
|
| 1491 |
+
"grad_norm": 0.14774499833583832,
|
| 1492 |
+
"learning_rate": 0.00011776447105788423,
|
| 1493 |
+
"loss": 1.056884765625,
|
| 1494 |
+
"step": 212
|
| 1495 |
+
},
|
| 1496 |
+
{
|
| 1497 |
+
"epoch": 0.8423133959466139,
|
| 1498 |
+
"grad_norm": 0.1311042457818985,
|
| 1499 |
+
"learning_rate": 0.00011736526946107784,
|
| 1500 |
+
"loss": 1.015238881111145,
|
| 1501 |
+
"step": 213
|
| 1502 |
+
},
|
| 1503 |
+
{
|
| 1504 |
+
"epoch": 0.8462679189322788,
|
| 1505 |
+
"grad_norm": 0.14513953030109406,
|
| 1506 |
+
"learning_rate": 0.00011696606786427147,
|
| 1507 |
+
"loss": 1.0354350805282593,
|
| 1508 |
+
"step": 214
|
| 1509 |
+
},
|
| 1510 |
+
{
|
| 1511 |
+
"epoch": 0.8502224419179436,
|
| 1512 |
+
"grad_norm": 0.1631115972995758,
|
| 1513 |
+
"learning_rate": 0.00011656686626746506,
|
| 1514 |
+
"loss": 0.9117772579193115,
|
| 1515 |
+
"step": 215
|
| 1516 |
+
},
|
| 1517 |
+
{
|
| 1518 |
+
"epoch": 0.8541769649036085,
|
| 1519 |
+
"grad_norm": 0.16442656517028809,
|
| 1520 |
+
"learning_rate": 0.0001161676646706587,
|
| 1521 |
+
"loss": 1.0395290851593018,
|
| 1522 |
+
"step": 216
|
| 1523 |
+
},
|
| 1524 |
+
{
|
| 1525 |
+
"epoch": 0.8581314878892734,
|
| 1526 |
+
"grad_norm": 0.14986057579517365,
|
| 1527 |
+
"learning_rate": 0.0001157684630738523,
|
| 1528 |
+
"loss": 0.9993502497673035,
|
| 1529 |
+
"step": 217
|
| 1530 |
+
},
|
| 1531 |
+
{
|
| 1532 |
+
"epoch": 0.8620860108749382,
|
| 1533 |
+
"grad_norm": 0.16269713640213013,
|
| 1534 |
+
"learning_rate": 0.00011536926147704592,
|
| 1535 |
+
"loss": 0.8923558592796326,
|
| 1536 |
+
"step": 218
|
| 1537 |
+
},
|
| 1538 |
+
{
|
| 1539 |
+
"epoch": 0.8660405338606031,
|
| 1540 |
+
"grad_norm": 0.14905333518981934,
|
| 1541 |
+
"learning_rate": 0.00011497005988023953,
|
| 1542 |
+
"loss": 0.846024751663208,
|
| 1543 |
+
"step": 219
|
| 1544 |
+
},
|
| 1545 |
+
{
|
| 1546 |
+
"epoch": 0.8699950568462679,
|
| 1547 |
+
"grad_norm": 0.14860469102859497,
|
| 1548 |
+
"learning_rate": 0.00011457085828343314,
|
| 1549 |
+
"loss": 0.933295488357544,
|
| 1550 |
+
"step": 220
|
| 1551 |
+
},
|
| 1552 |
+
{
|
| 1553 |
+
"epoch": 0.8739495798319328,
|
| 1554 |
+
"grad_norm": 0.15000364184379578,
|
| 1555 |
+
"learning_rate": 0.00011417165668662675,
|
| 1556 |
+
"loss": 0.9875235557556152,
|
| 1557 |
+
"step": 221
|
| 1558 |
+
},
|
| 1559 |
+
{
|
| 1560 |
+
"epoch": 0.8779041028175977,
|
| 1561 |
+
"grad_norm": 0.15674816071987152,
|
| 1562 |
+
"learning_rate": 0.00011377245508982037,
|
| 1563 |
+
"loss": 0.8917112946510315,
|
| 1564 |
+
"step": 222
|
| 1565 |
+
},
|
| 1566 |
+
{
|
| 1567 |
+
"epoch": 0.8818586258032625,
|
| 1568 |
+
"grad_norm": 0.16109170019626617,
|
| 1569 |
+
"learning_rate": 0.00011337325349301397,
|
| 1570 |
+
"loss": 0.9784120321273804,
|
| 1571 |
+
"step": 223
|
| 1572 |
+
},
|
| 1573 |
+
{
|
| 1574 |
+
"epoch": 0.8858131487889274,
|
| 1575 |
+
"grad_norm": 0.15252313017845154,
|
| 1576 |
+
"learning_rate": 0.0001129740518962076,
|
| 1577 |
+
"loss": 0.7924797534942627,
|
| 1578 |
+
"step": 224
|
| 1579 |
+
},
|
| 1580 |
+
{
|
| 1581 |
+
"epoch": 0.8897676717745922,
|
| 1582 |
+
"grad_norm": 0.15269458293914795,
|
| 1583 |
+
"learning_rate": 0.0001125748502994012,
|
| 1584 |
+
"loss": 0.9920112490653992,
|
| 1585 |
+
"step": 225
|
| 1586 |
+
},
|
| 1587 |
+
{
|
| 1588 |
+
"epoch": 0.893722194760257,
|
| 1589 |
+
"grad_norm": 0.17212046682834625,
|
| 1590 |
+
"learning_rate": 0.00011217564870259483,
|
| 1591 |
+
"loss": 0.9893192052841187,
|
| 1592 |
+
"step": 226
|
| 1593 |
+
},
|
| 1594 |
+
{
|
| 1595 |
+
"epoch": 0.897676717745922,
|
| 1596 |
+
"grad_norm": 0.15529413521289825,
|
| 1597 |
+
"learning_rate": 0.00011177644710578842,
|
| 1598 |
+
"loss": 1.0025854110717773,
|
| 1599 |
+
"step": 227
|
| 1600 |
+
},
|
| 1601 |
+
{
|
| 1602 |
+
"epoch": 0.9016312407315867,
|
| 1603 |
+
"grad_norm": 0.17252163589000702,
|
| 1604 |
+
"learning_rate": 0.00011137724550898203,
|
| 1605 |
+
"loss": 0.8992288112640381,
|
| 1606 |
+
"step": 228
|
| 1607 |
+
},
|
| 1608 |
+
{
|
| 1609 |
+
"epoch": 0.9055857637172516,
|
| 1610 |
+
"grad_norm": 0.1746063530445099,
|
| 1611 |
+
"learning_rate": 0.00011097804391217566,
|
| 1612 |
+
"loss": 0.8158121705055237,
|
| 1613 |
+
"step": 229
|
| 1614 |
+
},
|
| 1615 |
+
{
|
| 1616 |
+
"epoch": 0.9095402867029164,
|
| 1617 |
+
"grad_norm": 0.13989895582199097,
|
| 1618 |
+
"learning_rate": 0.00011057884231536925,
|
| 1619 |
+
"loss": 1.0338526964187622,
|
| 1620 |
+
"step": 230
|
| 1621 |
+
},
|
| 1622 |
+
{
|
| 1623 |
+
"epoch": 0.9134948096885813,
|
| 1624 |
+
"grad_norm": 0.16055206954479218,
|
| 1625 |
+
"learning_rate": 0.00011017964071856288,
|
| 1626 |
+
"loss": 0.9847093224525452,
|
| 1627 |
+
"step": 231
|
| 1628 |
+
},
|
| 1629 |
+
{
|
| 1630 |
+
"epoch": 0.9174493326742462,
|
| 1631 |
+
"grad_norm": 0.16400660574436188,
|
| 1632 |
+
"learning_rate": 0.00010978043912175649,
|
| 1633 |
+
"loss": 0.9464921951293945,
|
| 1634 |
+
"step": 232
|
| 1635 |
+
},
|
| 1636 |
+
{
|
| 1637 |
+
"epoch": 0.921403855659911,
|
| 1638 |
+
"grad_norm": 0.14987632632255554,
|
| 1639 |
+
"learning_rate": 0.00010938123752495011,
|
| 1640 |
+
"loss": 0.8394877910614014,
|
| 1641 |
+
"step": 233
|
| 1642 |
+
},
|
| 1643 |
+
{
|
| 1644 |
+
"epoch": 0.9253583786455759,
|
| 1645 |
+
"grad_norm": 0.16941316425800323,
|
| 1646 |
+
"learning_rate": 0.00010898203592814371,
|
| 1647 |
+
"loss": 0.8612609505653381,
|
| 1648 |
+
"step": 234
|
| 1649 |
+
},
|
| 1650 |
+
{
|
| 1651 |
+
"epoch": 0.9293129016312407,
|
| 1652 |
+
"grad_norm": 0.16219089925289154,
|
| 1653 |
+
"learning_rate": 0.00010858283433133733,
|
| 1654 |
+
"loss": 0.8607385158538818,
|
| 1655 |
+
"step": 235
|
| 1656 |
+
},
|
| 1657 |
+
{
|
| 1658 |
+
"epoch": 0.9332674246169056,
|
| 1659 |
+
"grad_norm": 0.1656905710697174,
|
| 1660 |
+
"learning_rate": 0.00010818363273453094,
|
| 1661 |
+
"loss": 1.0054378509521484,
|
| 1662 |
+
"step": 236
|
| 1663 |
+
},
|
| 1664 |
+
{
|
| 1665 |
+
"epoch": 0.9372219476025705,
|
| 1666 |
+
"grad_norm": 0.15206600725650787,
|
| 1667 |
+
"learning_rate": 0.00010778443113772456,
|
| 1668 |
+
"loss": 1.0277290344238281,
|
| 1669 |
+
"step": 237
|
| 1670 |
+
},
|
| 1671 |
+
{
|
| 1672 |
+
"epoch": 0.9411764705882353,
|
| 1673 |
+
"grad_norm": 0.1560627520084381,
|
| 1674 |
+
"learning_rate": 0.00010738522954091816,
|
| 1675 |
+
"loss": 0.9026745557785034,
|
| 1676 |
+
"step": 238
|
| 1677 |
+
},
|
| 1678 |
+
{
|
| 1679 |
+
"epoch": 0.9451309935739002,
|
| 1680 |
+
"grad_norm": 0.16465218365192413,
|
| 1681 |
+
"learning_rate": 0.00010698602794411178,
|
| 1682 |
+
"loss": 0.9656213521957397,
|
| 1683 |
+
"step": 239
|
| 1684 |
+
},
|
| 1685 |
+
{
|
| 1686 |
+
"epoch": 0.949085516559565,
|
| 1687 |
+
"grad_norm": 0.16291505098342896,
|
| 1688 |
+
"learning_rate": 0.00010658682634730539,
|
| 1689 |
+
"loss": 0.940356433391571,
|
| 1690 |
+
"step": 240
|
| 1691 |
+
},
|
| 1692 |
+
{
|
| 1693 |
+
"epoch": 0.9530400395452299,
|
| 1694 |
+
"grad_norm": 0.16686119139194489,
|
| 1695 |
+
"learning_rate": 0.00010618762475049902,
|
| 1696 |
+
"loss": 0.8976022601127625,
|
| 1697 |
+
"step": 241
|
| 1698 |
+
},
|
| 1699 |
+
{
|
| 1700 |
+
"epoch": 0.9569945625308948,
|
| 1701 |
+
"grad_norm": 0.1607900708913803,
|
| 1702 |
+
"learning_rate": 0.00010578842315369261,
|
| 1703 |
+
"loss": 0.9073758125305176,
|
| 1704 |
+
"step": 242
|
| 1705 |
+
},
|
| 1706 |
+
{
|
| 1707 |
+
"epoch": 0.9609490855165596,
|
| 1708 |
+
"grad_norm": 0.15350642800331116,
|
| 1709 |
+
"learning_rate": 0.00010538922155688624,
|
| 1710 |
+
"loss": 0.8798397779464722,
|
| 1711 |
+
"step": 243
|
| 1712 |
+
},
|
| 1713 |
+
{
|
| 1714 |
+
"epoch": 0.9649036085022245,
|
| 1715 |
+
"grad_norm": 0.13445277512073517,
|
| 1716 |
+
"learning_rate": 0.00010499001996007985,
|
| 1717 |
+
"loss": 0.8960294723510742,
|
| 1718 |
+
"step": 244
|
| 1719 |
+
},
|
| 1720 |
+
{
|
| 1721 |
+
"epoch": 0.9688581314878892,
|
| 1722 |
+
"grad_norm": 0.15072865784168243,
|
| 1723 |
+
"learning_rate": 0.00010459081836327347,
|
| 1724 |
+
"loss": 0.9184878468513489,
|
| 1725 |
+
"step": 245
|
| 1726 |
+
},
|
| 1727 |
+
{
|
| 1728 |
+
"epoch": 0.9728126544735541,
|
| 1729 |
+
"grad_norm": 0.1483660340309143,
|
| 1730 |
+
"learning_rate": 0.00010419161676646707,
|
| 1731 |
+
"loss": 1.000427484512329,
|
| 1732 |
+
"step": 246
|
| 1733 |
+
},
|
| 1734 |
+
{
|
| 1735 |
+
"epoch": 0.9767671774592189,
|
| 1736 |
+
"grad_norm": 0.16323480010032654,
|
| 1737 |
+
"learning_rate": 0.00010379241516966068,
|
| 1738 |
+
"loss": 0.9598352909088135,
|
| 1739 |
+
"step": 247
|
| 1740 |
+
},
|
| 1741 |
+
{
|
| 1742 |
+
"epoch": 0.9807217004448838,
|
| 1743 |
+
"grad_norm": 0.15449391305446625,
|
| 1744 |
+
"learning_rate": 0.0001033932135728543,
|
| 1745 |
+
"loss": 1.1161832809448242,
|
| 1746 |
+
"step": 248
|
| 1747 |
+
},
|
| 1748 |
+
{
|
| 1749 |
+
"epoch": 0.9846762234305487,
|
| 1750 |
+
"grad_norm": 0.15927913784980774,
|
| 1751 |
+
"learning_rate": 0.0001029940119760479,
|
| 1752 |
+
"loss": 0.9797319769859314,
|
| 1753 |
+
"step": 249
|
| 1754 |
+
},
|
| 1755 |
+
{
|
| 1756 |
+
"epoch": 0.9886307464162135,
|
| 1757 |
+
"grad_norm": 0.1711476743221283,
|
| 1758 |
+
"learning_rate": 0.00010259481037924152,
|
| 1759 |
+
"loss": 1.0162928104400635,
|
| 1760 |
+
"step": 250
|
| 1761 |
+
},
|
| 1762 |
+
{
|
| 1763 |
+
"epoch": 0.9925852694018784,
|
| 1764 |
+
"grad_norm": 0.15978345274925232,
|
| 1765 |
+
"learning_rate": 0.00010219560878243513,
|
| 1766 |
+
"loss": 0.8991297483444214,
|
| 1767 |
+
"step": 251
|
| 1768 |
+
},
|
| 1769 |
+
{
|
| 1770 |
+
"epoch": 0.9965397923875432,
|
| 1771 |
+
"grad_norm": 0.15758229792118073,
|
| 1772 |
+
"learning_rate": 0.00010179640718562875,
|
| 1773 |
+
"loss": 0.9805660843849182,
|
| 1774 |
+
"step": 252
|
| 1775 |
+
},
|
| 1776 |
+
{
|
| 1777 |
+
"epoch": 1.0,
|
| 1778 |
+
"grad_norm": 0.17985662817955017,
|
| 1779 |
+
"learning_rate": 0.00010139720558882235,
|
| 1780 |
+
"loss": 1.0062508583068848,
|
| 1781 |
+
"step": 253
|
| 1782 |
+
},
|
| 1783 |
+
{
|
| 1784 |
+
"epoch": 1.0039545229856648,
|
| 1785 |
+
"grad_norm": 0.16520218551158905,
|
| 1786 |
+
"learning_rate": 0.00010099800399201597,
|
| 1787 |
+
"loss": 1.0412254333496094,
|
| 1788 |
+
"step": 254
|
| 1789 |
+
},
|
| 1790 |
+
{
|
| 1791 |
+
"epoch": 1.0079090459713298,
|
| 1792 |
+
"grad_norm": 0.16414225101470947,
|
| 1793 |
+
"learning_rate": 0.00010059880239520958,
|
| 1794 |
+
"loss": 0.9039946794509888,
|
| 1795 |
+
"step": 255
|
| 1796 |
+
},
|
| 1797 |
+
{
|
| 1798 |
+
"epoch": 1.0118635689569946,
|
| 1799 |
+
"grad_norm": 0.15268850326538086,
|
| 1800 |
+
"learning_rate": 0.00010019960079840321,
|
| 1801 |
+
"loss": 0.9569994807243347,
|
| 1802 |
+
"step": 256
|
| 1803 |
+
},
|
| 1804 |
+
{
|
| 1805 |
+
"epoch": 1.0158180919426594,
|
| 1806 |
+
"grad_norm": 0.15560995042324066,
|
| 1807 |
+
"learning_rate": 9.98003992015968e-05,
|
| 1808 |
+
"loss": 1.1160976886749268,
|
| 1809 |
+
"step": 257
|
| 1810 |
+
},
|
| 1811 |
+
{
|
| 1812 |
+
"epoch": 1.0197726149283244,
|
| 1813 |
+
"grad_norm": 0.15324212610721588,
|
| 1814 |
+
"learning_rate": 9.940119760479042e-05,
|
| 1815 |
+
"loss": 1.0373631715774536,
|
| 1816 |
+
"step": 258
|
| 1817 |
+
},
|
| 1818 |
+
{
|
| 1819 |
+
"epoch": 1.0237271379139892,
|
| 1820 |
+
"grad_norm": 0.15931884944438934,
|
| 1821 |
+
"learning_rate": 9.900199600798404e-05,
|
| 1822 |
+
"loss": 0.9546667337417603,
|
| 1823 |
+
"step": 259
|
| 1824 |
+
},
|
| 1825 |
+
{
|
| 1826 |
+
"epoch": 1.027681660899654,
|
| 1827 |
+
"grad_norm": 0.16295117139816284,
|
| 1828 |
+
"learning_rate": 9.860279441117764e-05,
|
| 1829 |
+
"loss": 1.002548098564148,
|
| 1830 |
+
"step": 260
|
| 1831 |
+
},
|
| 1832 |
+
{
|
| 1833 |
+
"epoch": 1.0316361838853187,
|
| 1834 |
+
"grad_norm": 0.17251881957054138,
|
| 1835 |
+
"learning_rate": 9.820359281437126e-05,
|
| 1836 |
+
"loss": 0.922970175743103,
|
| 1837 |
+
"step": 261
|
| 1838 |
+
},
|
| 1839 |
+
{
|
| 1840 |
+
"epoch": 1.0355907068709838,
|
| 1841 |
+
"grad_norm": 0.1694485992193222,
|
| 1842 |
+
"learning_rate": 9.780439121756487e-05,
|
| 1843 |
+
"loss": 1.0178519487380981,
|
| 1844 |
+
"step": 262
|
| 1845 |
+
},
|
| 1846 |
+
{
|
| 1847 |
+
"epoch": 1.0395452298566485,
|
| 1848 |
+
"grad_norm": 0.16809332370758057,
|
| 1849 |
+
"learning_rate": 9.740518962075849e-05,
|
| 1850 |
+
"loss": 0.9237082004547119,
|
| 1851 |
+
"step": 263
|
| 1852 |
+
},
|
| 1853 |
+
{
|
| 1854 |
+
"epoch": 1.0434997528423133,
|
| 1855 |
+
"grad_norm": 0.16604508459568024,
|
| 1856 |
+
"learning_rate": 9.700598802395209e-05,
|
| 1857 |
+
"loss": 0.9441342353820801,
|
| 1858 |
+
"step": 264
|
| 1859 |
+
},
|
| 1860 |
+
{
|
| 1861 |
+
"epoch": 1.0474542758279783,
|
| 1862 |
+
"grad_norm": 0.16178838908672333,
|
| 1863 |
+
"learning_rate": 9.660678642714571e-05,
|
| 1864 |
+
"loss": 0.9563836455345154,
|
| 1865 |
+
"step": 265
|
| 1866 |
+
},
|
| 1867 |
+
{
|
| 1868 |
+
"epoch": 1.0514087988136431,
|
| 1869 |
+
"grad_norm": 0.17229746282100677,
|
| 1870 |
+
"learning_rate": 9.620758483033933e-05,
|
| 1871 |
+
"loss": 0.9493328928947449,
|
| 1872 |
+
"step": 266
|
| 1873 |
+
},
|
| 1874 |
+
{
|
| 1875 |
+
"epoch": 1.055363321799308,
|
| 1876 |
+
"grad_norm": 0.17116741836071014,
|
| 1877 |
+
"learning_rate": 9.580838323353294e-05,
|
| 1878 |
+
"loss": 0.7871612906455994,
|
| 1879 |
+
"step": 267
|
| 1880 |
+
},
|
| 1881 |
+
{
|
| 1882 |
+
"epoch": 1.0593178447849727,
|
| 1883 |
+
"grad_norm": 0.19620539247989655,
|
| 1884 |
+
"learning_rate": 9.540918163672655e-05,
|
| 1885 |
+
"loss": 0.7880710959434509,
|
| 1886 |
+
"step": 268
|
| 1887 |
+
},
|
| 1888 |
+
{
|
| 1889 |
+
"epoch": 1.0632723677706377,
|
| 1890 |
+
"grad_norm": 0.17312850058078766,
|
| 1891 |
+
"learning_rate": 9.500998003992016e-05,
|
| 1892 |
+
"loss": 0.8220570683479309,
|
| 1893 |
+
"step": 269
|
| 1894 |
+
},
|
| 1895 |
+
{
|
| 1896 |
+
"epoch": 1.0672268907563025,
|
| 1897 |
+
"grad_norm": 0.2012282907962799,
|
| 1898 |
+
"learning_rate": 9.461077844311378e-05,
|
| 1899 |
+
"loss": 0.8117949962615967,
|
| 1900 |
+
"step": 270
|
| 1901 |
+
},
|
| 1902 |
+
{
|
| 1903 |
+
"epoch": 1.0711814137419673,
|
| 1904 |
+
"grad_norm": 0.16529391705989838,
|
| 1905 |
+
"learning_rate": 9.42115768463074e-05,
|
| 1906 |
+
"loss": 1.000198483467102,
|
| 1907 |
+
"step": 271
|
| 1908 |
+
},
|
| 1909 |
+
{
|
| 1910 |
+
"epoch": 1.0751359367276323,
|
| 1911 |
+
"grad_norm": 0.17409488558769226,
|
| 1912 |
+
"learning_rate": 9.3812375249501e-05,
|
| 1913 |
+
"loss": 0.9956350922584534,
|
| 1914 |
+
"step": 272
|
| 1915 |
+
},
|
| 1916 |
+
{
|
| 1917 |
+
"epoch": 1.079090459713297,
|
| 1918 |
+
"grad_norm": 0.18472394347190857,
|
| 1919 |
+
"learning_rate": 9.341317365269462e-05,
|
| 1920 |
+
"loss": 1.0315921306610107,
|
| 1921 |
+
"step": 273
|
| 1922 |
+
},
|
| 1923 |
+
{
|
| 1924 |
+
"epoch": 1.0830449826989619,
|
| 1925 |
+
"grad_norm": 0.16941356658935547,
|
| 1926 |
+
"learning_rate": 9.301397205588823e-05,
|
| 1927 |
+
"loss": 0.95827716588974,
|
| 1928 |
+
"step": 274
|
| 1929 |
+
},
|
| 1930 |
+
{
|
| 1931 |
+
"epoch": 1.0869995056846269,
|
| 1932 |
+
"grad_norm": 0.16447168588638306,
|
| 1933 |
+
"learning_rate": 9.261477045908185e-05,
|
| 1934 |
+
"loss": 0.8616600036621094,
|
| 1935 |
+
"step": 275
|
| 1936 |
+
},
|
| 1937 |
+
{
|
| 1938 |
+
"epoch": 1.0909540286702917,
|
| 1939 |
+
"grad_norm": 0.19374088943004608,
|
| 1940 |
+
"learning_rate": 9.221556886227547e-05,
|
| 1941 |
+
"loss": 0.7994685173034668,
|
| 1942 |
+
"step": 276
|
| 1943 |
+
},
|
| 1944 |
+
{
|
| 1945 |
+
"epoch": 1.0949085516559565,
|
| 1946 |
+
"grad_norm": 0.16742141544818878,
|
| 1947 |
+
"learning_rate": 9.181636726546906e-05,
|
| 1948 |
+
"loss": 0.8132893443107605,
|
| 1949 |
+
"step": 277
|
| 1950 |
+
},
|
| 1951 |
+
{
|
| 1952 |
+
"epoch": 1.0988630746416213,
|
| 1953 |
+
"grad_norm": 0.20061492919921875,
|
| 1954 |
+
"learning_rate": 9.141716566866268e-05,
|
| 1955 |
+
"loss": 0.9603545665740967,
|
| 1956 |
+
"step": 278
|
| 1957 |
+
},
|
| 1958 |
+
{
|
| 1959 |
+
"epoch": 1.1028175976272863,
|
| 1960 |
+
"grad_norm": 0.16895346343517303,
|
| 1961 |
+
"learning_rate": 9.101796407185628e-05,
|
| 1962 |
+
"loss": 1.0123639106750488,
|
| 1963 |
+
"step": 279
|
| 1964 |
+
},
|
| 1965 |
+
{
|
| 1966 |
+
"epoch": 1.106772120612951,
|
| 1967 |
+
"grad_norm": 0.1848328411579132,
|
| 1968 |
+
"learning_rate": 9.06187624750499e-05,
|
| 1969 |
+
"loss": 0.9767408967018127,
|
| 1970 |
+
"step": 280
|
| 1971 |
+
},
|
| 1972 |
+
{
|
| 1973 |
+
"epoch": 1.1107266435986158,
|
| 1974 |
+
"grad_norm": 0.1762303113937378,
|
| 1975 |
+
"learning_rate": 9.021956087824352e-05,
|
| 1976 |
+
"loss": 1.0461487770080566,
|
| 1977 |
+
"step": 281
|
| 1978 |
+
},
|
| 1979 |
+
{
|
| 1980 |
+
"epoch": 1.1146811665842808,
|
| 1981 |
+
"grad_norm": 0.18317660689353943,
|
| 1982 |
+
"learning_rate": 8.982035928143712e-05,
|
| 1983 |
+
"loss": 0.8694061636924744,
|
| 1984 |
+
"step": 282
|
| 1985 |
+
},
|
| 1986 |
+
{
|
| 1987 |
+
"epoch": 1.1186356895699456,
|
| 1988 |
+
"grad_norm": 0.275339812040329,
|
| 1989 |
+
"learning_rate": 8.942115768463074e-05,
|
| 1990 |
+
"loss": 0.859940767288208,
|
| 1991 |
+
"step": 283
|
| 1992 |
+
},
|
| 1993 |
+
{
|
| 1994 |
+
"epoch": 1.1225902125556104,
|
| 1995 |
+
"grad_norm": 0.18906059861183167,
|
| 1996 |
+
"learning_rate": 8.902195608782435e-05,
|
| 1997 |
+
"loss": 0.9108320474624634,
|
| 1998 |
+
"step": 284
|
| 1999 |
+
},
|
| 2000 |
+
{
|
| 2001 |
+
"epoch": 1.1265447355412754,
|
| 2002 |
+
"grad_norm": 0.1821734756231308,
|
| 2003 |
+
"learning_rate": 8.862275449101797e-05,
|
| 2004 |
+
"loss": 0.8772467970848083,
|
| 2005 |
+
"step": 285
|
| 2006 |
+
},
|
| 2007 |
+
{
|
| 2008 |
+
"epoch": 1.1304992585269402,
|
| 2009 |
+
"grad_norm": 0.1746835559606552,
|
| 2010 |
+
"learning_rate": 8.822355289421159e-05,
|
| 2011 |
+
"loss": 0.9596457481384277,
|
| 2012 |
+
"step": 286
|
| 2013 |
+
},
|
| 2014 |
+
{
|
| 2015 |
+
"epoch": 1.134453781512605,
|
| 2016 |
+
"grad_norm": 0.17881642282009125,
|
| 2017 |
+
"learning_rate": 8.782435129740519e-05,
|
| 2018 |
+
"loss": 0.9305652379989624,
|
| 2019 |
+
"step": 287
|
| 2020 |
+
},
|
| 2021 |
+
{
|
| 2022 |
+
"epoch": 1.1384083044982698,
|
| 2023 |
+
"grad_norm": 0.18960846960544586,
|
| 2024 |
+
"learning_rate": 8.742514970059881e-05,
|
| 2025 |
+
"loss": 0.9538063406944275,
|
| 2026 |
+
"step": 288
|
| 2027 |
+
},
|
| 2028 |
+
{
|
| 2029 |
+
"epoch": 1.1423628274839348,
|
| 2030 |
+
"grad_norm": 0.17049920558929443,
|
| 2031 |
+
"learning_rate": 8.702594810379242e-05,
|
| 2032 |
+
"loss": 0.8245007395744324,
|
| 2033 |
+
"step": 289
|
| 2034 |
+
},
|
| 2035 |
+
{
|
| 2036 |
+
"epoch": 1.1463173504695996,
|
| 2037 |
+
"grad_norm": 0.18376107513904572,
|
| 2038 |
+
"learning_rate": 8.662674650698604e-05,
|
| 2039 |
+
"loss": 0.888431966304779,
|
| 2040 |
+
"step": 290
|
| 2041 |
+
},
|
| 2042 |
+
{
|
| 2043 |
+
"epoch": 1.1502718734552644,
|
| 2044 |
+
"grad_norm": 0.1948973834514618,
|
| 2045 |
+
"learning_rate": 8.622754491017964e-05,
|
| 2046 |
+
"loss": 0.9021210074424744,
|
| 2047 |
+
"step": 291
|
| 2048 |
+
},
|
| 2049 |
+
{
|
| 2050 |
+
"epoch": 1.1542263964409294,
|
| 2051 |
+
"grad_norm": 0.2063995599746704,
|
| 2052 |
+
"learning_rate": 8.582834331337326e-05,
|
| 2053 |
+
"loss": 0.9560433030128479,
|
| 2054 |
+
"step": 292
|
| 2055 |
+
},
|
| 2056 |
+
{
|
| 2057 |
+
"epoch": 1.1581809194265942,
|
| 2058 |
+
"grad_norm": 0.1944153606891632,
|
| 2059 |
+
"learning_rate": 8.542914171656688e-05,
|
| 2060 |
+
"loss": 0.9102143049240112,
|
| 2061 |
+
"step": 293
|
| 2062 |
+
},
|
| 2063 |
+
{
|
| 2064 |
+
"epoch": 1.162135442412259,
|
| 2065 |
+
"grad_norm": 0.19204482436180115,
|
| 2066 |
+
"learning_rate": 8.502994011976048e-05,
|
| 2067 |
+
"loss": 0.8926071524620056,
|
| 2068 |
+
"step": 294
|
| 2069 |
+
},
|
| 2070 |
+
{
|
| 2071 |
+
"epoch": 1.1660899653979238,
|
| 2072 |
+
"grad_norm": 0.20266790688037872,
|
| 2073 |
+
"learning_rate": 8.46307385229541e-05,
|
| 2074 |
+
"loss": 1.0379507541656494,
|
| 2075 |
+
"step": 295
|
| 2076 |
+
},
|
| 2077 |
+
{
|
| 2078 |
+
"epoch": 1.1700444883835888,
|
| 2079 |
+
"grad_norm": 0.18556608259677887,
|
| 2080 |
+
"learning_rate": 8.423153692614771e-05,
|
| 2081 |
+
"loss": 0.8484141826629639,
|
| 2082 |
+
"step": 296
|
| 2083 |
+
},
|
| 2084 |
+
{
|
| 2085 |
+
"epoch": 1.1739990113692536,
|
| 2086 |
+
"grad_norm": 0.20597191154956818,
|
| 2087 |
+
"learning_rate": 8.383233532934131e-05,
|
| 2088 |
+
"loss": 0.7798240780830383,
|
| 2089 |
+
"step": 297
|
| 2090 |
+
},
|
| 2091 |
+
{
|
| 2092 |
+
"epoch": 1.1779535343549183,
|
| 2093 |
+
"grad_norm": 0.18532755970954895,
|
| 2094 |
+
"learning_rate": 8.343313373253493e-05,
|
| 2095 |
+
"loss": 0.8546709418296814,
|
| 2096 |
+
"step": 298
|
| 2097 |
+
},
|
| 2098 |
+
{
|
| 2099 |
+
"epoch": 1.1819080573405834,
|
| 2100 |
+
"grad_norm": 0.19410572946071625,
|
| 2101 |
+
"learning_rate": 8.303393213572854e-05,
|
| 2102 |
+
"loss": 1.0295960903167725,
|
| 2103 |
+
"step": 299
|
| 2104 |
+
},
|
| 2105 |
+
{
|
| 2106 |
+
"epoch": 1.1858625803262481,
|
| 2107 |
+
"grad_norm": 0.1959802806377411,
|
| 2108 |
+
"learning_rate": 8.263473053892216e-05,
|
| 2109 |
+
"loss": 0.8178250193595886,
|
| 2110 |
+
"step": 300
|
| 2111 |
+
},
|
| 2112 |
+
{
|
| 2113 |
+
"epoch": 1.189817103311913,
|
| 2114 |
+
"grad_norm": 0.1843615621328354,
|
| 2115 |
+
"learning_rate": 8.223552894211578e-05,
|
| 2116 |
+
"loss": 0.9791051745414734,
|
| 2117 |
+
"step": 301
|
| 2118 |
+
},
|
| 2119 |
+
{
|
| 2120 |
+
"epoch": 1.193771626297578,
|
| 2121 |
+
"grad_norm": 0.19072717428207397,
|
| 2122 |
+
"learning_rate": 8.183632734530938e-05,
|
| 2123 |
+
"loss": 0.8327443599700928,
|
| 2124 |
+
"step": 302
|
| 2125 |
+
},
|
| 2126 |
+
{
|
| 2127 |
+
"epoch": 1.1977261492832427,
|
| 2128 |
+
"grad_norm": 0.18121100962162018,
|
| 2129 |
+
"learning_rate": 8.1437125748503e-05,
|
| 2130 |
+
"loss": 0.9626063704490662,
|
| 2131 |
+
"step": 303
|
| 2132 |
+
},
|
| 2133 |
+
{
|
| 2134 |
+
"epoch": 1.2016806722689075,
|
| 2135 |
+
"grad_norm": 0.1815863400697708,
|
| 2136 |
+
"learning_rate": 8.10379241516966e-05,
|
| 2137 |
+
"loss": 0.9914010763168335,
|
| 2138 |
+
"step": 304
|
| 2139 |
+
},
|
| 2140 |
+
{
|
| 2141 |
+
"epoch": 1.2056351952545725,
|
| 2142 |
+
"grad_norm": 0.17102938890457153,
|
| 2143 |
+
"learning_rate": 8.063872255489022e-05,
|
| 2144 |
+
"loss": 0.8116806149482727,
|
| 2145 |
+
"step": 305
|
| 2146 |
+
},
|
| 2147 |
+
{
|
| 2148 |
+
"epoch": 1.2095897182402373,
|
| 2149 |
+
"grad_norm": 0.18805690109729767,
|
| 2150 |
+
"learning_rate": 8.023952095808383e-05,
|
| 2151 |
+
"loss": 0.8748471140861511,
|
| 2152 |
+
"step": 306
|
| 2153 |
+
},
|
| 2154 |
+
{
|
| 2155 |
+
"epoch": 1.213544241225902,
|
| 2156 |
+
"grad_norm": 0.18052758276462555,
|
| 2157 |
+
"learning_rate": 7.984031936127745e-05,
|
| 2158 |
+
"loss": 0.9938658475875854,
|
| 2159 |
+
"step": 307
|
| 2160 |
+
},
|
| 2161 |
+
{
|
| 2162 |
+
"epoch": 1.2174987642115669,
|
| 2163 |
+
"grad_norm": 0.19235683977603912,
|
| 2164 |
+
"learning_rate": 7.944111776447107e-05,
|
| 2165 |
+
"loss": 1.1100804805755615,
|
| 2166 |
+
"step": 308
|
| 2167 |
+
},
|
| 2168 |
+
{
|
| 2169 |
+
"epoch": 1.221453287197232,
|
| 2170 |
+
"grad_norm": 0.18457858264446259,
|
| 2171 |
+
"learning_rate": 7.904191616766467e-05,
|
| 2172 |
+
"loss": 0.9189471006393433,
|
| 2173 |
+
"step": 309
|
| 2174 |
+
},
|
| 2175 |
+
{
|
| 2176 |
+
"epoch": 1.2254078101828967,
|
| 2177 |
+
"grad_norm": 0.18914006650447845,
|
| 2178 |
+
"learning_rate": 7.864271457085829e-05,
|
| 2179 |
+
"loss": 0.8920475840568542,
|
| 2180 |
+
"step": 310
|
| 2181 |
+
},
|
| 2182 |
+
{
|
| 2183 |
+
"epoch": 1.2293623331685615,
|
| 2184 |
+
"grad_norm": 0.19251829385757446,
|
| 2185 |
+
"learning_rate": 7.82435129740519e-05,
|
| 2186 |
+
"loss": 0.939852774143219,
|
| 2187 |
+
"step": 311
|
| 2188 |
+
},
|
| 2189 |
+
{
|
| 2190 |
+
"epoch": 1.2333168561542265,
|
| 2191 |
+
"grad_norm": 0.17629219591617584,
|
| 2192 |
+
"learning_rate": 7.784431137724552e-05,
|
| 2193 |
+
"loss": 1.0373430252075195,
|
| 2194 |
+
"step": 312
|
| 2195 |
+
},
|
| 2196 |
+
{
|
| 2197 |
+
"epoch": 1.2372713791398913,
|
| 2198 |
+
"grad_norm": 0.18412235379219055,
|
| 2199 |
+
"learning_rate": 7.744510978043914e-05,
|
| 2200 |
+
"loss": 0.9560478329658508,
|
| 2201 |
+
"step": 313
|
| 2202 |
+
},
|
| 2203 |
+
{
|
| 2204 |
+
"epoch": 1.241225902125556,
|
| 2205 |
+
"grad_norm": 0.191380113363266,
|
| 2206 |
+
"learning_rate": 7.704590818363274e-05,
|
| 2207 |
+
"loss": 0.8593218326568604,
|
| 2208 |
+
"step": 314
|
| 2209 |
+
},
|
| 2210 |
+
{
|
| 2211 |
+
"epoch": 1.2451804251112208,
|
| 2212 |
+
"grad_norm": 0.18198034167289734,
|
| 2213 |
+
"learning_rate": 7.664670658682636e-05,
|
| 2214 |
+
"loss": 0.8392094969749451,
|
| 2215 |
+
"step": 315
|
| 2216 |
+
},
|
| 2217 |
+
{
|
| 2218 |
+
"epoch": 1.2491349480968859,
|
| 2219 |
+
"grad_norm": 0.18671447038650513,
|
| 2220 |
+
"learning_rate": 7.624750499001996e-05,
|
| 2221 |
+
"loss": 0.9436869621276855,
|
| 2222 |
+
"step": 316
|
| 2223 |
+
},
|
| 2224 |
+
{
|
| 2225 |
+
"epoch": 1.2530894710825506,
|
| 2226 |
+
"grad_norm": 0.17818371951580048,
|
| 2227 |
+
"learning_rate": 7.584830339321357e-05,
|
| 2228 |
+
"loss": 0.886319100856781,
|
| 2229 |
+
"step": 317
|
| 2230 |
+
},
|
| 2231 |
+
{
|
| 2232 |
+
"epoch": 1.2570439940682157,
|
| 2233 |
+
"grad_norm": 0.18418045341968536,
|
| 2234 |
+
"learning_rate": 7.544910179640719e-05,
|
| 2235 |
+
"loss": 0.8711346983909607,
|
| 2236 |
+
"step": 318
|
| 2237 |
+
},
|
| 2238 |
+
{
|
| 2239 |
+
"epoch": 1.2609985170538804,
|
| 2240 |
+
"grad_norm": 0.18546275794506073,
|
| 2241 |
+
"learning_rate": 7.50499001996008e-05,
|
| 2242 |
+
"loss": 0.9372309446334839,
|
| 2243 |
+
"step": 319
|
| 2244 |
+
},
|
| 2245 |
+
{
|
| 2246 |
+
"epoch": 1.2649530400395452,
|
| 2247 |
+
"grad_norm": 0.19021575152873993,
|
| 2248 |
+
"learning_rate": 7.465069860279441e-05,
|
| 2249 |
+
"loss": 0.9252364039421082,
|
| 2250 |
+
"step": 320
|
| 2251 |
+
},
|
| 2252 |
+
{
|
| 2253 |
+
"epoch": 1.26890756302521,
|
| 2254 |
+
"grad_norm": 0.185474693775177,
|
| 2255 |
+
"learning_rate": 7.425149700598802e-05,
|
| 2256 |
+
"loss": 0.8521521687507629,
|
| 2257 |
+
"step": 321
|
| 2258 |
+
},
|
| 2259 |
+
{
|
| 2260 |
+
"epoch": 1.2728620860108748,
|
| 2261 |
+
"grad_norm": 0.20583201944828033,
|
| 2262 |
+
"learning_rate": 7.385229540918164e-05,
|
| 2263 |
+
"loss": 0.9617403149604797,
|
| 2264 |
+
"step": 322
|
| 2265 |
+
},
|
| 2266 |
+
{
|
| 2267 |
+
"epoch": 1.2768166089965398,
|
| 2268 |
+
"grad_norm": 0.19673901796340942,
|
| 2269 |
+
"learning_rate": 7.345309381237526e-05,
|
| 2270 |
+
"loss": 1.0314775705337524,
|
| 2271 |
+
"step": 323
|
| 2272 |
+
},
|
| 2273 |
+
{
|
| 2274 |
+
"epoch": 1.2807711319822046,
|
| 2275 |
+
"grad_norm": 0.18468639254570007,
|
| 2276 |
+
"learning_rate": 7.305389221556886e-05,
|
| 2277 |
+
"loss": 0.9060774445533752,
|
| 2278 |
+
"step": 324
|
| 2279 |
+
},
|
| 2280 |
+
{
|
| 2281 |
+
"epoch": 1.2847256549678696,
|
| 2282 |
+
"grad_norm": 0.18173182010650635,
|
| 2283 |
+
"learning_rate": 7.265469061876248e-05,
|
| 2284 |
+
"loss": 0.7989869117736816,
|
| 2285 |
+
"step": 325
|
| 2286 |
+
},
|
| 2287 |
+
{
|
| 2288 |
+
"epoch": 1.2886801779535344,
|
| 2289 |
+
"grad_norm": 0.1873815804719925,
|
| 2290 |
+
"learning_rate": 7.225548902195609e-05,
|
| 2291 |
+
"loss": 0.980597972869873,
|
| 2292 |
+
"step": 326
|
| 2293 |
+
},
|
| 2294 |
+
{
|
| 2295 |
+
"epoch": 1.2926347009391992,
|
| 2296 |
+
"grad_norm": 0.19111032783985138,
|
| 2297 |
+
"learning_rate": 7.18562874251497e-05,
|
| 2298 |
+
"loss": 0.9921055436134338,
|
| 2299 |
+
"step": 327
|
| 2300 |
+
},
|
| 2301 |
+
{
|
| 2302 |
+
"epoch": 1.296589223924864,
|
| 2303 |
+
"grad_norm": 0.1846132129430771,
|
| 2304 |
+
"learning_rate": 7.145708582834332e-05,
|
| 2305 |
+
"loss": 0.9843934178352356,
|
| 2306 |
+
"step": 328
|
| 2307 |
+
},
|
| 2308 |
+
{
|
| 2309 |
+
"epoch": 1.300543746910529,
|
| 2310 |
+
"grad_norm": 0.23496906459331512,
|
| 2311 |
+
"learning_rate": 7.105788423153693e-05,
|
| 2312 |
+
"loss": 1.000301480293274,
|
| 2313 |
+
"step": 329
|
| 2314 |
+
},
|
| 2315 |
+
{
|
| 2316 |
+
"epoch": 1.3044982698961938,
|
| 2317 |
+
"grad_norm": 0.1931312382221222,
|
| 2318 |
+
"learning_rate": 7.065868263473055e-05,
|
| 2319 |
+
"loss": 0.8303369283676147,
|
| 2320 |
+
"step": 330
|
| 2321 |
+
},
|
| 2322 |
+
{
|
| 2323 |
+
"epoch": 1.3084527928818586,
|
| 2324 |
+
"grad_norm": 0.185639426112175,
|
| 2325 |
+
"learning_rate": 7.025948103792415e-05,
|
| 2326 |
+
"loss": 0.898432731628418,
|
| 2327 |
+
"step": 331
|
| 2328 |
+
},
|
| 2329 |
+
{
|
| 2330 |
+
"epoch": 1.3124073158675236,
|
| 2331 |
+
"grad_norm": 0.20121130347251892,
|
| 2332 |
+
"learning_rate": 6.986027944111777e-05,
|
| 2333 |
+
"loss": 0.9152477383613586,
|
| 2334 |
+
"step": 332
|
| 2335 |
+
},
|
| 2336 |
+
{
|
| 2337 |
+
"epoch": 1.3163618388531884,
|
| 2338 |
+
"grad_norm": 0.21694481372833252,
|
| 2339 |
+
"learning_rate": 6.946107784431138e-05,
|
| 2340 |
+
"loss": 0.82918781042099,
|
| 2341 |
+
"step": 333
|
| 2342 |
+
},
|
| 2343 |
+
{
|
| 2344 |
+
"epoch": 1.3203163618388531,
|
| 2345 |
+
"grad_norm": 0.1961311399936676,
|
| 2346 |
+
"learning_rate": 6.9061876247505e-05,
|
| 2347 |
+
"loss": 0.7353627681732178,
|
| 2348 |
+
"step": 334
|
| 2349 |
+
},
|
| 2350 |
+
{
|
| 2351 |
+
"epoch": 1.324270884824518,
|
| 2352 |
+
"grad_norm": 0.18697749078273773,
|
| 2353 |
+
"learning_rate": 6.866267465069862e-05,
|
| 2354 |
+
"loss": 0.9976124167442322,
|
| 2355 |
+
"step": 335
|
| 2356 |
+
},
|
| 2357 |
+
{
|
| 2358 |
+
"epoch": 1.328225407810183,
|
| 2359 |
+
"grad_norm": 0.20015275478363037,
|
| 2360 |
+
"learning_rate": 6.826347305389222e-05,
|
| 2361 |
+
"loss": 0.9885960221290588,
|
| 2362 |
+
"step": 336
|
| 2363 |
+
},
|
| 2364 |
+
{
|
| 2365 |
+
"epoch": 1.3321799307958477,
|
| 2366 |
+
"grad_norm": 0.170740008354187,
|
| 2367 |
+
"learning_rate": 6.786427145708583e-05,
|
| 2368 |
+
"loss": 0.8704991936683655,
|
| 2369 |
+
"step": 337
|
| 2370 |
+
},
|
| 2371 |
+
{
|
| 2372 |
+
"epoch": 1.3361344537815127,
|
| 2373 |
+
"grad_norm": 0.2036600261926651,
|
| 2374 |
+
"learning_rate": 6.746506986027945e-05,
|
| 2375 |
+
"loss": 0.8373860120773315,
|
| 2376 |
+
"step": 338
|
| 2377 |
+
},
|
| 2378 |
+
{
|
| 2379 |
+
"epoch": 1.3400889767671775,
|
| 2380 |
+
"grad_norm": 0.1967741847038269,
|
| 2381 |
+
"learning_rate": 6.706586826347305e-05,
|
| 2382 |
+
"loss": 0.9344059228897095,
|
| 2383 |
+
"step": 339
|
| 2384 |
+
},
|
| 2385 |
+
{
|
| 2386 |
+
"epoch": 1.3440434997528423,
|
| 2387 |
+
"grad_norm": 0.18084974586963654,
|
| 2388 |
+
"learning_rate": 6.666666666666667e-05,
|
| 2389 |
+
"loss": 0.9507811069488525,
|
| 2390 |
+
"step": 340
|
| 2391 |
+
},
|
| 2392 |
+
{
|
| 2393 |
+
"epoch": 1.347998022738507,
|
| 2394 |
+
"grad_norm": 0.1890396773815155,
|
| 2395 |
+
"learning_rate": 6.626746506986028e-05,
|
| 2396 |
+
"loss": 0.9278057217597961,
|
| 2397 |
+
"step": 341
|
| 2398 |
+
},
|
| 2399 |
+
{
|
| 2400 |
+
"epoch": 1.351952545724172,
|
| 2401 |
+
"grad_norm": 0.20094835758209229,
|
| 2402 |
+
"learning_rate": 6.58682634730539e-05,
|
| 2403 |
+
"loss": 0.9359115362167358,
|
| 2404 |
+
"step": 342
|
| 2405 |
+
},
|
| 2406 |
+
{
|
| 2407 |
+
"epoch": 1.355907068709837,
|
| 2408 |
+
"grad_norm": 0.19261251389980316,
|
| 2409 |
+
"learning_rate": 6.54690618762475e-05,
|
| 2410 |
+
"loss": 0.9001166820526123,
|
| 2411 |
+
"step": 343
|
| 2412 |
+
},
|
| 2413 |
+
{
|
| 2414 |
+
"epoch": 1.3598615916955017,
|
| 2415 |
+
"grad_norm": 0.19089815020561218,
|
| 2416 |
+
"learning_rate": 6.506986027944112e-05,
|
| 2417 |
+
"loss": 0.9219706654548645,
|
| 2418 |
+
"step": 344
|
| 2419 |
+
},
|
| 2420 |
+
{
|
| 2421 |
+
"epoch": 1.3638161146811667,
|
| 2422 |
+
"grad_norm": 0.20358265936374664,
|
| 2423 |
+
"learning_rate": 6.467065868263474e-05,
|
| 2424 |
+
"loss": 0.8198409080505371,
|
| 2425 |
+
"step": 345
|
| 2426 |
+
},
|
| 2427 |
+
{
|
| 2428 |
+
"epoch": 1.3677706376668315,
|
| 2429 |
+
"grad_norm": 0.20188573002815247,
|
| 2430 |
+
"learning_rate": 6.427145708582834e-05,
|
| 2431 |
+
"loss": 0.9551017880439758,
|
| 2432 |
+
"step": 346
|
| 2433 |
+
},
|
| 2434 |
+
{
|
| 2435 |
+
"epoch": 1.3717251606524963,
|
| 2436 |
+
"grad_norm": 0.18636231124401093,
|
| 2437 |
+
"learning_rate": 6.387225548902196e-05,
|
| 2438 |
+
"loss": 1.007879376411438,
|
| 2439 |
+
"step": 347
|
| 2440 |
+
},
|
| 2441 |
+
{
|
| 2442 |
+
"epoch": 1.375679683638161,
|
| 2443 |
+
"grad_norm": 0.20502787828445435,
|
| 2444 |
+
"learning_rate": 6.347305389221557e-05,
|
| 2445 |
+
"loss": 0.7461677193641663,
|
| 2446 |
+
"step": 348
|
| 2447 |
+
},
|
| 2448 |
+
{
|
| 2449 |
+
"epoch": 1.379634206623826,
|
| 2450 |
+
"grad_norm": 0.1991313099861145,
|
| 2451 |
+
"learning_rate": 6.307385229540919e-05,
|
| 2452 |
+
"loss": 0.9965579509735107,
|
| 2453 |
+
"step": 349
|
| 2454 |
+
},
|
| 2455 |
+
{
|
| 2456 |
+
"epoch": 1.3835887296094909,
|
| 2457 |
+
"grad_norm": 0.18065449595451355,
|
| 2458 |
+
"learning_rate": 6.26746506986028e-05,
|
| 2459 |
+
"loss": 0.7860137820243835,
|
| 2460 |
+
"step": 350
|
| 2461 |
+
},
|
| 2462 |
+
{
|
| 2463 |
+
"epoch": 1.3875432525951557,
|
| 2464 |
+
"grad_norm": 0.1943470537662506,
|
| 2465 |
+
"learning_rate": 6.227544910179641e-05,
|
| 2466 |
+
"loss": 0.9140808582305908,
|
| 2467 |
+
"step": 351
|
| 2468 |
+
},
|
| 2469 |
+
{
|
| 2470 |
+
"epoch": 1.3914977755808207,
|
| 2471 |
+
"grad_norm": 0.18946987390518188,
|
| 2472 |
+
"learning_rate": 6.187624750499003e-05,
|
| 2473 |
+
"loss": 0.813644289970398,
|
| 2474 |
+
"step": 352
|
| 2475 |
+
},
|
| 2476 |
+
{
|
| 2477 |
+
"epoch": 1.3954522985664854,
|
| 2478 |
+
"grad_norm": 0.18935807049274445,
|
| 2479 |
+
"learning_rate": 6.147704590818363e-05,
|
| 2480 |
+
"loss": 0.7775428891181946,
|
| 2481 |
+
"step": 353
|
| 2482 |
+
},
|
| 2483 |
+
{
|
| 2484 |
+
"epoch": 1.3994068215521502,
|
| 2485 |
+
"grad_norm": 0.17212191224098206,
|
| 2486 |
+
"learning_rate": 6.107784431137725e-05,
|
| 2487 |
+
"loss": 0.791588306427002,
|
| 2488 |
+
"step": 354
|
| 2489 |
+
},
|
| 2490 |
+
{
|
| 2491 |
+
"epoch": 1.403361344537815,
|
| 2492 |
+
"grad_norm": 0.1845926195383072,
|
| 2493 |
+
"learning_rate": 6.0678642714570866e-05,
|
| 2494 |
+
"loss": 0.8854494690895081,
|
| 2495 |
+
"step": 355
|
| 2496 |
+
},
|
| 2497 |
+
{
|
| 2498 |
+
"epoch": 1.40731586752348,
|
| 2499 |
+
"grad_norm": 0.19545923173427582,
|
| 2500 |
+
"learning_rate": 6.027944111776448e-05,
|
| 2501 |
+
"loss": 0.9289248585700989,
|
| 2502 |
+
"step": 356
|
| 2503 |
+
},
|
| 2504 |
+
{
|
| 2505 |
+
"epoch": 1.4112703905091448,
|
| 2506 |
+
"grad_norm": 0.23064754903316498,
|
| 2507 |
+
"learning_rate": 5.988023952095808e-05,
|
| 2508 |
+
"loss": 0.8877400755882263,
|
| 2509 |
+
"step": 357
|
| 2510 |
+
},
|
| 2511 |
+
{
|
| 2512 |
+
"epoch": 1.4152249134948096,
|
| 2513 |
+
"grad_norm": 0.18640878796577454,
|
| 2514 |
+
"learning_rate": 5.9481037924151695e-05,
|
| 2515 |
+
"loss": 0.8007562160491943,
|
| 2516 |
+
"step": 358
|
| 2517 |
+
},
|
| 2518 |
+
{
|
| 2519 |
+
"epoch": 1.4191794364804746,
|
| 2520 |
+
"grad_norm": 0.196823850274086,
|
| 2521 |
+
"learning_rate": 5.908183632734531e-05,
|
| 2522 |
+
"loss": 0.9147028923034668,
|
| 2523 |
+
"step": 359
|
| 2524 |
+
},
|
| 2525 |
+
{
|
| 2526 |
+
"epoch": 1.4231339594661394,
|
| 2527 |
+
"grad_norm": 0.18492864072322845,
|
| 2528 |
+
"learning_rate": 5.868263473053892e-05,
|
| 2529 |
+
"loss": 1.0567796230316162,
|
| 2530 |
+
"step": 360
|
| 2531 |
+
},
|
| 2532 |
+
{
|
| 2533 |
+
"epoch": 1.4270884824518042,
|
| 2534 |
+
"grad_norm": 0.19871287047863007,
|
| 2535 |
+
"learning_rate": 5.828343313373253e-05,
|
| 2536 |
+
"loss": 0.858993411064148,
|
| 2537 |
+
"step": 361
|
| 2538 |
+
},
|
| 2539 |
+
{
|
| 2540 |
+
"epoch": 1.431043005437469,
|
| 2541 |
+
"grad_norm": 0.1865636110305786,
|
| 2542 |
+
"learning_rate": 5.788423153692615e-05,
|
| 2543 |
+
"loss": 0.9598185420036316,
|
| 2544 |
+
"step": 362
|
| 2545 |
+
},
|
| 2546 |
+
{
|
| 2547 |
+
"epoch": 1.434997528423134,
|
| 2548 |
+
"grad_norm": 0.1950787603855133,
|
| 2549 |
+
"learning_rate": 5.748502994011976e-05,
|
| 2550 |
+
"loss": 0.866109311580658,
|
| 2551 |
+
"step": 363
|
| 2552 |
+
},
|
| 2553 |
+
{
|
| 2554 |
+
"epoch": 1.4389520514087988,
|
| 2555 |
+
"grad_norm": 0.19537241756916046,
|
| 2556 |
+
"learning_rate": 5.7085828343313375e-05,
|
| 2557 |
+
"loss": 0.9231923818588257,
|
| 2558 |
+
"step": 364
|
| 2559 |
+
},
|
| 2560 |
+
{
|
| 2561 |
+
"epoch": 1.4429065743944638,
|
| 2562 |
+
"grad_norm": 0.1848801076412201,
|
| 2563 |
+
"learning_rate": 5.668662674650699e-05,
|
| 2564 |
+
"loss": 0.9152354001998901,
|
| 2565 |
+
"step": 365
|
| 2566 |
+
},
|
| 2567 |
+
{
|
| 2568 |
+
"epoch": 1.4468610973801286,
|
| 2569 |
+
"grad_norm": 0.19516348838806152,
|
| 2570 |
+
"learning_rate": 5.62874251497006e-05,
|
| 2571 |
+
"loss": 0.9022656679153442,
|
| 2572 |
+
"step": 366
|
| 2573 |
+
},
|
| 2574 |
+
{
|
| 2575 |
+
"epoch": 1.4508156203657934,
|
| 2576 |
+
"grad_norm": 0.1875586360692978,
|
| 2577 |
+
"learning_rate": 5.588822355289421e-05,
|
| 2578 |
+
"loss": 0.9204172492027283,
|
| 2579 |
+
"step": 367
|
| 2580 |
+
},
|
| 2581 |
+
{
|
| 2582 |
+
"epoch": 1.4547701433514582,
|
| 2583 |
+
"grad_norm": 0.19239316880702972,
|
| 2584 |
+
"learning_rate": 5.548902195608783e-05,
|
| 2585 |
+
"loss": 0.9383366107940674,
|
| 2586 |
+
"step": 368
|
| 2587 |
+
},
|
| 2588 |
+
{
|
| 2589 |
+
"epoch": 1.4587246663371232,
|
| 2590 |
+
"grad_norm": 0.19720111787319183,
|
| 2591 |
+
"learning_rate": 5.508982035928144e-05,
|
| 2592 |
+
"loss": 0.961857795715332,
|
| 2593 |
+
"step": 369
|
| 2594 |
+
},
|
| 2595 |
+
{
|
| 2596 |
+
"epoch": 1.462679189322788,
|
| 2597 |
+
"grad_norm": 0.213732048869133,
|
| 2598 |
+
"learning_rate": 5.4690618762475055e-05,
|
| 2599 |
+
"loss": 0.8472163081169128,
|
| 2600 |
+
"step": 370
|
| 2601 |
+
},
|
| 2602 |
+
{
|
| 2603 |
+
"epoch": 1.4666337123084527,
|
| 2604 |
+
"grad_norm": 0.1938932090997696,
|
| 2605 |
+
"learning_rate": 5.429141716566867e-05,
|
| 2606 |
+
"loss": 0.8686806559562683,
|
| 2607 |
+
"step": 371
|
| 2608 |
+
},
|
| 2609 |
+
{
|
| 2610 |
+
"epoch": 1.4705882352941178,
|
| 2611 |
+
"grad_norm": 0.19152139127254486,
|
| 2612 |
+
"learning_rate": 5.389221556886228e-05,
|
| 2613 |
+
"loss": 0.8825725317001343,
|
| 2614 |
+
"step": 372
|
| 2615 |
+
},
|
| 2616 |
+
{
|
| 2617 |
+
"epoch": 1.4745427582797825,
|
| 2618 |
+
"grad_norm": 0.19953249394893646,
|
| 2619 |
+
"learning_rate": 5.349301397205589e-05,
|
| 2620 |
+
"loss": 1.0075819492340088,
|
| 2621 |
+
"step": 373
|
| 2622 |
+
},
|
| 2623 |
+
{
|
| 2624 |
+
"epoch": 1.4784972812654473,
|
| 2625 |
+
"grad_norm": 0.19870170950889587,
|
| 2626 |
+
"learning_rate": 5.309381237524951e-05,
|
| 2627 |
+
"loss": 0.8033257126808167,
|
| 2628 |
+
"step": 374
|
| 2629 |
+
},
|
| 2630 |
+
{
|
| 2631 |
+
"epoch": 1.4824518042511121,
|
| 2632 |
+
"grad_norm": 0.21190565824508667,
|
| 2633 |
+
"learning_rate": 5.269461077844312e-05,
|
| 2634 |
+
"loss": 0.9029344916343689,
|
| 2635 |
+
"step": 375
|
| 2636 |
+
},
|
| 2637 |
+
{
|
| 2638 |
+
"epoch": 1.4864063272367771,
|
| 2639 |
+
"grad_norm": 0.20307143032550812,
|
| 2640 |
+
"learning_rate": 5.2295409181636734e-05,
|
| 2641 |
+
"loss": 0.9491141438484192,
|
| 2642 |
+
"step": 376
|
| 2643 |
+
},
|
| 2644 |
+
{
|
| 2645 |
+
"epoch": 1.490360850222442,
|
| 2646 |
+
"grad_norm": 0.21230387687683105,
|
| 2647 |
+
"learning_rate": 5.189620758483034e-05,
|
| 2648 |
+
"loss": 0.9077959060668945,
|
| 2649 |
+
"step": 377
|
| 2650 |
+
},
|
| 2651 |
+
{
|
| 2652 |
+
"epoch": 1.4943153732081067,
|
| 2653 |
+
"grad_norm": 0.20202863216400146,
|
| 2654 |
+
"learning_rate": 5.149700598802395e-05,
|
| 2655 |
+
"loss": 0.9439913630485535,
|
| 2656 |
+
"step": 378
|
| 2657 |
+
},
|
| 2658 |
+
{
|
| 2659 |
+
"epoch": 1.4982698961937717,
|
| 2660 |
+
"grad_norm": 0.1929042786359787,
|
| 2661 |
+
"learning_rate": 5.1097804391217564e-05,
|
| 2662 |
+
"loss": 0.9769095778465271,
|
| 2663 |
+
"step": 379
|
| 2664 |
+
},
|
| 2665 |
+
{
|
| 2666 |
+
"epoch": 1.5022244191794365,
|
| 2667 |
+
"grad_norm": 0.21215690672397614,
|
| 2668 |
+
"learning_rate": 5.0698602794411176e-05,
|
| 2669 |
+
"loss": 0.7954482436180115,
|
| 2670 |
+
"step": 380
|
| 2671 |
+
},
|
| 2672 |
+
{
|
| 2673 |
+
"epoch": 1.5061789421651013,
|
| 2674 |
+
"grad_norm": 0.19395476579666138,
|
| 2675 |
+
"learning_rate": 5.029940119760479e-05,
|
| 2676 |
+
"loss": 0.8338028192520142,
|
| 2677 |
+
"step": 381
|
| 2678 |
+
},
|
| 2679 |
+
{
|
| 2680 |
+
"epoch": 1.510133465150766,
|
| 2681 |
+
"grad_norm": 0.2140199989080429,
|
| 2682 |
+
"learning_rate": 4.99001996007984e-05,
|
| 2683 |
+
"loss": 0.7927234172821045,
|
| 2684 |
+
"step": 382
|
| 2685 |
+
},
|
| 2686 |
+
{
|
| 2687 |
+
"epoch": 1.514087988136431,
|
| 2688 |
+
"grad_norm": 0.20886200666427612,
|
| 2689 |
+
"learning_rate": 4.950099800399202e-05,
|
| 2690 |
+
"loss": 0.9537109136581421,
|
| 2691 |
+
"step": 383
|
| 2692 |
+
},
|
| 2693 |
+
{
|
| 2694 |
+
"epoch": 1.5180425111220959,
|
| 2695 |
+
"grad_norm": 0.21319736540317535,
|
| 2696 |
+
"learning_rate": 4.910179640718563e-05,
|
| 2697 |
+
"loss": 0.7856256365776062,
|
| 2698 |
+
"step": 384
|
| 2699 |
+
},
|
| 2700 |
+
{
|
| 2701 |
+
"epoch": 1.5219970341077609,
|
| 2702 |
+
"grad_norm": 0.21167606115341187,
|
| 2703 |
+
"learning_rate": 4.8702594810379244e-05,
|
| 2704 |
+
"loss": 0.9939818978309631,
|
| 2705 |
+
"step": 385
|
| 2706 |
+
},
|
| 2707 |
+
{
|
| 2708 |
+
"epoch": 1.5259515570934257,
|
| 2709 |
+
"grad_norm": 0.20061740279197693,
|
| 2710 |
+
"learning_rate": 4.8303393213572856e-05,
|
| 2711 |
+
"loss": 0.9484026432037354,
|
| 2712 |
+
"step": 386
|
| 2713 |
+
},
|
| 2714 |
+
{
|
| 2715 |
+
"epoch": 1.5299060800790905,
|
| 2716 |
+
"grad_norm": 0.2059990018606186,
|
| 2717 |
+
"learning_rate": 4.790419161676647e-05,
|
| 2718 |
+
"loss": 0.8519854545593262,
|
| 2719 |
+
"step": 387
|
| 2720 |
+
},
|
| 2721 |
+
{
|
| 2722 |
+
"epoch": 1.5338606030647552,
|
| 2723 |
+
"grad_norm": 0.1978318989276886,
|
| 2724 |
+
"learning_rate": 4.750499001996008e-05,
|
| 2725 |
+
"loss": 0.8361144065856934,
|
| 2726 |
+
"step": 388
|
| 2727 |
+
},
|
| 2728 |
+
{
|
| 2729 |
+
"epoch": 1.53781512605042,
|
| 2730 |
+
"grad_norm": 0.2057390809059143,
|
| 2731 |
+
"learning_rate": 4.71057884231537e-05,
|
| 2732 |
+
"loss": 0.8409650325775146,
|
| 2733 |
+
"step": 389
|
| 2734 |
+
},
|
| 2735 |
+
{
|
| 2736 |
+
"epoch": 1.541769649036085,
|
| 2737 |
+
"grad_norm": 0.1905912309885025,
|
| 2738 |
+
"learning_rate": 4.670658682634731e-05,
|
| 2739 |
+
"loss": 0.8437705039978027,
|
| 2740 |
+
"step": 390
|
| 2741 |
+
},
|
| 2742 |
+
{
|
| 2743 |
+
"epoch": 1.5457241720217498,
|
| 2744 |
+
"grad_norm": 0.23166270554065704,
|
| 2745 |
+
"learning_rate": 4.630738522954092e-05,
|
| 2746 |
+
"loss": 1.1090465784072876,
|
| 2747 |
+
"step": 391
|
| 2748 |
+
},
|
| 2749 |
+
{
|
| 2750 |
+
"epoch": 1.5496786950074148,
|
| 2751 |
+
"grad_norm": 0.19709770381450653,
|
| 2752 |
+
"learning_rate": 4.590818363273453e-05,
|
| 2753 |
+
"loss": 0.9146186709403992,
|
| 2754 |
+
"step": 392
|
| 2755 |
+
},
|
| 2756 |
+
{
|
| 2757 |
+
"epoch": 1.5536332179930796,
|
| 2758 |
+
"grad_norm": 0.24494928121566772,
|
| 2759 |
+
"learning_rate": 4.550898203592814e-05,
|
| 2760 |
+
"loss": 0.8773439526557922,
|
| 2761 |
+
"step": 393
|
| 2762 |
+
},
|
| 2763 |
+
{
|
| 2764 |
+
"epoch": 1.5575877409787444,
|
| 2765 |
+
"grad_norm": 0.1936318725347519,
|
| 2766 |
+
"learning_rate": 4.510978043912176e-05,
|
| 2767 |
+
"loss": 0.8627104163169861,
|
| 2768 |
+
"step": 394
|
| 2769 |
+
},
|
| 2770 |
+
{
|
| 2771 |
+
"epoch": 1.5615422639644092,
|
| 2772 |
+
"grad_norm": 0.1917494237422943,
|
| 2773 |
+
"learning_rate": 4.471057884231537e-05,
|
| 2774 |
+
"loss": 0.9282490015029907,
|
| 2775 |
+
"step": 395
|
| 2776 |
+
},
|
| 2777 |
+
{
|
| 2778 |
+
"epoch": 1.565496786950074,
|
| 2779 |
+
"grad_norm": 0.20790532231330872,
|
| 2780 |
+
"learning_rate": 4.4311377245508984e-05,
|
| 2781 |
+
"loss": 0.9672350883483887,
|
| 2782 |
+
"step": 396
|
| 2783 |
+
},
|
| 2784 |
+
{
|
| 2785 |
+
"epoch": 1.569451309935739,
|
| 2786 |
+
"grad_norm": 0.21338993310928345,
|
| 2787 |
+
"learning_rate": 4.3912175648702596e-05,
|
| 2788 |
+
"loss": 0.9289650321006775,
|
| 2789 |
+
"step": 397
|
| 2790 |
+
},
|
| 2791 |
+
{
|
| 2792 |
+
"epoch": 1.573405832921404,
|
| 2793 |
+
"grad_norm": 0.20963947474956512,
|
| 2794 |
+
"learning_rate": 4.351297405189621e-05,
|
| 2795 |
+
"loss": 0.9762122631072998,
|
| 2796 |
+
"step": 398
|
| 2797 |
+
},
|
| 2798 |
+
{
|
| 2799 |
+
"epoch": 1.5773603559070688,
|
| 2800 |
+
"grad_norm": 0.20937083661556244,
|
| 2801 |
+
"learning_rate": 4.311377245508982e-05,
|
| 2802 |
+
"loss": 0.9213401079177856,
|
| 2803 |
+
"step": 399
|
| 2804 |
+
},
|
| 2805 |
+
{
|
| 2806 |
+
"epoch": 1.5813148788927336,
|
| 2807 |
+
"grad_norm": 0.20903795957565308,
|
| 2808 |
+
"learning_rate": 4.271457085828344e-05,
|
| 2809 |
+
"loss": 0.9836536645889282,
|
| 2810 |
+
"step": 400
|
| 2811 |
+
},
|
| 2812 |
+
{
|
| 2813 |
+
"epoch": 1.5852694018783984,
|
| 2814 |
+
"grad_norm": 0.20544159412384033,
|
| 2815 |
+
"learning_rate": 4.231536926147705e-05,
|
| 2816 |
+
"loss": 0.8152018785476685,
|
| 2817 |
+
"step": 401
|
| 2818 |
+
},
|
| 2819 |
+
{
|
| 2820 |
+
"epoch": 1.5892239248640632,
|
| 2821 |
+
"grad_norm": 0.1830497533082962,
|
| 2822 |
+
"learning_rate": 4.191616766467066e-05,
|
| 2823 |
+
"loss": 0.9987653493881226,
|
| 2824 |
+
"step": 402
|
| 2825 |
+
},
|
| 2826 |
+
{
|
| 2827 |
+
"epoch": 1.5931784478497282,
|
| 2828 |
+
"grad_norm": 0.1913018822669983,
|
| 2829 |
+
"learning_rate": 4.151696606786427e-05,
|
| 2830 |
+
"loss": 0.867318332195282,
|
| 2831 |
+
"step": 403
|
| 2832 |
+
},
|
| 2833 |
+
{
|
| 2834 |
+
"epoch": 1.597132970835393,
|
| 2835 |
+
"grad_norm": 0.19043081998825073,
|
| 2836 |
+
"learning_rate": 4.111776447105789e-05,
|
| 2837 |
+
"loss": 0.8715974688529968,
|
| 2838 |
+
"step": 404
|
| 2839 |
+
},
|
| 2840 |
+
{
|
| 2841 |
+
"epoch": 1.601087493821058,
|
| 2842 |
+
"grad_norm": 0.18875691294670105,
|
| 2843 |
+
"learning_rate": 4.07185628742515e-05,
|
| 2844 |
+
"loss": 0.904281735420227,
|
| 2845 |
+
"step": 405
|
| 2846 |
+
},
|
| 2847 |
+
{
|
| 2848 |
+
"epoch": 1.6050420168067228,
|
| 2849 |
+
"grad_norm": 0.2014741152524948,
|
| 2850 |
+
"learning_rate": 4.031936127744511e-05,
|
| 2851 |
+
"loss": 0.9567676186561584,
|
| 2852 |
+
"step": 406
|
| 2853 |
+
},
|
| 2854 |
+
{
|
| 2855 |
+
"epoch": 1.6089965397923875,
|
| 2856 |
+
"grad_norm": 0.17751306295394897,
|
| 2857 |
+
"learning_rate": 3.9920159680638724e-05,
|
| 2858 |
+
"loss": 0.8185122013092041,
|
| 2859 |
+
"step": 407
|
| 2860 |
+
},
|
| 2861 |
+
{
|
| 2862 |
+
"epoch": 1.6129510627780523,
|
| 2863 |
+
"grad_norm": 0.20318663120269775,
|
| 2864 |
+
"learning_rate": 3.9520958083832336e-05,
|
| 2865 |
+
"loss": 0.9554027318954468,
|
| 2866 |
+
"step": 408
|
| 2867 |
+
},
|
| 2868 |
+
{
|
| 2869 |
+
"epoch": 1.6169055857637171,
|
| 2870 |
+
"grad_norm": 0.1872302144765854,
|
| 2871 |
+
"learning_rate": 3.912175648702595e-05,
|
| 2872 |
+
"loss": 0.9051477313041687,
|
| 2873 |
+
"step": 409
|
| 2874 |
+
},
|
| 2875 |
+
{
|
| 2876 |
+
"epoch": 1.6208601087493821,
|
| 2877 |
+
"grad_norm": 0.21332977712154388,
|
| 2878 |
+
"learning_rate": 3.872255489021957e-05,
|
| 2879 |
+
"loss": 0.944547712802887,
|
| 2880 |
+
"step": 410
|
| 2881 |
+
},
|
| 2882 |
+
{
|
| 2883 |
+
"epoch": 1.624814631735047,
|
| 2884 |
+
"grad_norm": 0.18820840120315552,
|
| 2885 |
+
"learning_rate": 3.832335329341318e-05,
|
| 2886 |
+
"loss": 0.9271795749664307,
|
| 2887 |
+
"step": 411
|
| 2888 |
+
},
|
| 2889 |
+
{
|
| 2890 |
+
"epoch": 1.628769154720712,
|
| 2891 |
+
"grad_norm": 0.2146555483341217,
|
| 2892 |
+
"learning_rate": 3.7924151696606785e-05,
|
| 2893 |
+
"loss": 0.8307335376739502,
|
| 2894 |
+
"step": 412
|
| 2895 |
+
},
|
| 2896 |
+
{
|
| 2897 |
+
"epoch": 1.6327236777063767,
|
| 2898 |
+
"grad_norm": 0.2196199744939804,
|
| 2899 |
+
"learning_rate": 3.75249500998004e-05,
|
| 2900 |
+
"loss": 0.8816962242126465,
|
| 2901 |
+
"step": 413
|
| 2902 |
+
},
|
| 2903 |
+
{
|
| 2904 |
+
"epoch": 1.6366782006920415,
|
| 2905 |
+
"grad_norm": 0.1986646205186844,
|
| 2906 |
+
"learning_rate": 3.712574850299401e-05,
|
| 2907 |
+
"loss": 0.9725742936134338,
|
| 2908 |
+
"step": 414
|
| 2909 |
+
},
|
| 2910 |
+
{
|
| 2911 |
+
"epoch": 1.6406327236777063,
|
| 2912 |
+
"grad_norm": 0.18679843842983246,
|
| 2913 |
+
"learning_rate": 3.672654690618763e-05,
|
| 2914 |
+
"loss": 0.9385328888893127,
|
| 2915 |
+
"step": 415
|
| 2916 |
+
},
|
| 2917 |
+
{
|
| 2918 |
+
"epoch": 1.644587246663371,
|
| 2919 |
+
"grad_norm": 0.18637174367904663,
|
| 2920 |
+
"learning_rate": 3.632734530938124e-05,
|
| 2921 |
+
"loss": 0.9450833201408386,
|
| 2922 |
+
"step": 416
|
| 2923 |
+
},
|
| 2924 |
+
{
|
| 2925 |
+
"epoch": 1.648541769649036,
|
| 2926 |
+
"grad_norm": 0.20110389590263367,
|
| 2927 |
+
"learning_rate": 3.592814371257485e-05,
|
| 2928 |
+
"loss": 0.9009251594543457,
|
| 2929 |
+
"step": 417
|
| 2930 |
+
},
|
| 2931 |
+
{
|
| 2932 |
+
"epoch": 1.652496292634701,
|
| 2933 |
+
"grad_norm": 0.22820661962032318,
|
| 2934 |
+
"learning_rate": 3.5528942115768465e-05,
|
| 2935 |
+
"loss": 0.9610739350318909,
|
| 2936 |
+
"step": 418
|
| 2937 |
+
},
|
| 2938 |
+
{
|
| 2939 |
+
"epoch": 1.656450815620366,
|
| 2940 |
+
"grad_norm": 0.20116274058818817,
|
| 2941 |
+
"learning_rate": 3.512974051896208e-05,
|
| 2942 |
+
"loss": 0.9576656818389893,
|
| 2943 |
+
"step": 419
|
| 2944 |
+
},
|
| 2945 |
+
{
|
| 2946 |
+
"epoch": 1.6604053386060307,
|
| 2947 |
+
"grad_norm": 0.19964101910591125,
|
| 2948 |
+
"learning_rate": 3.473053892215569e-05,
|
| 2949 |
+
"loss": 0.9899635314941406,
|
| 2950 |
+
"step": 420
|
| 2951 |
+
},
|
| 2952 |
+
{
|
| 2953 |
+
"epoch": 1.6643598615916955,
|
| 2954 |
+
"grad_norm": 0.21723447740077972,
|
| 2955 |
+
"learning_rate": 3.433133732534931e-05,
|
| 2956 |
+
"loss": 0.7412058115005493,
|
| 2957 |
+
"step": 421
|
| 2958 |
+
},
|
| 2959 |
+
{
|
| 2960 |
+
"epoch": 1.6683143845773603,
|
| 2961 |
+
"grad_norm": 0.21164871752262115,
|
| 2962 |
+
"learning_rate": 3.393213572854291e-05,
|
| 2963 |
+
"loss": 1.107064127922058,
|
| 2964 |
+
"step": 422
|
| 2965 |
+
},
|
| 2966 |
+
{
|
| 2967 |
+
"epoch": 1.6722689075630253,
|
| 2968 |
+
"grad_norm": 0.20853902399539948,
|
| 2969 |
+
"learning_rate": 3.3532934131736525e-05,
|
| 2970 |
+
"loss": 1.0325095653533936,
|
| 2971 |
+
"step": 423
|
| 2972 |
+
},
|
| 2973 |
+
{
|
| 2974 |
+
"epoch": 1.67622343054869,
|
| 2975 |
+
"grad_norm": 0.1993069052696228,
|
| 2976 |
+
"learning_rate": 3.313373253493014e-05,
|
| 2977 |
+
"loss": 0.8811869621276855,
|
| 2978 |
+
"step": 424
|
| 2979 |
+
},
|
| 2980 |
+
{
|
| 2981 |
+
"epoch": 1.680177953534355,
|
| 2982 |
+
"grad_norm": 0.2086590826511383,
|
| 2983 |
+
"learning_rate": 3.273453093812375e-05,
|
| 2984 |
+
"loss": 0.8045269846916199,
|
| 2985 |
+
"step": 425
|
| 2986 |
+
},
|
| 2987 |
+
{
|
| 2988 |
+
"epoch": 1.6841324765200198,
|
| 2989 |
+
"grad_norm": 0.19502981007099152,
|
| 2990 |
+
"learning_rate": 3.233532934131737e-05,
|
| 2991 |
+
"loss": 1.0250900983810425,
|
| 2992 |
+
"step": 426
|
| 2993 |
+
},
|
| 2994 |
+
{
|
| 2995 |
+
"epoch": 1.6880869995056846,
|
| 2996 |
+
"grad_norm": 0.1864766627550125,
|
| 2997 |
+
"learning_rate": 3.193612774451098e-05,
|
| 2998 |
+
"loss": 0.7248687744140625,
|
| 2999 |
+
"step": 427
|
| 3000 |
+
},
|
| 3001 |
+
{
|
| 3002 |
+
"epoch": 1.6920415224913494,
|
| 3003 |
+
"grad_norm": 0.20849153399467468,
|
| 3004 |
+
"learning_rate": 3.153692614770459e-05,
|
| 3005 |
+
"loss": 0.8926454186439514,
|
| 3006 |
+
"step": 428
|
| 3007 |
+
},
|
| 3008 |
+
{
|
| 3009 |
+
"epoch": 1.6959960454770142,
|
| 3010 |
+
"grad_norm": 0.20903082191944122,
|
| 3011 |
+
"learning_rate": 3.1137724550898205e-05,
|
| 3012 |
+
"loss": 0.8962099552154541,
|
| 3013 |
+
"step": 429
|
| 3014 |
+
},
|
| 3015 |
+
{
|
| 3016 |
+
"epoch": 1.6999505684626792,
|
| 3017 |
+
"grad_norm": 0.1925486922264099,
|
| 3018 |
+
"learning_rate": 3.073852295409182e-05,
|
| 3019 |
+
"loss": 0.9962249994277954,
|
| 3020 |
+
"step": 430
|
| 3021 |
+
},
|
| 3022 |
+
{
|
| 3023 |
+
"epoch": 1.703905091448344,
|
| 3024 |
+
"grad_norm": 0.20653171837329865,
|
| 3025 |
+
"learning_rate": 3.0339321357285433e-05,
|
| 3026 |
+
"loss": 0.8618420958518982,
|
| 3027 |
+
"step": 431
|
| 3028 |
+
},
|
| 3029 |
+
{
|
| 3030 |
+
"epoch": 1.707859614434009,
|
| 3031 |
+
"grad_norm": 0.21909548342227936,
|
| 3032 |
+
"learning_rate": 2.994011976047904e-05,
|
| 3033 |
+
"loss": 1.0019207000732422,
|
| 3034 |
+
"step": 432
|
| 3035 |
+
},
|
| 3036 |
+
{
|
| 3037 |
+
"epoch": 1.7118141374196738,
|
| 3038 |
+
"grad_norm": 0.1952063888311386,
|
| 3039 |
+
"learning_rate": 2.9540918163672654e-05,
|
| 3040 |
+
"loss": 0.7654832005500793,
|
| 3041 |
+
"step": 433
|
| 3042 |
+
},
|
| 3043 |
+
{
|
| 3044 |
+
"epoch": 1.7157686604053386,
|
| 3045 |
+
"grad_norm": 0.20689325034618378,
|
| 3046 |
+
"learning_rate": 2.9141716566866266e-05,
|
| 3047 |
+
"loss": 0.9330834746360779,
|
| 3048 |
+
"step": 434
|
| 3049 |
+
},
|
| 3050 |
+
{
|
| 3051 |
+
"epoch": 1.7197231833910034,
|
| 3052 |
+
"grad_norm": 0.20880067348480225,
|
| 3053 |
+
"learning_rate": 2.874251497005988e-05,
|
| 3054 |
+
"loss": 0.9945483803749084,
|
| 3055 |
+
"step": 435
|
| 3056 |
+
},
|
| 3057 |
+
{
|
| 3058 |
+
"epoch": 1.7236777063766682,
|
| 3059 |
+
"grad_norm": 0.18159128725528717,
|
| 3060 |
+
"learning_rate": 2.8343313373253494e-05,
|
| 3061 |
+
"loss": 0.9131766557693481,
|
| 3062 |
+
"step": 436
|
| 3063 |
+
},
|
| 3064 |
+
{
|
| 3065 |
+
"epoch": 1.7276322293623332,
|
| 3066 |
+
"grad_norm": 0.19038815796375275,
|
| 3067 |
+
"learning_rate": 2.7944111776447106e-05,
|
| 3068 |
+
"loss": 0.8274218440055847,
|
| 3069 |
+
"step": 437
|
| 3070 |
+
},
|
| 3071 |
+
{
|
| 3072 |
+
"epoch": 1.731586752347998,
|
| 3073 |
+
"grad_norm": 0.20182478427886963,
|
| 3074 |
+
"learning_rate": 2.754491017964072e-05,
|
| 3075 |
+
"loss": 0.8425199389457703,
|
| 3076 |
+
"step": 438
|
| 3077 |
+
},
|
| 3078 |
+
{
|
| 3079 |
+
"epoch": 1.735541275333663,
|
| 3080 |
+
"grad_norm": 0.20511551201343536,
|
| 3081 |
+
"learning_rate": 2.7145708582834333e-05,
|
| 3082 |
+
"loss": 0.8865355253219604,
|
| 3083 |
+
"step": 439
|
| 3084 |
+
},
|
| 3085 |
+
{
|
| 3086 |
+
"epoch": 1.7394957983193278,
|
| 3087 |
+
"grad_norm": 0.2129596769809723,
|
| 3088 |
+
"learning_rate": 2.6746506986027946e-05,
|
| 3089 |
+
"loss": 0.9642688035964966,
|
| 3090 |
+
"step": 440
|
| 3091 |
+
},
|
| 3092 |
+
{
|
| 3093 |
+
"epoch": 1.7434503213049926,
|
| 3094 |
+
"grad_norm": 0.20142444968223572,
|
| 3095 |
+
"learning_rate": 2.634730538922156e-05,
|
| 3096 |
+
"loss": 1.0012695789337158,
|
| 3097 |
+
"step": 441
|
| 3098 |
+
},
|
| 3099 |
+
{
|
| 3100 |
+
"epoch": 1.7474048442906573,
|
| 3101 |
+
"grad_norm": 0.20275944471359253,
|
| 3102 |
+
"learning_rate": 2.594810379241517e-05,
|
| 3103 |
+
"loss": 0.8509343266487122,
|
| 3104 |
+
"step": 442
|
| 3105 |
+
},
|
| 3106 |
+
{
|
| 3107 |
+
"epoch": 1.7513593672763221,
|
| 3108 |
+
"grad_norm": 0.20978280901908875,
|
| 3109 |
+
"learning_rate": 2.5548902195608782e-05,
|
| 3110 |
+
"loss": 0.8017287850379944,
|
| 3111 |
+
"step": 443
|
| 3112 |
+
},
|
| 3113 |
+
{
|
| 3114 |
+
"epoch": 1.7553138902619871,
|
| 3115 |
+
"grad_norm": 0.2030046135187149,
|
| 3116 |
+
"learning_rate": 2.5149700598802394e-05,
|
| 3117 |
+
"loss": 0.9278156757354736,
|
| 3118 |
+
"step": 444
|
| 3119 |
+
},
|
| 3120 |
+
{
|
| 3121 |
+
"epoch": 1.7592684132476522,
|
| 3122 |
+
"grad_norm": 0.2252151072025299,
|
| 3123 |
+
"learning_rate": 2.475049900199601e-05,
|
| 3124 |
+
"loss": 0.9012824296951294,
|
| 3125 |
+
"step": 445
|
| 3126 |
+
},
|
| 3127 |
+
{
|
| 3128 |
+
"epoch": 1.763222936233317,
|
| 3129 |
+
"grad_norm": 0.20253197848796844,
|
| 3130 |
+
"learning_rate": 2.4351297405189622e-05,
|
| 3131 |
+
"loss": 0.8817086815834045,
|
| 3132 |
+
"step": 446
|
| 3133 |
+
},
|
| 3134 |
+
{
|
| 3135 |
+
"epoch": 1.7671774592189817,
|
| 3136 |
+
"grad_norm": 0.22054298222064972,
|
| 3137 |
+
"learning_rate": 2.3952095808383234e-05,
|
| 3138 |
+
"loss": 0.9003196954727173,
|
| 3139 |
+
"step": 447
|
| 3140 |
+
},
|
| 3141 |
+
{
|
| 3142 |
+
"epoch": 1.7711319822046465,
|
| 3143 |
+
"grad_norm": 0.2037447839975357,
|
| 3144 |
+
"learning_rate": 2.355289421157685e-05,
|
| 3145 |
+
"loss": 0.8119622468948364,
|
| 3146 |
+
"step": 448
|
| 3147 |
+
},
|
| 3148 |
+
{
|
| 3149 |
+
"epoch": 1.7750865051903113,
|
| 3150 |
+
"grad_norm": 0.21892541646957397,
|
| 3151 |
+
"learning_rate": 2.315369261477046e-05,
|
| 3152 |
+
"loss": 0.9405473470687866,
|
| 3153 |
+
"step": 449
|
| 3154 |
+
},
|
| 3155 |
+
{
|
| 3156 |
+
"epoch": 1.7790410281759763,
|
| 3157 |
+
"grad_norm": 0.20719027519226074,
|
| 3158 |
+
"learning_rate": 2.275449101796407e-05,
|
| 3159 |
+
"loss": 0.8205680847167969,
|
| 3160 |
+
"step": 450
|
| 3161 |
+
},
|
| 3162 |
+
{
|
| 3163 |
+
"epoch": 1.782995551161641,
|
| 3164 |
+
"grad_norm": 0.20699091255664825,
|
| 3165 |
+
"learning_rate": 2.2355289421157686e-05,
|
| 3166 |
+
"loss": 0.8950562477111816,
|
| 3167 |
+
"step": 451
|
| 3168 |
+
},
|
| 3169 |
+
{
|
| 3170 |
+
"epoch": 1.7869500741473061,
|
| 3171 |
+
"grad_norm": 0.19834667444229126,
|
| 3172 |
+
"learning_rate": 2.1956087824351298e-05,
|
| 3173 |
+
"loss": 0.8222463130950928,
|
| 3174 |
+
"step": 452
|
| 3175 |
+
},
|
| 3176 |
+
{
|
| 3177 |
+
"epoch": 1.790904597132971,
|
| 3178 |
+
"grad_norm": 0.19083131849765778,
|
| 3179 |
+
"learning_rate": 2.155688622754491e-05,
|
| 3180 |
+
"loss": 0.9506749510765076,
|
| 3181 |
+
"step": 453
|
| 3182 |
+
},
|
| 3183 |
+
{
|
| 3184 |
+
"epoch": 1.7948591201186357,
|
| 3185 |
+
"grad_norm": 0.21305403113365173,
|
| 3186 |
+
"learning_rate": 2.1157684630738526e-05,
|
| 3187 |
+
"loss": 0.833589494228363,
|
| 3188 |
+
"step": 454
|
| 3189 |
+
},
|
| 3190 |
+
{
|
| 3191 |
+
"epoch": 1.7988136431043005,
|
| 3192 |
+
"grad_norm": 0.21756337583065033,
|
| 3193 |
+
"learning_rate": 2.0758483033932134e-05,
|
| 3194 |
+
"loss": 0.7631924152374268,
|
| 3195 |
+
"step": 455
|
| 3196 |
+
},
|
| 3197 |
+
{
|
| 3198 |
+
"epoch": 1.8027681660899653,
|
| 3199 |
+
"grad_norm": 0.20038093626499176,
|
| 3200 |
+
"learning_rate": 2.035928143712575e-05,
|
| 3201 |
+
"loss": 1.037887692451477,
|
| 3202 |
+
"step": 456
|
| 3203 |
+
},
|
| 3204 |
+
{
|
| 3205 |
+
"epoch": 1.8067226890756303,
|
| 3206 |
+
"grad_norm": 0.20598313212394714,
|
| 3207 |
+
"learning_rate": 1.9960079840319362e-05,
|
| 3208 |
+
"loss": 0.9344784617424011,
|
| 3209 |
+
"step": 457
|
| 3210 |
+
},
|
| 3211 |
+
{
|
| 3212 |
+
"epoch": 1.810677212061295,
|
| 3213 |
+
"grad_norm": 0.19507546722888947,
|
| 3214 |
+
"learning_rate": 1.9560878243512974e-05,
|
| 3215 |
+
"loss": 0.9052549600601196,
|
| 3216 |
+
"step": 458
|
| 3217 |
+
},
|
| 3218 |
+
{
|
| 3219 |
+
"epoch": 1.81463173504696,
|
| 3220 |
+
"grad_norm": 0.20818522572517395,
|
| 3221 |
+
"learning_rate": 1.916167664670659e-05,
|
| 3222 |
+
"loss": 0.8613417148590088,
|
| 3223 |
+
"step": 459
|
| 3224 |
+
},
|
| 3225 |
+
{
|
| 3226 |
+
"epoch": 1.8185862580326249,
|
| 3227 |
+
"grad_norm": 0.18870945274829865,
|
| 3228 |
+
"learning_rate": 1.87624750499002e-05,
|
| 3229 |
+
"loss": 0.793246865272522,
|
| 3230 |
+
"step": 460
|
| 3231 |
+
},
|
| 3232 |
+
{
|
| 3233 |
+
"epoch": 1.8225407810182896,
|
| 3234 |
+
"grad_norm": 0.21497206389904022,
|
| 3235 |
+
"learning_rate": 1.8363273453093814e-05,
|
| 3236 |
+
"loss": 0.9714541435241699,
|
| 3237 |
+
"step": 461
|
| 3238 |
+
},
|
| 3239 |
+
{
|
| 3240 |
+
"epoch": 1.8264953040039544,
|
| 3241 |
+
"grad_norm": 0.21383772790431976,
|
| 3242 |
+
"learning_rate": 1.7964071856287426e-05,
|
| 3243 |
+
"loss": 1.0128012895584106,
|
| 3244 |
+
"step": 462
|
| 3245 |
+
},
|
| 3246 |
+
{
|
| 3247 |
+
"epoch": 1.8304498269896192,
|
| 3248 |
+
"grad_norm": 0.21303774416446686,
|
| 3249 |
+
"learning_rate": 1.756487025948104e-05,
|
| 3250 |
+
"loss": 0.8802911639213562,
|
| 3251 |
+
"step": 463
|
| 3252 |
+
},
|
| 3253 |
+
{
|
| 3254 |
+
"epoch": 1.8344043499752842,
|
| 3255 |
+
"grad_norm": 0.18445076048374176,
|
| 3256 |
+
"learning_rate": 1.7165668662674654e-05,
|
| 3257 |
+
"loss": 1.0024535655975342,
|
| 3258 |
+
"step": 464
|
| 3259 |
+
},
|
| 3260 |
+
{
|
| 3261 |
+
"epoch": 1.8383588729609492,
|
| 3262 |
+
"grad_norm": 0.20212195813655853,
|
| 3263 |
+
"learning_rate": 1.6766467065868263e-05,
|
| 3264 |
+
"loss": 1.1197216510772705,
|
| 3265 |
+
"step": 465
|
| 3266 |
+
},
|
| 3267 |
+
{
|
| 3268 |
+
"epoch": 1.842313395946614,
|
| 3269 |
+
"grad_norm": 0.2088390588760376,
|
| 3270 |
+
"learning_rate": 1.6367265469061875e-05,
|
| 3271 |
+
"loss": 0.9632558226585388,
|
| 3272 |
+
"step": 466
|
| 3273 |
+
},
|
| 3274 |
+
{
|
| 3275 |
+
"epoch": 1.8462679189322788,
|
| 3276 |
+
"grad_norm": 0.20784492790699005,
|
| 3277 |
+
"learning_rate": 1.596806387225549e-05,
|
| 3278 |
+
"loss": 0.734897255897522,
|
| 3279 |
+
"step": 467
|
| 3280 |
+
},
|
| 3281 |
+
{
|
| 3282 |
+
"epoch": 1.8502224419179436,
|
| 3283 |
+
"grad_norm": 0.19716675579547882,
|
| 3284 |
+
"learning_rate": 1.5568862275449103e-05,
|
| 3285 |
+
"loss": 0.9311115145683289,
|
| 3286 |
+
"step": 468
|
| 3287 |
+
},
|
| 3288 |
+
{
|
| 3289 |
+
"epoch": 1.8541769649036084,
|
| 3290 |
+
"grad_norm": 0.20142972469329834,
|
| 3291 |
+
"learning_rate": 1.5169660678642716e-05,
|
| 3292 |
+
"loss": 0.9369468688964844,
|
| 3293 |
+
"step": 469
|
| 3294 |
+
},
|
| 3295 |
+
{
|
| 3296 |
+
"epoch": 1.8581314878892734,
|
| 3297 |
+
"grad_norm": 0.20825821161270142,
|
| 3298 |
+
"learning_rate": 1.4770459081836327e-05,
|
| 3299 |
+
"loss": 1.135632038116455,
|
| 3300 |
+
"step": 470
|
| 3301 |
+
},
|
| 3302 |
+
{
|
| 3303 |
+
"epoch": 1.8620860108749382,
|
| 3304 |
+
"grad_norm": 0.20531900227069855,
|
| 3305 |
+
"learning_rate": 1.437125748502994e-05,
|
| 3306 |
+
"loss": 0.817120373249054,
|
| 3307 |
+
"step": 471
|
| 3308 |
+
},
|
| 3309 |
+
{
|
| 3310 |
+
"epoch": 1.8660405338606032,
|
| 3311 |
+
"grad_norm": 0.2101070135831833,
|
| 3312 |
+
"learning_rate": 1.3972055888223553e-05,
|
| 3313 |
+
"loss": 0.8993910551071167,
|
| 3314 |
+
"step": 472
|
| 3315 |
+
},
|
| 3316 |
+
{
|
| 3317 |
+
"epoch": 1.869995056846268,
|
| 3318 |
+
"grad_norm": 0.1841268241405487,
|
| 3319 |
+
"learning_rate": 1.3572854291417167e-05,
|
| 3320 |
+
"loss": 0.9081541299819946,
|
| 3321 |
+
"step": 473
|
| 3322 |
+
},
|
| 3323 |
+
{
|
| 3324 |
+
"epoch": 1.8739495798319328,
|
| 3325 |
+
"grad_norm": 0.19929905235767365,
|
| 3326 |
+
"learning_rate": 1.317365269461078e-05,
|
| 3327 |
+
"loss": 0.8846563696861267,
|
| 3328 |
+
"step": 474
|
| 3329 |
+
},
|
| 3330 |
+
{
|
| 3331 |
+
"epoch": 1.8779041028175976,
|
| 3332 |
+
"grad_norm": 0.21067850291728973,
|
| 3333 |
+
"learning_rate": 1.2774451097804391e-05,
|
| 3334 |
+
"loss": 0.8508776426315308,
|
| 3335 |
+
"step": 475
|
| 3336 |
+
},
|
| 3337 |
+
{
|
| 3338 |
+
"epoch": 1.8818586258032624,
|
| 3339 |
+
"grad_norm": 0.19630905985832214,
|
| 3340 |
+
"learning_rate": 1.2375249500998005e-05,
|
| 3341 |
+
"loss": 0.9228495359420776,
|
| 3342 |
+
"step": 476
|
| 3343 |
+
},
|
| 3344 |
+
{
|
| 3345 |
+
"epoch": 1.8858131487889274,
|
| 3346 |
+
"grad_norm": 0.22471991181373596,
|
| 3347 |
+
"learning_rate": 1.1976047904191617e-05,
|
| 3348 |
+
"loss": 0.8360285758972168,
|
| 3349 |
+
"step": 477
|
| 3350 |
+
},
|
| 3351 |
+
{
|
| 3352 |
+
"epoch": 1.8897676717745922,
|
| 3353 |
+
"grad_norm": 0.18797798454761505,
|
| 3354 |
+
"learning_rate": 1.157684630738523e-05,
|
| 3355 |
+
"loss": 0.8709227442741394,
|
| 3356 |
+
"step": 478
|
| 3357 |
+
},
|
| 3358 |
+
{
|
| 3359 |
+
"epoch": 1.8937221947602572,
|
| 3360 |
+
"grad_norm": 0.20813153684139252,
|
| 3361 |
+
"learning_rate": 1.1177644710578843e-05,
|
| 3362 |
+
"loss": 1.0134263038635254,
|
| 3363 |
+
"step": 479
|
| 3364 |
+
},
|
| 3365 |
+
{
|
| 3366 |
+
"epoch": 1.897676717745922,
|
| 3367 |
+
"grad_norm": 0.20971933007240295,
|
| 3368 |
+
"learning_rate": 1.0778443113772455e-05,
|
| 3369 |
+
"loss": 0.8753843307495117,
|
| 3370 |
+
"step": 480
|
| 3371 |
+
},
|
| 3372 |
+
{
|
| 3373 |
+
"epoch": 1.9016312407315867,
|
| 3374 |
+
"grad_norm": 0.19115358591079712,
|
| 3375 |
+
"learning_rate": 1.0379241516966067e-05,
|
| 3376 |
+
"loss": 0.8529820442199707,
|
| 3377 |
+
"step": 481
|
| 3378 |
+
},
|
| 3379 |
+
{
|
| 3380 |
+
"epoch": 1.9055857637172515,
|
| 3381 |
+
"grad_norm": 0.23405560851097107,
|
| 3382 |
+
"learning_rate": 9.980039920159681e-06,
|
| 3383 |
+
"loss": 0.9483686685562134,
|
| 3384 |
+
"step": 482
|
| 3385 |
+
},
|
| 3386 |
+
{
|
| 3387 |
+
"epoch": 1.9095402867029163,
|
| 3388 |
+
"grad_norm": 0.23471501469612122,
|
| 3389 |
+
"learning_rate": 9.580838323353295e-06,
|
| 3390 |
+
"loss": 0.9428390860557556,
|
| 3391 |
+
"step": 483
|
| 3392 |
+
},
|
| 3393 |
+
{
|
| 3394 |
+
"epoch": 1.9134948096885813,
|
| 3395 |
+
"grad_norm": 0.21804015338420868,
|
| 3396 |
+
"learning_rate": 9.181636726546907e-06,
|
| 3397 |
+
"loss": 0.8681425452232361,
|
| 3398 |
+
"step": 484
|
| 3399 |
+
},
|
| 3400 |
+
{
|
| 3401 |
+
"epoch": 1.9174493326742463,
|
| 3402 |
+
"grad_norm": 0.20593298971652985,
|
| 3403 |
+
"learning_rate": 8.78243512974052e-06,
|
| 3404 |
+
"loss": 0.9854474067687988,
|
| 3405 |
+
"step": 485
|
| 3406 |
+
},
|
| 3407 |
+
{
|
| 3408 |
+
"epoch": 1.9214038556599111,
|
| 3409 |
+
"grad_norm": 0.20268182456493378,
|
| 3410 |
+
"learning_rate": 8.383233532934131e-06,
|
| 3411 |
+
"loss": 0.8769013285636902,
|
| 3412 |
+
"step": 486
|
| 3413 |
+
},
|
| 3414 |
+
{
|
| 3415 |
+
"epoch": 1.925358378645576,
|
| 3416 |
+
"grad_norm": 0.20166848599910736,
|
| 3417 |
+
"learning_rate": 7.984031936127745e-06,
|
| 3418 |
+
"loss": 0.8038483262062073,
|
| 3419 |
+
"step": 487
|
| 3420 |
+
},
|
| 3421 |
+
{
|
| 3422 |
+
"epoch": 1.9293129016312407,
|
| 3423 |
+
"grad_norm": 0.20438715815544128,
|
| 3424 |
+
"learning_rate": 7.584830339321358e-06,
|
| 3425 |
+
"loss": 0.8820595741271973,
|
| 3426 |
+
"step": 488
|
| 3427 |
+
},
|
| 3428 |
+
{
|
| 3429 |
+
"epoch": 1.9332674246169055,
|
| 3430 |
+
"grad_norm": 0.22562836110591888,
|
| 3431 |
+
"learning_rate": 7.18562874251497e-06,
|
| 3432 |
+
"loss": 0.8607326745986938,
|
| 3433 |
+
"step": 489
|
| 3434 |
+
},
|
| 3435 |
+
{
|
| 3436 |
+
"epoch": 1.9372219476025705,
|
| 3437 |
+
"grad_norm": 0.24032363295555115,
|
| 3438 |
+
"learning_rate": 6.786427145708583e-06,
|
| 3439 |
+
"loss": 0.9563621878623962,
|
| 3440 |
+
"step": 490
|
| 3441 |
+
},
|
| 3442 |
+
{
|
| 3443 |
+
"epoch": 1.9411764705882353,
|
| 3444 |
+
"grad_norm": 0.22053006291389465,
|
| 3445 |
+
"learning_rate": 6.3872255489021955e-06,
|
| 3446 |
+
"loss": 0.9206818342208862,
|
| 3447 |
+
"step": 491
|
| 3448 |
+
},
|
| 3449 |
+
{
|
| 3450 |
+
"epoch": 1.9451309935739003,
|
| 3451 |
+
"grad_norm": 0.201730415225029,
|
| 3452 |
+
"learning_rate": 5.9880239520958085e-06,
|
| 3453 |
+
"loss": 0.8101902604103088,
|
| 3454 |
+
"step": 492
|
| 3455 |
+
},
|
| 3456 |
+
{
|
| 3457 |
+
"epoch": 1.949085516559565,
|
| 3458 |
+
"grad_norm": 0.2032587230205536,
|
| 3459 |
+
"learning_rate": 5.5888223552894215e-06,
|
| 3460 |
+
"loss": 0.8570499420166016,
|
| 3461 |
+
"step": 493
|
| 3462 |
+
},
|
| 3463 |
+
{
|
| 3464 |
+
"epoch": 1.9530400395452299,
|
| 3465 |
+
"grad_norm": 0.21162763237953186,
|
| 3466 |
+
"learning_rate": 5.189620758483034e-06,
|
| 3467 |
+
"loss": 0.8345759510993958,
|
| 3468 |
+
"step": 494
|
| 3469 |
+
},
|
| 3470 |
+
{
|
| 3471 |
+
"epoch": 1.9569945625308947,
|
| 3472 |
+
"grad_norm": 0.19271045923233032,
|
| 3473 |
+
"learning_rate": 4.7904191616766475e-06,
|
| 3474 |
+
"loss": 0.9391037821769714,
|
| 3475 |
+
"step": 495
|
| 3476 |
+
},
|
| 3477 |
+
{
|
| 3478 |
+
"epoch": 1.9609490855165594,
|
| 3479 |
+
"grad_norm": 0.20637477934360504,
|
| 3480 |
+
"learning_rate": 4.39121756487026e-06,
|
| 3481 |
+
"loss": 0.8780249953269958,
|
| 3482 |
+
"step": 496
|
| 3483 |
+
},
|
| 3484 |
+
{
|
| 3485 |
+
"epoch": 1.9649036085022245,
|
| 3486 |
+
"grad_norm": 0.20201361179351807,
|
| 3487 |
+
"learning_rate": 3.992015968063873e-06,
|
| 3488 |
+
"loss": 0.9155791997909546,
|
| 3489 |
+
"step": 497
|
| 3490 |
+
},
|
| 3491 |
+
{
|
| 3492 |
+
"epoch": 1.9688581314878892,
|
| 3493 |
+
"grad_norm": 0.21115684509277344,
|
| 3494 |
+
"learning_rate": 3.592814371257485e-06,
|
| 3495 |
+
"loss": 0.7178623676300049,
|
| 3496 |
+
"step": 498
|
| 3497 |
+
},
|
| 3498 |
+
{
|
| 3499 |
+
"epoch": 1.9728126544735543,
|
| 3500 |
+
"grad_norm": 0.22672364115715027,
|
| 3501 |
+
"learning_rate": 3.1936127744510977e-06,
|
| 3502 |
+
"loss": 0.8261544704437256,
|
| 3503 |
+
"step": 499
|
| 3504 |
+
},
|
| 3505 |
+
{
|
| 3506 |
+
"epoch": 1.976767177459219,
|
| 3507 |
+
"grad_norm": 0.18823912739753723,
|
| 3508 |
+
"learning_rate": 2.7944111776447107e-06,
|
| 3509 |
+
"loss": 0.8679497241973877,
|
| 3510 |
+
"step": 500
|
| 3511 |
+
},
|
| 3512 |
+
{
|
| 3513 |
+
"epoch": 1.9807217004448838,
|
| 3514 |
+
"grad_norm": 0.2101755440235138,
|
| 3515 |
+
"learning_rate": 2.3952095808383237e-06,
|
| 3516 |
+
"loss": 0.950539231300354,
|
| 3517 |
+
"step": 501
|
| 3518 |
+
},
|
| 3519 |
+
{
|
| 3520 |
+
"epoch": 1.9846762234305486,
|
| 3521 |
+
"grad_norm": 0.21033525466918945,
|
| 3522 |
+
"learning_rate": 1.9960079840319363e-06,
|
| 3523 |
+
"loss": 0.7960910797119141,
|
| 3524 |
+
"step": 502
|
| 3525 |
+
},
|
| 3526 |
+
{
|
| 3527 |
+
"epoch": 1.9886307464162134,
|
| 3528 |
+
"grad_norm": 0.21546964347362518,
|
| 3529 |
+
"learning_rate": 1.5968063872255489e-06,
|
| 3530 |
+
"loss": 0.9860714077949524,
|
| 3531 |
+
"step": 503
|
| 3532 |
+
},
|
| 3533 |
+
{
|
| 3534 |
+
"epoch": 1.9925852694018784,
|
| 3535 |
+
"grad_norm": 0.19765757024288177,
|
| 3536 |
+
"learning_rate": 1.1976047904191619e-06,
|
| 3537 |
+
"loss": 1.025565505027771,
|
| 3538 |
+
"step": 504
|
| 3539 |
+
},
|
| 3540 |
+
{
|
| 3541 |
+
"epoch": 1.9965397923875432,
|
| 3542 |
+
"grad_norm": 0.20392237603664398,
|
| 3543 |
+
"learning_rate": 7.984031936127744e-07,
|
| 3544 |
+
"loss": 0.8897353410720825,
|
| 3545 |
+
"step": 505
|
| 3546 |
+
},
|
| 3547 |
+
{
|
| 3548 |
+
"epoch": 2.0,
|
| 3549 |
+
"grad_norm": 0.2248964011669159,
|
| 3550 |
+
"learning_rate": 3.992015968063872e-07,
|
| 3551 |
+
"loss": 0.7869975566864014,
|
| 3552 |
+
"step": 506
|
| 3553 |
+
}
|
| 3554 |
+
],
|
| 3555 |
+
"logging_steps": 1,
|
| 3556 |
+
"max_steps": 506,
|
| 3557 |
+
"num_input_tokens_seen": 0,
|
| 3558 |
+
"num_train_epochs": 2,
|
| 3559 |
+
"save_steps": 100,
|
| 3560 |
+
"stateful_callbacks": {
|
| 3561 |
+
"TrainerControl": {
|
| 3562 |
+
"args": {
|
| 3563 |
+
"should_epoch_stop": false,
|
| 3564 |
+
"should_evaluate": false,
|
| 3565 |
+
"should_log": false,
|
| 3566 |
+
"should_save": true,
|
| 3567 |
+
"should_training_stop": true
|
| 3568 |
+
},
|
| 3569 |
+
"attributes": {}
|
| 3570 |
+
}
|
| 3571 |
+
},
|
| 3572 |
+
"total_flos": 2.5380020501658665e+18,
|
| 3573 |
+
"train_batch_size": 8,
|
| 3574 |
+
"trial_name": null,
|
| 3575 |
+
"trial_params": null
|
| 3576 |
+
}
|