Hugging Face's logo

shanjay
/
mgc-ds

TensorBoard
Model card Files Metrics Community
shanjay commited on
Commit
2f590c8
•
1 Parent(s): bec7a44

Upload folder using huggingface_hub

Browse files
Files changed (28) hide show
  1. train/checkpoint-500/optimizer.pt +3 -0
  2. train/checkpoint-500/pytorch_model.bin +3 -0
  3. train/checkpoint-500/rng_state.pth +3 -0
  4. train/checkpoint-500/scheduler.pt +3 -0
  5. train/checkpoint-500/trainer_state.json +3016 -0
  6. train/checkpoint-500/training_args.bin +3 -0
  7. train/runs/Dec09_21-43-47_1252a6b6660f/1702158227.8969784/events.out.tfevents.1702158227.1252a6b6660f +3 -0
  8. train/runs/Dec09_21-43-47_1252a6b6660f/events.out.tfevents.1702158227.1252a6b6660f +3 -0
  9. train/runs/Dec09_21-44-36_1252a6b6660f/1702158276.6846008/events.out.tfevents.1702158276.1252a6b6660f +3 -0
  10. train/runs/Dec09_21-44-36_1252a6b6660f/events.out.tfevents.1702158276.1252a6b6660f +3 -0
  11. train/runs/Dec09_21-45-42_1252a6b6660f/1702158343.1430557/events.out.tfevents.1702158343.1252a6b6660f.4035.1 +3 -0
  12. train/runs/Dec09_21-45-42_1252a6b6660f/events.out.tfevents.1702158343.1252a6b6660f.4035.0 +3 -0
  13. train/runs/Dec09_21-45-55_1252a6b6660f/1702158355.5142772/events.out.tfevents.1702158355.1252a6b6660f.4035.3 +3 -0
  14. train/runs/Dec09_21-45-55_1252a6b6660f/events.out.tfevents.1702158355.1252a6b6660f.4035.2 +3 -0
  15. train/runs/Dec09_21-46-30_1252a6b6660f/1702158390.361645/events.out.tfevents.1702158390.1252a6b6660f.4035.5 +3 -0
  16. train/runs/Dec09_21-46-30_1252a6b6660f/events.out.tfevents.1702158390.1252a6b6660f.4035.4 +3 -0
  17. train/runs/Dec09_22-20-54_1252a6b6660f/1702160454.3142843/events.out.tfevents.1702160454.1252a6b6660f.4035.7 +3 -0
  18. train/runs/Dec09_22-20-54_1252a6b6660f/events.out.tfevents.1702160454.1252a6b6660f.4035.6 +3 -0
  19. train/runs/Dec09_22-21-11_1252a6b6660f/1702160471.8232346/events.out.tfevents.1702160471.1252a6b6660f.4035.9 +3 -0
  20. train/runs/Dec09_22-21-11_1252a6b6660f/events.out.tfevents.1702160471.1252a6b6660f.4035.8 +3 -0
  21. train/runs/Dec09_22-22-57_1252a6b6660f/1702160577.833262/events.out.tfevents.1702160577.1252a6b6660f.16555.1 +3 -0
  22. train/runs/Dec09_22-22-57_1252a6b6660f/events.out.tfevents.1702160577.1252a6b6660f.16555.0 +3 -0
  23. train/runs/Dec09_22-23-50_1252a6b6660f/1702160631.116859/events.out.tfevents.1702160631.1252a6b6660f.16555.3 +3 -0
  24. train/runs/Dec09_22-23-50_1252a6b6660f/events.out.tfevents.1702160631.1252a6b6660f.16555.2 +3 -0
  25. train/runs/Dec09_23-07-43_1252a6b6660f/1702163264.08709/events.out.tfevents.1702163264.1252a6b6660f.29008.1 +3 -0
  26. train/runs/Dec09_23-07-43_1252a6b6660f/events.out.tfevents.1702163264.1252a6b6660f.29008.0 +3 -0
  27. train/runs/Dec10_10-09-29_2b7f3ae7a72d/1702202970.7887614/events.out.tfevents.1702202970.2b7f3ae7a72d.3821.1 +3 -0
  28. train/runs/Dec10_10-09-29_2b7f3ae7a72d/events.out.tfevents.1702202970.2b7f3ae7a72d.3821.0 +3 -0
train/checkpoint-500/optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c5ef4632d6fe1f46088e9bad0a9bfe55c3d29be294e4b05a615aa34dea8f1fba
3
+ size 16962618
train/checkpoint-500/pytorch_model.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:a8ba7560d681817acf04e523bb991d742183c6fdc75fd6a8552d950fa0539806
3
+ size 4329833190
train/checkpoint-500/rng_state.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:2417ca57d43f178a4765c78011b314ec1064d3e4f25c2787ae31bc927de73429
3
+ size 14244
train/checkpoint-500/scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:70fd9d3e8c2d24b7c67dc28dcc82da976bf7ed853f547683ebf5f1ce5d499bb9
3
+ size 1064
train/checkpoint-500/trainer_state.json ADDED
@@ -0,0 +1,3016 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": null,
3
+ "best_model_checkpoint": null,
4
+ "epoch": 2.229654403567447,
5
+ "global_step": 500,
6
+ "is_hyper_param_search": false,
7
+ "is_local_process_zero": true,
8
+ "is_world_process_zero": true,
9
+ "log_history": [
10
+ {
11
+ "epoch": 0.0,
12
+ "learning_rate": 0.0,
13
+ "loss": 9.2554,
14
+ "step": 1
15
+ },
16
+ {
17
+ "epoch": 0.01,
18
+ "learning_rate": 6.25e-06,
19
+ "loss": 9.5951,
20
+ "step": 2
21
+ },
22
+ {
23
+ "epoch": 0.01,
24
+ "learning_rate": 1.25e-05,
25
+ "loss": 9.968,
26
+ "step": 3
27
+ },
28
+ {
29
+ "epoch": 0.02,
30
+ "learning_rate": 1.8750000000000002e-05,
31
+ "loss": 9.3248,
32
+ "step": 4
33
+ },
34
+ {
35
+ "epoch": 0.02,
36
+ "learning_rate": 2.5e-05,
37
+ "loss": 8.4478,
38
+ "step": 5
39
+ },
40
+ {
41
+ "epoch": 0.03,
42
+ "learning_rate": 3.125e-05,
43
+ "loss": 8.8715,
44
+ "step": 6
45
+ },
46
+ {
47
+ "epoch": 0.03,
48
+ "learning_rate": 3.7500000000000003e-05,
49
+ "loss": 9.1145,
50
+ "step": 7
51
+ },
52
+ {
53
+ "epoch": 0.04,
54
+ "learning_rate": 4.375e-05,
55
+ "loss": 8.4062,
56
+ "step": 8
57
+ },
58
+ {
59
+ "epoch": 0.04,
60
+ "learning_rate": 5e-05,
61
+ "loss": 8.7625,
62
+ "step": 9
63
+ },
64
+ {
65
+ "epoch": 0.04,
66
+ "learning_rate": 5.6250000000000005e-05,
67
+ "loss": 7.9871,
68
+ "step": 10
69
+ },
70
+ {
71
+ "epoch": 0.05,
72
+ "learning_rate": 6.25e-05,
73
+ "loss": 7.8684,
74
+ "step": 11
75
+ },
76
+ {
77
+ "epoch": 0.05,
78
+ "learning_rate": 6.875e-05,
79
+ "loss": 7.4262,
80
+ "step": 12
81
+ },
82
+ {
83
+ "epoch": 0.06,
84
+ "learning_rate": 7.500000000000001e-05,
85
+ "loss": 7.0718,
86
+ "step": 13
87
+ },
88
+ {
89
+ "epoch": 0.06,
90
+ "learning_rate": 8.125000000000001e-05,
91
+ "loss": 6.7281,
92
+ "step": 14
93
+ },
94
+ {
95
+ "epoch": 0.07,
96
+ "learning_rate": 8.75e-05,
97
+ "loss": 6.6335,
98
+ "step": 15
99
+ },
100
+ {
101
+ "epoch": 0.07,
102
+ "learning_rate": 9.375e-05,
103
+ "loss": 5.8393,
104
+ "step": 16
105
+ },
106
+ {
107
+ "epoch": 0.08,
108
+ "learning_rate": 0.0001,
109
+ "loss": 5.3873,
110
+ "step": 17
111
+ },
112
+ {
113
+ "epoch": 0.08,
114
+ "learning_rate": 0.00010625000000000001,
115
+ "loss": 4.9589,
116
+ "step": 18
117
+ },
118
+ {
119
+ "epoch": 0.08,
120
+ "learning_rate": 0.00011250000000000001,
121
+ "loss": 4.7494,
122
+ "step": 19
123
+ },
124
+ {
125
+ "epoch": 0.09,
126
+ "learning_rate": 0.00011875,
127
+ "loss": 4.3132,
128
+ "step": 20
129
+ },
130
+ {
131
+ "epoch": 0.09,
132
+ "learning_rate": 0.000125,
133
+ "loss": 4.1908,
134
+ "step": 21
135
+ },
136
+ {
137
+ "epoch": 0.1,
138
+ "learning_rate": 0.00013125000000000002,
139
+ "loss": 3.8402,
140
+ "step": 22
141
+ },
142
+ {
143
+ "epoch": 0.1,
144
+ "learning_rate": 0.0001375,
145
+ "loss": 3.4314,
146
+ "step": 23
147
+ },
148
+ {
149
+ "epoch": 0.11,
150
+ "learning_rate": 0.00014375,
151
+ "loss": 3.2802,
152
+ "step": 24
153
+ },
154
+ {
155
+ "epoch": 0.11,
156
+ "learning_rate": 0.00015000000000000001,
157
+ "loss": 3.1288,
158
+ "step": 25
159
+ },
160
+ {
161
+ "epoch": 0.12,
162
+ "learning_rate": 0.00015625,
163
+ "loss": 3.0646,
164
+ "step": 26
165
+ },
166
+ {
167
+ "epoch": 0.12,
168
+ "learning_rate": 0.00016250000000000002,
169
+ "loss": 3.0644,
170
+ "step": 27
171
+ },
172
+ {
173
+ "epoch": 0.12,
174
+ "learning_rate": 0.00016875,
175
+ "loss": 2.4856,
176
+ "step": 28
177
+ },
178
+ {
179
+ "epoch": 0.13,
180
+ "learning_rate": 0.000175,
181
+ "loss": 2.3275,
182
+ "step": 29
183
+ },
184
+ {
185
+ "epoch": 0.13,
186
+ "learning_rate": 0.00018125000000000001,
187
+ "loss": 2.1863,
188
+ "step": 30
189
+ },
190
+ {
191
+ "epoch": 0.14,
192
+ "learning_rate": 0.0001875,
193
+ "loss": 1.8291,
194
+ "step": 31
195
+ },
196
+ {
197
+ "epoch": 0.14,
198
+ "learning_rate": 0.00019375000000000002,
199
+ "loss": 1.7732,
200
+ "step": 32
201
+ },
202
+ {
203
+ "epoch": 0.15,
204
+ "learning_rate": 0.0002,
205
+ "loss": 1.6838,
206
+ "step": 33
207
+ },
208
+ {
209
+ "epoch": 0.15,
210
+ "learning_rate": 0.00019999866506037345,
211
+ "loss": 1.3643,
212
+ "step": 34
213
+ },
214
+ {
215
+ "epoch": 0.16,
216
+ "learning_rate": 0.00019999466027713507,
217
+ "loss": 1.4643,
218
+ "step": 35
219
+ },
220
+ {
221
+ "epoch": 0.16,
222
+ "learning_rate": 0.00019998798575720776,
223
+ "loss": 1.6028,
224
+ "step": 36
225
+ },
226
+ {
227
+ "epoch": 0.16,
228
+ "learning_rate": 0.00019997864167879312,
229
+ "loss": 1.4206,
230
+ "step": 37
231
+ },
232
+ {
233
+ "epoch": 0.17,
234
+ "learning_rate": 0.00019996662829136676,
235
+ "loss": 1.4531,
236
+ "step": 38
237
+ },
238
+ {
239
+ "epoch": 0.17,
240
+ "learning_rate": 0.0001999519459156716,
241
+ "loss": 1.2041,
242
+ "step": 39
243
+ },
244
+ {
245
+ "epoch": 0.18,
246
+ "learning_rate": 0.0001999345949437094,
247
+ "loss": 1.1949,
248
+ "step": 40
249
+ },
250
+ {
251
+ "epoch": 0.18,
252
+ "learning_rate": 0.0001999145758387301,
253
+ "loss": 0.9885,
254
+ "step": 41
255
+ },
256
+ {
257
+ "epoch": 0.19,
258
+ "learning_rate": 0.0001998918891352197,
259
+ "loss": 1.1611,
260
+ "step": 42
261
+ },
262
+ {
263
+ "epoch": 0.19,
264
+ "learning_rate": 0.00019986653543888568,
265
+ "loss": 0.9241,
266
+ "step": 43
267
+ },
268
+ {
269
+ "epoch": 0.2,
270
+ "learning_rate": 0.00019983851542664126,
271
+ "loss": 1.3583,
272
+ "step": 44
273
+ },
274
+ {
275
+ "epoch": 0.2,
276
+ "learning_rate": 0.00019980782984658683,
277
+ "loss": 1.1377,
278
+ "step": 45
279
+ },
280
+ {
281
+ "epoch": 0.21,
282
+ "learning_rate": 0.00019977447951799034,
283
+ "loss": 1.2204,
284
+ "step": 46
285
+ },
286
+ {
287
+ "epoch": 0.21,
288
+ "learning_rate": 0.00019973846533126533,
289
+ "loss": 0.9928,
290
+ "step": 47
291
+ },
292
+ {
293
+ "epoch": 0.21,
294
+ "learning_rate": 0.00019969978824794707,
295
+ "loss": 1.0598,
296
+ "step": 48
297
+ },
298
+ {
299
+ "epoch": 0.22,
300
+ "learning_rate": 0.000199658449300667,
301
+ "loss": 1.25,
302
+ "step": 49
303
+ },
304
+ {
305
+ "epoch": 0.22,
306
+ "learning_rate": 0.00019961444959312508,
307
+ "loss": 1.3713,
308
+ "step": 50
309
+ },
310
+ {
311
+ "epoch": 0.23,
312
+ "learning_rate": 0.0001995677903000604,
313
+ "loss": 0.9969,
314
+ "step": 51
315
+ },
316
+ {
317
+ "epoch": 0.23,
318
+ "learning_rate": 0.0001995184726672197,
319
+ "loss": 0.9356,
320
+ "step": 52
321
+ },
322
+ {
323
+ "epoch": 0.24,
324
+ "learning_rate": 0.00019946649801132427,
325
+ "loss": 0.834,
326
+ "step": 53
327
+ },
328
+ {
329
+ "epoch": 0.24,
330
+ "learning_rate": 0.00019941186772003464,
331
+ "loss": 0.9744,
332
+ "step": 54
333
+ },
334
+ {
335
+ "epoch": 0.25,
336
+ "learning_rate": 0.00019935458325191365,
337
+ "loss": 0.8644,
338
+ "step": 55
339
+ },
340
+ {
341
+ "epoch": 0.25,
342
+ "learning_rate": 0.0001992946461363874,
343
+ "loss": 1.1741,
344
+ "step": 56
345
+ },
346
+ {
347
+ "epoch": 0.25,
348
+ "learning_rate": 0.0001992320579737045,
349
+ "loss": 0.7969,
350
+ "step": 57
351
+ },
352
+ {
353
+ "epoch": 0.26,
354
+ "learning_rate": 0.00019916682043489336,
355
+ "loss": 0.9504,
356
+ "step": 58
357
+ },
358
+ {
359
+ "epoch": 0.26,
360
+ "learning_rate": 0.00019909893526171745,
361
+ "loss": 1.3963,
362
+ "step": 59
363
+ },
364
+ {
365
+ "epoch": 0.27,
366
+ "learning_rate": 0.00019902840426662895,
367
+ "loss": 1.0668,
368
+ "step": 60
369
+ },
370
+ {
371
+ "epoch": 0.27,
372
+ "learning_rate": 0.00019902840426662895,
373
+ "loss": 0.6902,
374
+ "step": 61
375
+ },
376
+ {
377
+ "epoch": 0.28,
378
+ "learning_rate": 0.00019895522933272028,
379
+ "loss": 1.2662,
380
+ "step": 62
381
+ },
382
+ {
383
+ "epoch": 0.28,
384
+ "learning_rate": 0.00019895522933272028,
385
+ "loss": 2.0447,
386
+ "step": 63
387
+ },
388
+ {
389
+ "epoch": 0.29,
390
+ "learning_rate": 0.00019895522933272028,
391
+ "loss": 0.7568,
392
+ "step": 64
393
+ },
394
+ {
395
+ "epoch": 0.29,
396
+ "learning_rate": 0.00019895522933272028,
397
+ "loss": 0.9146,
398
+ "step": 65
399
+ },
400
+ {
401
+ "epoch": 0.29,
402
+ "learning_rate": 0.00019887941241367377,
403
+ "loss": 1.1332,
404
+ "step": 66
405
+ },
406
+ {
407
+ "epoch": 0.3,
408
+ "learning_rate": 0.00019880095553370967,
409
+ "loss": 1.1715,
410
+ "step": 67
411
+ },
412
+ {
413
+ "epoch": 0.3,
414
+ "learning_rate": 0.0001987198607875319,
415
+ "loss": 1.3658,
416
+ "step": 68
417
+ },
418
+ {
419
+ "epoch": 0.31,
420
+ "learning_rate": 0.0001987198607875319,
421
+ "loss": 0.6815,
422
+ "step": 69
423
+ },
424
+ {
425
+ "epoch": 0.31,
426
+ "learning_rate": 0.0001987198607875319,
427
+ "loss": 0.0,
428
+ "step": 70
429
+ },
430
+ {
431
+ "epoch": 0.32,
432
+ "learning_rate": 0.0001987198607875319,
433
+ "loss": 0.635,
434
+ "step": 71
435
+ },
436
+ {
437
+ "epoch": 0.32,
438
+ "learning_rate": 0.0001987198607875319,
439
+ "loss": 0.5615,
440
+ "step": 72
441
+ },
442
+ {
443
+ "epoch": 0.33,
444
+ "learning_rate": 0.0001987198607875319,
445
+ "loss": 0.7602,
446
+ "step": 73
447
+ },
448
+ {
449
+ "epoch": 0.33,
450
+ "learning_rate": 0.0001987198607875319,
451
+ "loss": 0.6762,
452
+ "step": 74
453
+ },
454
+ {
455
+ "epoch": 0.33,
456
+ "learning_rate": 0.0001987198607875319,
457
+ "loss": 0.5942,
458
+ "step": 75
459
+ },
460
+ {
461
+ "epoch": 0.34,
462
+ "learning_rate": 0.0001987198607875319,
463
+ "loss": 1.3533,
464
+ "step": 76
465
+ },
466
+ {
467
+ "epoch": 0.34,
468
+ "learning_rate": 0.0001987198607875319,
469
+ "loss": 0.0,
470
+ "step": 77
471
+ },
472
+ {
473
+ "epoch": 0.35,
474
+ "learning_rate": 0.0001987198607875319,
475
+ "loss": 1.4797,
476
+ "step": 78
477
+ },
478
+ {
479
+ "epoch": 0.35,
480
+ "learning_rate": 0.0001987198607875319,
481
+ "loss": 1.0598,
482
+ "step": 79
483
+ },
484
+ {
485
+ "epoch": 0.36,
486
+ "learning_rate": 0.0001987198607875319,
487
+ "loss": 0.8546,
488
+ "step": 80
489
+ },
490
+ {
491
+ "epoch": 0.36,
492
+ "learning_rate": 0.0001987198607875319,
493
+ "loss": 0.5734,
494
+ "step": 81
495
+ },
496
+ {
497
+ "epoch": 0.37,
498
+ "learning_rate": 0.0001987198607875319,
499
+ "loss": 0.0866,
500
+ "step": 82
501
+ },
502
+ {
503
+ "epoch": 0.37,
504
+ "learning_rate": 0.0001987198607875319,
505
+ "loss": 0.238,
506
+ "step": 83
507
+ },
508
+ {
509
+ "epoch": 0.37,
510
+ "learning_rate": 0.0001987198607875319,
511
+ "loss": 0.8479,
512
+ "step": 84
513
+ },
514
+ {
515
+ "epoch": 0.38,
516
+ "learning_rate": 0.0001987198607875319,
517
+ "loss": 1.0991,
518
+ "step": 85
519
+ },
520
+ {
521
+ "epoch": 0.38,
522
+ "learning_rate": 0.0001987198607875319,
523
+ "loss": 1.9502,
524
+ "step": 86
525
+ },
526
+ {
527
+ "epoch": 0.39,
528
+ "learning_rate": 0.0001987198607875319,
529
+ "loss": 0.5053,
530
+ "step": 87
531
+ },
532
+ {
533
+ "epoch": 0.39,
534
+ "learning_rate": 0.0001987198607875319,
535
+ "loss": 0.4622,
536
+ "step": 88
537
+ },
538
+ {
539
+ "epoch": 0.4,
540
+ "learning_rate": 0.0001987198607875319,
541
+ "loss": 0.349,
542
+ "step": 89
543
+ },
544
+ {
545
+ "epoch": 0.4,
546
+ "learning_rate": 0.0001987198607875319,
547
+ "loss": 0.0,
548
+ "step": 90
549
+ },
550
+ {
551
+ "epoch": 0.41,
552
+ "learning_rate": 0.0001987198607875319,
553
+ "loss": 0.8208,
554
+ "step": 91
555
+ },
556
+ {
557
+ "epoch": 0.41,
558
+ "learning_rate": 0.0001987198607875319,
559
+ "loss": 0.5862,
560
+ "step": 92
561
+ },
562
+ {
563
+ "epoch": 0.41,
564
+ "learning_rate": 0.0001987198607875319,
565
+ "loss": 0.1249,
566
+ "step": 93
567
+ },
568
+ {
569
+ "epoch": 0.42,
570
+ "learning_rate": 0.0001987198607875319,
571
+ "loss": 0.8549,
572
+ "step": 94
573
+ },
574
+ {
575
+ "epoch": 0.42,
576
+ "learning_rate": 0.0001987198607875319,
577
+ "loss": 0.76,
578
+ "step": 95
579
+ },
580
+ {
581
+ "epoch": 0.43,
582
+ "learning_rate": 0.0001987198607875319,
583
+ "loss": 0.4468,
584
+ "step": 96
585
+ },
586
+ {
587
+ "epoch": 0.43,
588
+ "learning_rate": 0.0001987198607875319,
589
+ "loss": 1.3109,
590
+ "step": 97
591
+ },
592
+ {
593
+ "epoch": 0.44,
594
+ "learning_rate": 0.0001987198607875319,
595
+ "loss": 0.8511,
596
+ "step": 98
597
+ },
598
+ {
599
+ "epoch": 0.44,
600
+ "learning_rate": 0.0001987198607875319,
601
+ "loss": 0.9792,
602
+ "step": 99
603
+ },
604
+ {
605
+ "epoch": 0.45,
606
+ "learning_rate": 0.0001987198607875319,
607
+ "loss": 1.1989,
608
+ "step": 100
609
+ },
610
+ {
611
+ "epoch": 0.45,
612
+ "learning_rate": 0.0001987198607875319,
613
+ "loss": 0.0,
614
+ "step": 101
615
+ },
616
+ {
617
+ "epoch": 0.45,
618
+ "learning_rate": 0.0001987198607875319,
619
+ "loss": 0.438,
620
+ "step": 102
621
+ },
622
+ {
623
+ "epoch": 0.46,
624
+ "learning_rate": 0.0001987198607875319,
625
+ "loss": 0.3758,
626
+ "step": 103
627
+ },
628
+ {
629
+ "epoch": 0.46,
630
+ "learning_rate": 0.0001987198607875319,
631
+ "loss": 1.2202,
632
+ "step": 104
633
+ },
634
+ {
635
+ "epoch": 0.47,
636
+ "learning_rate": 0.0001987198607875319,
637
+ "loss": 1.6621,
638
+ "step": 105
639
+ },
640
+ {
641
+ "epoch": 0.47,
642
+ "learning_rate": 0.0001987198607875319,
643
+ "loss": 1.695,
644
+ "step": 106
645
+ },
646
+ {
647
+ "epoch": 0.48,
648
+ "learning_rate": 0.00019863613034027224,
649
+ "loss": 0.6176,
650
+ "step": 107
651
+ },
652
+ {
653
+ "epoch": 0.48,
654
+ "learning_rate": 0.00019863613034027224,
655
+ "loss": 1.211,
656
+ "step": 108
657
+ },
658
+ {
659
+ "epoch": 0.49,
660
+ "learning_rate": 0.00019863613034027224,
661
+ "loss": 1.0064,
662
+ "step": 109
663
+ },
664
+ {
665
+ "epoch": 0.49,
666
+ "learning_rate": 0.00019863613034027224,
667
+ "loss": 0.0,
668
+ "step": 110
669
+ },
670
+ {
671
+ "epoch": 0.49,
672
+ "learning_rate": 0.00019863613034027224,
673
+ "loss": 3.5679,
674
+ "step": 111
675
+ },
676
+ {
677
+ "epoch": 0.5,
678
+ "learning_rate": 0.00019863613034027224,
679
+ "loss": 0.0,
680
+ "step": 112
681
+ },
682
+ {
683
+ "epoch": 0.5,
684
+ "learning_rate": 0.00019863613034027224,
685
+ "loss": 0.0,
686
+ "step": 113
687
+ },
688
+ {
689
+ "epoch": 0.51,
690
+ "learning_rate": 0.00019863613034027224,
691
+ "loss": 0.227,
692
+ "step": 114
693
+ },
694
+ {
695
+ "epoch": 0.51,
696
+ "learning_rate": 0.00019863613034027224,
697
+ "loss": 0.0,
698
+ "step": 115
699
+ },
700
+ {
701
+ "epoch": 0.52,
702
+ "learning_rate": 0.00019863613034027224,
703
+ "loss": 0.0,
704
+ "step": 116
705
+ },
706
+ {
707
+ "epoch": 0.52,
708
+ "learning_rate": 0.00019863613034027224,
709
+ "loss": 1.4384,
710
+ "step": 117
711
+ },
712
+ {
713
+ "epoch": 0.53,
714
+ "learning_rate": 0.00019863613034027224,
715
+ "loss": 0.0,
716
+ "step": 118
717
+ },
718
+ {
719
+ "epoch": 0.53,
720
+ "learning_rate": 0.00019863613034027224,
721
+ "loss": 0.0,
722
+ "step": 119
723
+ },
724
+ {
725
+ "epoch": 0.54,
726
+ "learning_rate": 0.00019863613034027224,
727
+ "loss": 0.2177,
728
+ "step": 120
729
+ },
730
+ {
731
+ "epoch": 0.54,
732
+ "learning_rate": 0.00019863613034027224,
733
+ "loss": 0.4402,
734
+ "step": 121
735
+ },
736
+ {
737
+ "epoch": 0.54,
738
+ "learning_rate": 0.00019863613034027224,
739
+ "loss": 0.744,
740
+ "step": 122
741
+ },
742
+ {
743
+ "epoch": 0.55,
744
+ "learning_rate": 0.00019863613034027224,
745
+ "loss": 0.5766,
746
+ "step": 123
747
+ },
748
+ {
749
+ "epoch": 0.55,
750
+ "learning_rate": 0.00019863613034027224,
751
+ "loss": 0.1018,
752
+ "step": 124
753
+ },
754
+ {
755
+ "epoch": 0.56,
756
+ "learning_rate": 0.00019863613034027224,
757
+ "loss": 0.0,
758
+ "step": 125
759
+ },
760
+ {
761
+ "epoch": 0.56,
762
+ "learning_rate": 0.00019863613034027224,
763
+ "loss": 0.0,
764
+ "step": 126
765
+ },
766
+ {
767
+ "epoch": 0.57,
768
+ "learning_rate": 0.00019863613034027224,
769
+ "loss": 0.0,
770
+ "step": 127
771
+ },
772
+ {
773
+ "epoch": 0.57,
774
+ "learning_rate": 0.00019863613034027224,
775
+ "loss": 0.0,
776
+ "step": 128
777
+ },
778
+ {
779
+ "epoch": 0.58,
780
+ "learning_rate": 0.00019863613034027224,
781
+ "loss": 0.0,
782
+ "step": 129
783
+ },
784
+ {
785
+ "epoch": 0.58,
786
+ "learning_rate": 0.00019863613034027224,
787
+ "loss": 0.8572,
788
+ "step": 130
789
+ },
790
+ {
791
+ "epoch": 0.58,
792
+ "learning_rate": 0.00019863613034027224,
793
+ "loss": 0.0,
794
+ "step": 131
795
+ },
796
+ {
797
+ "epoch": 0.59,
798
+ "learning_rate": 0.00019863613034027224,
799
+ "loss": 0.0,
800
+ "step": 132
801
+ },
802
+ {
803
+ "epoch": 0.59,
804
+ "learning_rate": 0.00019863613034027224,
805
+ "loss": 0.2196,
806
+ "step": 133
807
+ },
808
+ {
809
+ "epoch": 0.6,
810
+ "learning_rate": 0.00019863613034027224,
811
+ "loss": 0.4427,
812
+ "step": 134
813
+ },
814
+ {
815
+ "epoch": 0.6,
816
+ "learning_rate": 0.00019863613034027224,
817
+ "loss": 0.1912,
818
+ "step": 135
819
+ },
820
+ {
821
+ "epoch": 0.61,
822
+ "learning_rate": 0.00019863613034027224,
823
+ "loss": 0.8012,
824
+ "step": 136
825
+ },
826
+ {
827
+ "epoch": 0.61,
828
+ "learning_rate": 0.00019863613034027224,
829
+ "loss": 0.0,
830
+ "step": 137
831
+ },
832
+ {
833
+ "epoch": 0.62,
834
+ "learning_rate": 0.00019863613034027224,
835
+ "loss": 0.0,
836
+ "step": 138
837
+ },
838
+ {
839
+ "epoch": 0.62,
840
+ "learning_rate": 0.00019863613034027224,
841
+ "loss": 0.51,
842
+ "step": 139
843
+ },
844
+ {
845
+ "epoch": 0.62,
846
+ "learning_rate": 0.00019863613034027224,
847
+ "loss": 0.1927,
848
+ "step": 140
849
+ },
850
+ {
851
+ "epoch": 0.63,
852
+ "learning_rate": 0.00019863613034027224,
853
+ "loss": 0.0,
854
+ "step": 141
855
+ },
856
+ {
857
+ "epoch": 0.63,
858
+ "learning_rate": 0.00019863613034027224,
859
+ "loss": 0.0,
860
+ "step": 142
861
+ },
862
+ {
863
+ "epoch": 0.64,
864
+ "learning_rate": 0.00019863613034027224,
865
+ "loss": 0.2303,
866
+ "step": 143
867
+ },
868
+ {
869
+ "epoch": 0.64,
870
+ "learning_rate": 0.00019863613034027224,
871
+ "loss": 0.0,
872
+ "step": 144
873
+ },
874
+ {
875
+ "epoch": 0.65,
876
+ "learning_rate": 0.00019863613034027224,
877
+ "loss": 0.1468,
878
+ "step": 145
879
+ },
880
+ {
881
+ "epoch": 0.65,
882
+ "learning_rate": 0.00019863613034027224,
883
+ "loss": 0.4611,
884
+ "step": 146
885
+ },
886
+ {
887
+ "epoch": 0.66,
888
+ "learning_rate": 0.00019863613034027224,
889
+ "loss": 0.126,
890
+ "step": 147
891
+ },
892
+ {
893
+ "epoch": 0.66,
894
+ "learning_rate": 0.00019863613034027224,
895
+ "loss": 1.1252,
896
+ "step": 148
897
+ },
898
+ {
899
+ "epoch": 0.66,
900
+ "learning_rate": 0.00019863613034027224,
901
+ "loss": 0.0,
902
+ "step": 149
903
+ },
904
+ {
905
+ "epoch": 0.67,
906
+ "learning_rate": 0.00019863613034027224,
907
+ "loss": 0.0,
908
+ "step": 150
909
+ },
910
+ {
911
+ "epoch": 0.67,
912
+ "learning_rate": 0.00019863613034027224,
913
+ "loss": 0.0,
914
+ "step": 151
915
+ },
916
+ {
917
+ "epoch": 0.68,
918
+ "learning_rate": 0.00019863613034027224,
919
+ "loss": 0.677,
920
+ "step": 152
921
+ },
922
+ {
923
+ "epoch": 0.68,
924
+ "learning_rate": 0.00019863613034027224,
925
+ "loss": 0.0,
926
+ "step": 153
927
+ },
928
+ {
929
+ "epoch": 0.69,
930
+ "learning_rate": 0.00019863613034027224,
931
+ "loss": 0.0,
932
+ "step": 154
933
+ },
934
+ {
935
+ "epoch": 0.69,
936
+ "learning_rate": 0.00019863613034027224,
937
+ "loss": 0.0,
938
+ "step": 155
939
+ },
940
+ {
941
+ "epoch": 0.7,
942
+ "learning_rate": 0.00019863613034027224,
943
+ "loss": 0.0,
944
+ "step": 156
945
+ },
946
+ {
947
+ "epoch": 0.7,
948
+ "learning_rate": 0.00019863613034027224,
949
+ "loss": 1.2064,
950
+ "step": 157
951
+ },
952
+ {
953
+ "epoch": 0.7,
954
+ "learning_rate": 0.00019863613034027224,
955
+ "loss": 0.4441,
956
+ "step": 158
957
+ },
958
+ {
959
+ "epoch": 0.71,
960
+ "learning_rate": 0.00019863613034027224,
961
+ "loss": 0.355,
962
+ "step": 159
963
+ },
964
+ {
965
+ "epoch": 0.71,
966
+ "learning_rate": 0.00019863613034027224,
967
+ "loss": 0.4723,
968
+ "step": 160
969
+ },
970
+ {
971
+ "epoch": 0.72,
972
+ "learning_rate": 0.00019863613034027224,
973
+ "loss": 0.4213,
974
+ "step": 161
975
+ },
976
+ {
977
+ "epoch": 0.72,
978
+ "learning_rate": 0.00019863613034027224,
979
+ "loss": 0.0,
980
+ "step": 162
981
+ },
982
+ {
983
+ "epoch": 0.73,
984
+ "learning_rate": 0.00019863613034027224,
985
+ "loss": 0.655,
986
+ "step": 163
987
+ },
988
+ {
989
+ "epoch": 0.73,
990
+ "learning_rate": 0.00019863613034027224,
991
+ "loss": 0.5043,
992
+ "step": 164
993
+ },
994
+ {
995
+ "epoch": 0.74,
996
+ "learning_rate": 0.00019863613034027224,
997
+ "loss": 3.0041,
998
+ "step": 165
999
+ },
1000
+ {
1001
+ "epoch": 0.74,
1002
+ "learning_rate": 0.00019863613034027224,
1003
+ "loss": 0.6247,
1004
+ "step": 166
1005
+ },
1006
+ {
1007
+ "epoch": 0.74,
1008
+ "learning_rate": 0.00019863613034027224,
1009
+ "loss": 0.4243,
1010
+ "step": 167
1011
+ },
1012
+ {
1013
+ "epoch": 0.75,
1014
+ "learning_rate": 0.00019863613034027224,
1015
+ "loss": 0.357,
1016
+ "step": 168
1017
+ },
1018
+ {
1019
+ "epoch": 0.75,
1020
+ "learning_rate": 0.00019863613034027224,
1021
+ "loss": 0.0,
1022
+ "step": 169
1023
+ },
1024
+ {
1025
+ "epoch": 0.76,
1026
+ "learning_rate": 0.00019863613034027224,
1027
+ "loss": 0.0,
1028
+ "step": 170
1029
+ },
1030
+ {
1031
+ "epoch": 0.76,
1032
+ "learning_rate": 0.00019863613034027224,
1033
+ "loss": 0.0,
1034
+ "step": 171
1035
+ },
1036
+ {
1037
+ "epoch": 0.77,
1038
+ "learning_rate": 0.00019863613034027224,
1039
+ "loss": 0.0,
1040
+ "step": 172
1041
+ },
1042
+ {
1043
+ "epoch": 0.77,
1044
+ "learning_rate": 0.00019863613034027224,
1045
+ "loss": 0.3537,
1046
+ "step": 173
1047
+ },
1048
+ {
1049
+ "epoch": 0.78,
1050
+ "learning_rate": 0.00019863613034027224,
1051
+ "loss": 0.5713,
1052
+ "step": 174
1053
+ },
1054
+ {
1055
+ "epoch": 0.78,
1056
+ "learning_rate": 0.00019863613034027224,
1057
+ "loss": 0.886,
1058
+ "step": 175
1059
+ },
1060
+ {
1061
+ "epoch": 0.78,
1062
+ "learning_rate": 0.00019863613034027224,
1063
+ "loss": 1.0761,
1064
+ "step": 176
1065
+ },
1066
+ {
1067
+ "epoch": 0.79,
1068
+ "learning_rate": 0.00019863613034027224,
1069
+ "loss": 1.2531,
1070
+ "step": 177
1071
+ },
1072
+ {
1073
+ "epoch": 0.79,
1074
+ "learning_rate": 0.00019863613034027224,
1075
+ "loss": 0.7733,
1076
+ "step": 178
1077
+ },
1078
+ {
1079
+ "epoch": 0.8,
1080
+ "learning_rate": 0.00019863613034027224,
1081
+ "loss": 0.0,
1082
+ "step": 179
1083
+ },
1084
+ {
1085
+ "epoch": 0.8,
1086
+ "learning_rate": 0.00019863613034027224,
1087
+ "loss": 0.3724,
1088
+ "step": 180
1089
+ },
1090
+ {
1091
+ "epoch": 0.81,
1092
+ "learning_rate": 0.00019863613034027224,
1093
+ "loss": 0.1415,
1094
+ "step": 181
1095
+ },
1096
+ {
1097
+ "epoch": 0.81,
1098
+ "learning_rate": 0.00019863613034027224,
1099
+ "loss": 1.0002,
1100
+ "step": 182
1101
+ },
1102
+ {
1103
+ "epoch": 0.82,
1104
+ "learning_rate": 0.00019863613034027224,
1105
+ "loss": 0.2229,
1106
+ "step": 183
1107
+ },
1108
+ {
1109
+ "epoch": 0.82,
1110
+ "learning_rate": 0.00019863613034027224,
1111
+ "loss": 0.7544,
1112
+ "step": 184
1113
+ },
1114
+ {
1115
+ "epoch": 0.82,
1116
+ "learning_rate": 0.00019863613034027224,
1117
+ "loss": 1.2167,
1118
+ "step": 185
1119
+ },
1120
+ {
1121
+ "epoch": 0.83,
1122
+ "learning_rate": 0.00019863613034027224,
1123
+ "loss": 0.4988,
1124
+ "step": 186
1125
+ },
1126
+ {
1127
+ "epoch": 0.83,
1128
+ "learning_rate": 0.00019863613034027224,
1129
+ "loss": 0.9973,
1130
+ "step": 187
1131
+ },
1132
+ {
1133
+ "epoch": 0.84,
1134
+ "learning_rate": 0.00019863613034027224,
1135
+ "loss": 0.7653,
1136
+ "step": 188
1137
+ },
1138
+ {
1139
+ "epoch": 0.84,
1140
+ "learning_rate": 0.00019863613034027224,
1141
+ "loss": 0.2718,
1142
+ "step": 189
1143
+ },
1144
+ {
1145
+ "epoch": 0.85,
1146
+ "learning_rate": 0.00019863613034027224,
1147
+ "loss": 1.0434,
1148
+ "step": 190
1149
+ },
1150
+ {
1151
+ "epoch": 0.85,
1152
+ "learning_rate": 0.00019863613034027224,
1153
+ "loss": 0.0,
1154
+ "step": 191
1155
+ },
1156
+ {
1157
+ "epoch": 0.86,
1158
+ "learning_rate": 0.00019863613034027224,
1159
+ "loss": 0.0,
1160
+ "step": 192
1161
+ },
1162
+ {
1163
+ "epoch": 0.86,
1164
+ "learning_rate": 0.00019863613034027224,
1165
+ "loss": 0.0,
1166
+ "step": 193
1167
+ },
1168
+ {
1169
+ "epoch": 0.87,
1170
+ "learning_rate": 0.00019863613034027224,
1171
+ "loss": 0.5341,
1172
+ "step": 194
1173
+ },
1174
+ {
1175
+ "epoch": 0.87,
1176
+ "learning_rate": 0.00019863613034027224,
1177
+ "loss": 0.7153,
1178
+ "step": 195
1179
+ },
1180
+ {
1181
+ "epoch": 0.87,
1182
+ "learning_rate": 0.00019863613034027224,
1183
+ "loss": 1.2904,
1184
+ "step": 196
1185
+ },
1186
+ {
1187
+ "epoch": 0.88,
1188
+ "learning_rate": 0.00019863613034027224,
1189
+ "loss": 0.2691,
1190
+ "step": 197
1191
+ },
1192
+ {
1193
+ "epoch": 0.88,
1194
+ "learning_rate": 0.00019863613034027224,
1195
+ "loss": 0.2125,
1196
+ "step": 198
1197
+ },
1198
+ {
1199
+ "epoch": 0.89,
1200
+ "learning_rate": 0.00019863613034027224,
1201
+ "loss": 0.0,
1202
+ "step": 199
1203
+ },
1204
+ {
1205
+ "epoch": 0.89,
1206
+ "learning_rate": 0.00019863613034027224,
1207
+ "loss": 0.536,
1208
+ "step": 200
1209
+ },
1210
+ {
1211
+ "epoch": 0.9,
1212
+ "learning_rate": 0.00019863613034027224,
1213
+ "loss": 0.6531,
1214
+ "step": 201
1215
+ },
1216
+ {
1217
+ "epoch": 0.9,
1218
+ "learning_rate": 0.00019863613034027224,
1219
+ "loss": 0.0,
1220
+ "step": 202
1221
+ },
1222
+ {
1223
+ "epoch": 0.91,
1224
+ "learning_rate": 0.00019863613034027224,
1225
+ "loss": 0.0,
1226
+ "step": 203
1227
+ },
1228
+ {
1229
+ "epoch": 0.91,
1230
+ "learning_rate": 0.00019863613034027224,
1231
+ "loss": 0.2077,
1232
+ "step": 204
1233
+ },
1234
+ {
1235
+ "epoch": 0.91,
1236
+ "learning_rate": 0.00019863613034027224,
1237
+ "loss": 0.1067,
1238
+ "step": 205
1239
+ },
1240
+ {
1241
+ "epoch": 0.92,
1242
+ "learning_rate": 0.00019863613034027224,
1243
+ "loss": 0.369,
1244
+ "step": 206
1245
+ },
1246
+ {
1247
+ "epoch": 0.92,
1248
+ "learning_rate": 0.00019863613034027224,
1249
+ "loss": 0.0,
1250
+ "step": 207
1251
+ },
1252
+ {
1253
+ "epoch": 0.93,
1254
+ "learning_rate": 0.00019863613034027224,
1255
+ "loss": 0.0,
1256
+ "step": 208
1257
+ },
1258
+ {
1259
+ "epoch": 0.93,
1260
+ "learning_rate": 0.00019863613034027224,
1261
+ "loss": 1.6003,
1262
+ "step": 209
1263
+ },
1264
+ {
1265
+ "epoch": 0.94,
1266
+ "learning_rate": 0.00019863613034027224,
1267
+ "loss": 0.0,
1268
+ "step": 210
1269
+ },
1270
+ {
1271
+ "epoch": 0.94,
1272
+ "learning_rate": 0.00019863613034027224,
1273
+ "loss": 0.3425,
1274
+ "step": 211
1275
+ },
1276
+ {
1277
+ "epoch": 0.95,
1278
+ "learning_rate": 0.00019863613034027224,
1279
+ "loss": 0.0,
1280
+ "step": 212
1281
+ },
1282
+ {
1283
+ "epoch": 0.95,
1284
+ "learning_rate": 0.00019863613034027224,
1285
+ "loss": 0.3939,
1286
+ "step": 213
1287
+ },
1288
+ {
1289
+ "epoch": 0.95,
1290
+ "learning_rate": 0.00019863613034027224,
1291
+ "loss": 0.1561,
1292
+ "step": 214
1293
+ },
1294
+ {
1295
+ "epoch": 0.96,
1296
+ "learning_rate": 0.00019863613034027224,
1297
+ "loss": 0.0,
1298
+ "step": 215
1299
+ },
1300
+ {
1301
+ "epoch": 0.96,
1302
+ "learning_rate": 0.00019863613034027224,
1303
+ "loss": 0.0,
1304
+ "step": 216
1305
+ },
1306
+ {
1307
+ "epoch": 0.97,
1308
+ "learning_rate": 0.00019863613034027224,
1309
+ "loss": 0.8341,
1310
+ "step": 217
1311
+ },
1312
+ {
1313
+ "epoch": 0.97,
1314
+ "learning_rate": 0.00019863613034027224,
1315
+ "loss": 0.0,
1316
+ "step": 218
1317
+ },
1318
+ {
1319
+ "epoch": 0.98,
1320
+ "learning_rate": 0.00019863613034027224,
1321
+ "loss": 0.7648,
1322
+ "step": 219
1323
+ },
1324
+ {
1325
+ "epoch": 0.98,
1326
+ "learning_rate": 0.00019863613034027224,
1327
+ "loss": 0.0884,
1328
+ "step": 220
1329
+ },
1330
+ {
1331
+ "epoch": 0.99,
1332
+ "learning_rate": 0.00019863613034027224,
1333
+ "loss": 0.1772,
1334
+ "step": 221
1335
+ },
1336
+ {
1337
+ "epoch": 0.99,
1338
+ "learning_rate": 0.00019863613034027224,
1339
+ "loss": 0.0,
1340
+ "step": 222
1341
+ },
1342
+ {
1343
+ "epoch": 0.99,
1344
+ "learning_rate": 0.00019863613034027224,
1345
+ "loss": 0.943,
1346
+ "step": 223
1347
+ },
1348
+ {
1349
+ "epoch": 1.0,
1350
+ "learning_rate": 0.00019863613034027224,
1351
+ "loss": 1.9696,
1352
+ "step": 224
1353
+ },
1354
+ {
1355
+ "epoch": 1.0,
1356
+ "learning_rate": 0.00019863613034027224,
1357
+ "loss": 0.1053,
1358
+ "step": 225
1359
+ },
1360
+ {
1361
+ "epoch": 1.01,
1362
+ "learning_rate": 0.00019863613034027224,
1363
+ "loss": 0.1738,
1364
+ "step": 226
1365
+ },
1366
+ {
1367
+ "epoch": 1.01,
1368
+ "learning_rate": 0.00019863613034027224,
1369
+ "loss": 0.0,
1370
+ "step": 227
1371
+ },
1372
+ {
1373
+ "epoch": 1.02,
1374
+ "learning_rate": 0.00019863613034027224,
1375
+ "loss": 0.0,
1376
+ "step": 228
1377
+ },
1378
+ {
1379
+ "epoch": 1.02,
1380
+ "learning_rate": 0.00019863613034027224,
1381
+ "loss": 0.9189,
1382
+ "step": 229
1383
+ },
1384
+ {
1385
+ "epoch": 1.03,
1386
+ "learning_rate": 0.00019863613034027224,
1387
+ "loss": 0.4056,
1388
+ "step": 230
1389
+ },
1390
+ {
1391
+ "epoch": 1.03,
1392
+ "learning_rate": 0.0001985497664274326,
1393
+ "loss": 0.1359,
1394
+ "step": 231
1395
+ },
1396
+ {
1397
+ "epoch": 1.03,
1398
+ "learning_rate": 0.0001984607713548251,
1399
+ "loss": 0.6309,
1400
+ "step": 232
1401
+ },
1402
+ {
1403
+ "epoch": 1.04,
1404
+ "learning_rate": 0.0001983691474985108,
1405
+ "loss": 0.0,
1406
+ "step": 233
1407
+ },
1408
+ {
1409
+ "epoch": 1.04,
1410
+ "learning_rate": 0.00019827489730473596,
1411
+ "loss": 0.3888,
1412
+ "step": 234
1413
+ },
1414
+ {
1415
+ "epoch": 1.05,
1416
+ "learning_rate": 0.00019817802328986697,
1417
+ "loss": 0.787,
1418
+ "step": 235
1419
+ },
1420
+ {
1421
+ "epoch": 1.05,
1422
+ "learning_rate": 0.00019807852804032305,
1423
+ "loss": 0.0,
1424
+ "step": 236
1425
+ },
1426
+ {
1427
+ "epoch": 1.06,
1428
+ "learning_rate": 0.00019797641421250725,
1429
+ "loss": 0.0,
1430
+ "step": 237
1431
+ },
1432
+ {
1433
+ "epoch": 1.06,
1434
+ "learning_rate": 0.00019787168453273544,
1435
+ "loss": 0.1077,
1436
+ "step": 238
1437
+ },
1438
+ {
1439
+ "epoch": 1.07,
1440
+ "learning_rate": 0.00019776434179716366,
1441
+ "loss": 0.0,
1442
+ "step": 239
1443
+ },
1444
+ {
1445
+ "epoch": 1.07,
1446
+ "learning_rate": 0.00019765438887171327,
1447
+ "loss": 0.1153,
1448
+ "step": 240
1449
+ },
1450
+ {
1451
+ "epoch": 1.07,
1452
+ "learning_rate": 0.0001975418286919947,
1453
+ "loss": 0.0,
1454
+ "step": 241
1455
+ },
1456
+ {
1457
+ "epoch": 1.08,
1458
+ "learning_rate": 0.00019742666426322876,
1459
+ "loss": 0.0,
1460
+ "step": 242
1461
+ },
1462
+ {
1463
+ "epoch": 1.08,
1464
+ "learning_rate": 0.0001973088986601667,
1465
+ "loss": 0.0,
1466
+ "step": 243
1467
+ },
1468
+ {
1469
+ "epoch": 1.09,
1470
+ "learning_rate": 0.00019718853502700783,
1471
+ "loss": 1.2459,
1472
+ "step": 244
1473
+ },
1474
+ {
1475
+ "epoch": 1.09,
1476
+ "learning_rate": 0.0001970655765773159,
1477
+ "loss": 0.1139,
1478
+ "step": 245
1479
+ },
1480
+ {
1481
+ "epoch": 1.1,
1482
+ "learning_rate": 0.00019694002659393305,
1483
+ "loss": 0.2691,
1484
+ "step": 246
1485
+ },
1486
+ {
1487
+ "epoch": 1.1,
1488
+ "learning_rate": 0.00019681188842889222,
1489
+ "loss": 0.0,
1490
+ "step": 247
1491
+ },
1492
+ {
1493
+ "epoch": 1.11,
1494
+ "learning_rate": 0.00019668116550332766,
1495
+ "loss": 0.4363,
1496
+ "step": 248
1497
+ },
1498
+ {
1499
+ "epoch": 1.11,
1500
+ "learning_rate": 0.0001965478613073837,
1501
+ "loss": 0.0,
1502
+ "step": 249
1503
+ },
1504
+ {
1505
+ "epoch": 1.11,
1506
+ "learning_rate": 0.00019641197940012137,
1507
+ "loss": 0.3793,
1508
+ "step": 250
1509
+ },
1510
+ {
1511
+ "epoch": 1.12,
1512
+ "learning_rate": 0.00019627352340942353,
1513
+ "loss": 0.0,
1514
+ "step": 251
1515
+ },
1516
+ {
1517
+ "epoch": 1.12,
1518
+ "learning_rate": 0.00019613249703189796,
1519
+ "loss": 0.1207,
1520
+ "step": 252
1521
+ },
1522
+ {
1523
+ "epoch": 1.13,
1524
+ "learning_rate": 0.00019598890403277864,
1525
+ "loss": 0.0872,
1526
+ "step": 253
1527
+ },
1528
+ {
1529
+ "epoch": 1.13,
1530
+ "learning_rate": 0.0001958427482458253,
1531
+ "loss": 0.0,
1532
+ "step": 254
1533
+ },
1534
+ {
1535
+ "epoch": 1.14,
1536
+ "learning_rate": 0.0001956940335732209,
1537
+ "loss": 0.0,
1538
+ "step": 255
1539
+ },
1540
+ {
1541
+ "epoch": 1.14,
1542
+ "learning_rate": 0.00019554276398546768,
1543
+ "loss": 0.9294,
1544
+ "step": 256
1545
+ },
1546
+ {
1547
+ "epoch": 1.15,
1548
+ "learning_rate": 0.000195388943521281,
1549
+ "loss": 0.6593,
1550
+ "step": 257
1551
+ },
1552
+ {
1553
+ "epoch": 1.15,
1554
+ "learning_rate": 0.00019523257628748146,
1555
+ "loss": 0.5178,
1556
+ "step": 258
1557
+ },
1558
+ {
1559
+ "epoch": 1.15,
1560
+ "learning_rate": 0.00019507366645888543,
1561
+ "loss": 0.0,
1562
+ "step": 259
1563
+ },
1564
+ {
1565
+ "epoch": 1.16,
1566
+ "learning_rate": 0.00019491221827819347,
1567
+ "loss": 0.6078,
1568
+ "step": 260
1569
+ },
1570
+ {
1571
+ "epoch": 1.16,
1572
+ "learning_rate": 0.00019474823605587703,
1573
+ "loss": 0.6807,
1574
+ "step": 261
1575
+ },
1576
+ {
1577
+ "epoch": 1.17,
1578
+ "learning_rate": 0.00019458172417006347,
1579
+ "loss": 0.2233,
1580
+ "step": 262
1581
+ },
1582
+ {
1583
+ "epoch": 1.17,
1584
+ "learning_rate": 0.00019441268706641907,
1585
+ "loss": 0.0,
1586
+ "step": 263
1587
+ },
1588
+ {
1589
+ "epoch": 1.18,
1590
+ "learning_rate": 0.00019424112925803039,
1591
+ "loss": 0.6479,
1592
+ "step": 264
1593
+ },
1594
+ {
1595
+ "epoch": 1.18,
1596
+ "learning_rate": 0.00019406705532528374,
1597
+ "loss": 0.0,
1598
+ "step": 265
1599
+ },
1600
+ {
1601
+ "epoch": 1.19,
1602
+ "learning_rate": 0.00019389046991574298,
1603
+ "loss": 0.0,
1604
+ "step": 266
1605
+ },
1606
+ {
1607
+ "epoch": 1.19,
1608
+ "learning_rate": 0.00019371137774402527,
1609
+ "loss": 0.2012,
1610
+ "step": 267
1611
+ },
1612
+ {
1613
+ "epoch": 1.2,
1614
+ "learning_rate": 0.0001935297835916754,
1615
+ "loss": 0.0,
1616
+ "step": 268
1617
+ },
1618
+ {
1619
+ "epoch": 1.2,
1620
+ "learning_rate": 0.00019334569230703794,
1621
+ "loss": 0.0,
1622
+ "step": 269
1623
+ },
1624
+ {
1625
+ "epoch": 1.2,
1626
+ "learning_rate": 0.0001931591088051279,
1627
+ "loss": 0.4641,
1628
+ "step": 270
1629
+ },
1630
+ {
1631
+ "epoch": 1.21,
1632
+ "learning_rate": 0.0001929700380674995,
1633
+ "loss": 0.0,
1634
+ "step": 271
1635
+ },
1636
+ {
1637
+ "epoch": 1.21,
1638
+ "learning_rate": 0.00019277848514211317,
1639
+ "loss": 0.2607,
1640
+ "step": 272
1641
+ },
1642
+ {
1643
+ "epoch": 1.22,
1644
+ "learning_rate": 0.00019258445514320065,
1645
+ "loss": 0.0,
1646
+ "step": 273
1647
+ },
1648
+ {
1649
+ "epoch": 1.22,
1650
+ "learning_rate": 0.0001923879532511287,
1651
+ "loss": 0.8596,
1652
+ "step": 274
1653
+ },
1654
+ {
1655
+ "epoch": 1.23,
1656
+ "learning_rate": 0.0001921889847122605,
1657
+ "loss": 0.315,
1658
+ "step": 275
1659
+ },
1660
+ {
1661
+ "epoch": 1.23,
1662
+ "learning_rate": 0.00019198755483881583,
1663
+ "loss": 0.0,
1664
+ "step": 276
1665
+ },
1666
+ {
1667
+ "epoch": 1.24,
1668
+ "learning_rate": 0.0001917836690087291,
1669
+ "loss": 0.0,
1670
+ "step": 277
1671
+ },
1672
+ {
1673
+ "epoch": 1.24,
1674
+ "learning_rate": 0.00019157733266550575,
1675
+ "loss": 1.0424,
1676
+ "step": 278
1677
+ },
1678
+ {
1679
+ "epoch": 1.24,
1680
+ "learning_rate": 0.00019136855131807705,
1681
+ "loss": 0.181,
1682
+ "step": 279
1683
+ },
1684
+ {
1685
+ "epoch": 1.25,
1686
+ "learning_rate": 0.0001911573305406528,
1687
+ "loss": 0.0,
1688
+ "step": 280
1689
+ },
1690
+ {
1691
+ "epoch": 1.25,
1692
+ "learning_rate": 0.00019094367597257282,
1693
+ "loss": 0.4583,
1694
+ "step": 281
1695
+ },
1696
+ {
1697
+ "epoch": 1.26,
1698
+ "learning_rate": 0.000190727593318156,
1699
+ "loss": 0.0,
1700
+ "step": 282
1701
+ },
1702
+ {
1703
+ "epoch": 1.26,
1704
+ "learning_rate": 0.00019050908834654834,
1705
+ "loss": 0.0898,
1706
+ "step": 283
1707
+ },
1708
+ {
1709
+ "epoch": 1.27,
1710
+ "learning_rate": 0.00019028816689156878,
1711
+ "loss": 2.3381,
1712
+ "step": 284
1713
+ },
1714
+ {
1715
+ "epoch": 1.27,
1716
+ "learning_rate": 0.00019006483485155338,
1717
+ "loss": 0.0,
1718
+ "step": 285
1719
+ },
1720
+ {
1721
+ "epoch": 1.28,
1722
+ "learning_rate": 0.0001898390981891979,
1723
+ "loss": 0.4441,
1724
+ "step": 286
1725
+ },
1726
+ {
1727
+ "epoch": 1.28,
1728
+ "learning_rate": 0.0001896109629313987,
1729
+ "loss": 0.0,
1730
+ "step": 287
1731
+ },
1732
+ {
1733
+ "epoch": 1.28,
1734
+ "learning_rate": 0.0001893804351690917,
1735
+ "loss": 0.3223,
1736
+ "step": 288
1737
+ },
1738
+ {
1739
+ "epoch": 1.29,
1740
+ "learning_rate": 0.0001891475210570898,
1741
+ "loss": 0.2783,
1742
+ "step": 289
1743
+ },
1744
+ {
1745
+ "epoch": 1.29,
1746
+ "learning_rate": 0.00018891222681391851,
1747
+ "loss": 0.8561,
1748
+ "step": 290
1749
+ },
1750
+ {
1751
+ "epoch": 1.3,
1752
+ "learning_rate": 0.00018867455872165008,
1753
+ "loss": 0.7544,
1754
+ "step": 291
1755
+ },
1756
+ {
1757
+ "epoch": 1.3,
1758
+ "learning_rate": 0.00018843452312573554,
1759
+ "loss": 1.5485,
1760
+ "step": 292
1761
+ },
1762
+ {
1763
+ "epoch": 1.31,
1764
+ "learning_rate": 0.0001881921264348355,
1765
+ "loss": 0.0,
1766
+ "step": 293
1767
+ },
1768
+ {
1769
+ "epoch": 1.31,
1770
+ "learning_rate": 0.0001879473751206489,
1771
+ "loss": 0.0,
1772
+ "step": 294
1773
+ },
1774
+ {
1775
+ "epoch": 1.32,
1776
+ "learning_rate": 0.00018770027571774031,
1777
+ "loss": 0.0,
1778
+ "step": 295
1779
+ },
1780
+ {
1781
+ "epoch": 1.32,
1782
+ "learning_rate": 0.00018745083482336544,
1783
+ "loss": 0.0,
1784
+ "step": 296
1785
+ },
1786
+ {
1787
+ "epoch": 1.32,
1788
+ "learning_rate": 0.00018719905909729494,
1789
+ "loss": 3.7086,
1790
+ "step": 297
1791
+ },
1792
+ {
1793
+ "epoch": 1.33,
1794
+ "learning_rate": 0.0001869449552616367,
1795
+ "loss": 1.2451,
1796
+ "step": 298
1797
+ },
1798
+ {
1799
+ "epoch": 1.33,
1800
+ "learning_rate": 0.00018668853010065634,
1801
+ "loss": 0.5947,
1802
+ "step": 299
1803
+ },
1804
+ {
1805
+ "epoch": 1.34,
1806
+ "learning_rate": 0.00018642979046059593,
1807
+ "loss": 1.9308,
1808
+ "step": 300
1809
+ },
1810
+ {
1811
+ "epoch": 1.34,
1812
+ "learning_rate": 0.00018616874324949159,
1813
+ "loss": 0.0,
1814
+ "step": 301
1815
+ },
1816
+ {
1817
+ "epoch": 1.35,
1818
+ "learning_rate": 0.00018590539543698854,
1819
+ "loss": 0.8493,
1820
+ "step": 302
1821
+ },
1822
+ {
1823
+ "epoch": 1.35,
1824
+ "learning_rate": 0.0001856397540541554,
1825
+ "loss": 0.7325,
1826
+ "step": 303
1827
+ },
1828
+ {
1829
+ "epoch": 1.36,
1830
+ "learning_rate": 0.0001853718261932964,
1831
+ "loss": 0.1978,
1832
+ "step": 304
1833
+ },
1834
+ {
1835
+ "epoch": 1.36,
1836
+ "learning_rate": 0.00018510161900776187,
1837
+ "loss": 0.0,
1838
+ "step": 305
1839
+ },
1840
+ {
1841
+ "epoch": 1.36,
1842
+ "learning_rate": 0.00018482913971175737,
1843
+ "loss": 0.2106,
1844
+ "step": 306
1845
+ },
1846
+ {
1847
+ "epoch": 1.37,
1848
+ "learning_rate": 0.00018455439558015115,
1849
+ "loss": 0.0,
1850
+ "step": 307
1851
+ },
1852
+ {
1853
+ "epoch": 1.37,
1854
+ "learning_rate": 0.00018427739394827973,
1855
+ "loss": 0.5134,
1856
+ "step": 308
1857
+ },
1858
+ {
1859
+ "epoch": 1.38,
1860
+ "learning_rate": 0.00018399814221175227,
1861
+ "loss": 0.1162,
1862
+ "step": 309
1863
+ },
1864
+ {
1865
+ "epoch": 1.38,
1866
+ "learning_rate": 0.00018371664782625287,
1867
+ "loss": 0.5827,
1868
+ "step": 310
1869
+ },
1870
+ {
1871
+ "epoch": 1.39,
1872
+ "learning_rate": 0.00018343291830734176,
1873
+ "loss": 0.0,
1874
+ "step": 311
1875
+ },
1876
+ {
1877
+ "epoch": 1.39,
1878
+ "learning_rate": 0.00018314696123025454,
1879
+ "loss": 1.7252,
1880
+ "step": 312
1881
+ },
1882
+ {
1883
+ "epoch": 1.4,
1884
+ "learning_rate": 0.00018285878422969983,
1885
+ "loss": 0.0,
1886
+ "step": 313
1887
+ },
1888
+ {
1889
+ "epoch": 1.4,
1890
+ "learning_rate": 0.0001825683949996556,
1891
+ "loss": 0.2794,
1892
+ "step": 314
1893
+ },
1894
+ {
1895
+ "epoch": 1.4,
1896
+ "learning_rate": 0.00018227580129316366,
1897
+ "loss": 0.0,
1898
+ "step": 315
1899
+ },
1900
+ {
1901
+ "epoch": 1.41,
1902
+ "learning_rate": 0.00018198101092212267,
1903
+ "loss": 0.6174,
1904
+ "step": 316
1905
+ },
1906
+ {
1907
+ "epoch": 1.41,
1908
+ "learning_rate": 0.00018168403175707954,
1909
+ "loss": 0.1033,
1910
+ "step": 317
1911
+ },
1912
+ {
1913
+ "epoch": 1.42,
1914
+ "learning_rate": 0.0001813848717270195,
1915
+ "loss": 0.0,
1916
+ "step": 318
1917
+ },
1918
+ {
1919
+ "epoch": 1.42,
1920
+ "learning_rate": 0.00018108353881915402,
1921
+ "loss": 0.4028,
1922
+ "step": 319
1923
+ },
1924
+ {
1925
+ "epoch": 1.43,
1926
+ "learning_rate": 0.00018078004107870797,
1927
+ "loss": 0.0,
1928
+ "step": 320
1929
+ },
1930
+ {
1931
+ "epoch": 1.43,
1932
+ "learning_rate": 0.00018047438660870446,
1933
+ "loss": 0.0,
1934
+ "step": 321
1935
+ },
1936
+ {
1937
+ "epoch": 1.44,
1938
+ "learning_rate": 0.00018016658356974884,
1939
+ "loss": 1.2,
1940
+ "step": 322
1941
+ },
1942
+ {
1943
+ "epoch": 1.44,
1944
+ "learning_rate": 0.0001798566401798106,
1945
+ "loss": 1.112,
1946
+ "step": 323
1947
+ },
1948
+ {
1949
+ "epoch": 1.44,
1950
+ "learning_rate": 0.00017954456471400393,
1951
+ "loss": 0.0,
1952
+ "step": 324
1953
+ },
1954
+ {
1955
+ "epoch": 1.45,
1956
+ "learning_rate": 0.00017923036550436704,
1957
+ "loss": 0.3934,
1958
+ "step": 325
1959
+ },
1960
+ {
1961
+ "epoch": 1.45,
1962
+ "learning_rate": 0.00017891405093963938,
1963
+ "loss": 0.3015,
1964
+ "step": 326
1965
+ },
1966
+ {
1967
+ "epoch": 1.46,
1968
+ "learning_rate": 0.00017859562946503788,
1969
+ "loss": 0.0726,
1970
+ "step": 327
1971
+ },
1972
+ {
1973
+ "epoch": 1.46,
1974
+ "learning_rate": 0.00017827510958203147,
1975
+ "loss": 0.9278,
1976
+ "step": 328
1977
+ },
1978
+ {
1979
+ "epoch": 1.47,
1980
+ "learning_rate": 0.00017795249984811396,
1981
+ "loss": 0.6865,
1982
+ "step": 329
1983
+ },
1984
+ {
1985
+ "epoch": 1.47,
1986
+ "learning_rate": 0.00017762780887657574,
1987
+ "loss": 0.1301,
1988
+ "step": 330
1989
+ },
1990
+ {
1991
+ "epoch": 1.48,
1992
+ "learning_rate": 0.0001773010453362737,
1993
+ "loss": 0.3788,
1994
+ "step": 331
1995
+ },
1996
+ {
1997
+ "epoch": 1.48,
1998
+ "learning_rate": 0.0001769722179513998,
1999
+ "loss": 0.1105,
2000
+ "step": 332
2001
+ },
2002
+ {
2003
+ "epoch": 1.48,
2004
+ "learning_rate": 0.00017664133550124815,
2005
+ "loss": 0.8738,
2006
+ "step": 333
2007
+ },
2008
+ {
2009
+ "epoch": 1.49,
2010
+ "learning_rate": 0.00017630840681998066,
2011
+ "loss": 0.4048,
2012
+ "step": 334
2013
+ },
2014
+ {
2015
+ "epoch": 1.49,
2016
+ "learning_rate": 0.00017597344079639112,
2017
+ "loss": 0.9362,
2018
+ "step": 335
2019
+ },
2020
+ {
2021
+ "epoch": 1.5,
2022
+ "learning_rate": 0.00017563644637366788,
2023
+ "loss": 0.0,
2024
+ "step": 336
2025
+ },
2026
+ {
2027
+ "epoch": 1.5,
2028
+ "learning_rate": 0.00017529743254915512,
2029
+ "loss": 0.0,
2030
+ "step": 337
2031
+ },
2032
+ {
2033
+ "epoch": 1.51,
2034
+ "learning_rate": 0.0001749564083741126,
2035
+ "loss": 0.24,
2036
+ "step": 338
2037
+ },
2038
+ {
2039
+ "epoch": 1.51,
2040
+ "learning_rate": 0.00017461338295347406,
2041
+ "loss": 0.0,
2042
+ "step": 339
2043
+ },
2044
+ {
2045
+ "epoch": 1.52,
2046
+ "learning_rate": 0.000174268365445604,
2047
+ "loss": 0.0,
2048
+ "step": 340
2049
+ },
2050
+ {
2051
+ "epoch": 1.52,
2052
+ "learning_rate": 0.0001739213650620533,
2053
+ "loss": 0.1102,
2054
+ "step": 341
2055
+ },
2056
+ {
2057
+ "epoch": 1.53,
2058
+ "learning_rate": 0.00017357239106731317,
2059
+ "loss": 0.8462,
2060
+ "step": 342
2061
+ },
2062
+ {
2063
+ "epoch": 1.53,
2064
+ "learning_rate": 0.00017322145277856794,
2065
+ "loss": 0.0,
2066
+ "step": 343
2067
+ },
2068
+ {
2069
+ "epoch": 1.53,
2070
+ "learning_rate": 0.00017286855956544613,
2071
+ "loss": 0.103,
2072
+ "step": 344
2073
+ },
2074
+ {
2075
+ "epoch": 1.54,
2076
+ "learning_rate": 0.0001725137208497705,
2077
+ "loss": 0.3364,
2078
+ "step": 345
2079
+ },
2080
+ {
2081
+ "epoch": 1.54,
2082
+ "learning_rate": 0.0001721569461053062,
2083
+ "loss": 0.1886,
2084
+ "step": 346
2085
+ },
2086
+ {
2087
+ "epoch": 1.55,
2088
+ "learning_rate": 0.0001717982448575082,
2089
+ "loss": 0.0,
2090
+ "step": 347
2091
+ },
2092
+ {
2093
+ "epoch": 1.55,
2094
+ "learning_rate": 0.00017143762668326667,
2095
+ "loss": 1.0661,
2096
+ "step": 348
2097
+ },
2098
+ {
2099
+ "epoch": 1.56,
2100
+ "learning_rate": 0.00017107510121065138,
2101
+ "loss": 0.9661,
2102
+ "step": 349
2103
+ },
2104
+ {
2105
+ "epoch": 1.56,
2106
+ "learning_rate": 0.00017071067811865476,
2107
+ "loss": 0.254,
2108
+ "step": 350
2109
+ },
2110
+ {
2111
+ "epoch": 1.57,
2112
+ "learning_rate": 0.0001703443671369333,
2113
+ "loss": 0.4995,
2114
+ "step": 351
2115
+ },
2116
+ {
2117
+ "epoch": 1.57,
2118
+ "learning_rate": 0.00016997617804554796,
2119
+ "loss": 0.0,
2120
+ "step": 352
2121
+ },
2122
+ {
2123
+ "epoch": 1.57,
2124
+ "learning_rate": 0.00016960612067470288,
2125
+ "loss": 0.2468,
2126
+ "step": 353
2127
+ },
2128
+ {
2129
+ "epoch": 1.58,
2130
+ "learning_rate": 0.00016923420490448296,
2131
+ "loss": 0.0,
2132
+ "step": 354
2133
+ },
2134
+ {
2135
+ "epoch": 1.58,
2136
+ "learning_rate": 0.0001688604406645903,
2137
+ "loss": 0.0,
2138
+ "step": 355
2139
+ },
2140
+ {
2141
+ "epoch": 1.59,
2142
+ "learning_rate": 0.00016848483793407873,
2143
+ "loss": 0.0,
2144
+ "step": 356
2145
+ },
2146
+ {
2147
+ "epoch": 1.59,
2148
+ "learning_rate": 0.00016810740674108764,
2149
+ "loss": 0.0,
2150
+ "step": 357
2151
+ },
2152
+ {
2153
+ "epoch": 1.6,
2154
+ "learning_rate": 0.00016772815716257412,
2155
+ "loss": 0.6732,
2156
+ "step": 358
2157
+ },
2158
+ {
2159
+ "epoch": 1.6,
2160
+ "learning_rate": 0.00016734709932404403,
2161
+ "loss": 0.0,
2162
+ "step": 359
2163
+ },
2164
+ {
2165
+ "epoch": 1.61,
2166
+ "learning_rate": 0.00016696424339928152,
2167
+ "loss": 0.0,
2168
+ "step": 360
2169
+ },
2170
+ {
2171
+ "epoch": 1.61,
2172
+ "learning_rate": 0.00016657959961007747,
2173
+ "loss": 0.1854,
2174
+ "step": 361
2175
+ },
2176
+ {
2177
+ "epoch": 1.61,
2178
+ "learning_rate": 0.00016619317822595667,
2179
+ "loss": 0.0,
2180
+ "step": 362
2181
+ },
2182
+ {
2183
+ "epoch": 1.62,
2184
+ "learning_rate": 0.00016580498956390342,
2185
+ "loss": 1.3683,
2186
+ "step": 363
2187
+ },
2188
+ {
2189
+ "epoch": 1.62,
2190
+ "learning_rate": 0.00016541504398808631,
2191
+ "loss": 0.6392,
2192
+ "step": 364
2193
+ },
2194
+ {
2195
+ "epoch": 1.63,
2196
+ "learning_rate": 0.00016502335190958135,
2197
+ "loss": 0.0,
2198
+ "step": 365
2199
+ },
2200
+ {
2201
+ "epoch": 1.63,
2202
+ "learning_rate": 0.00016462992378609407,
2203
+ "loss": 0.0,
2204
+ "step": 366
2205
+ },
2206
+ {
2207
+ "epoch": 1.64,
2208
+ "learning_rate": 0.00016423477012168038,
2209
+ "loss": 0.0,
2210
+ "step": 367
2211
+ },
2212
+ {
2213
+ "epoch": 1.64,
2214
+ "learning_rate": 0.00016383790146646588,
2215
+ "loss": 2.2779,
2216
+ "step": 368
2217
+ },
2218
+ {
2219
+ "epoch": 1.65,
2220
+ "learning_rate": 0.00016343932841636456,
2221
+ "loss": 0.0,
2222
+ "step": 369
2223
+ },
2224
+ {
2225
+ "epoch": 1.65,
2226
+ "learning_rate": 0.0001630390616127955,
2227
+ "loss": 1.1432,
2228
+ "step": 370
2229
+ },
2230
+ {
2231
+ "epoch": 1.65,
2232
+ "learning_rate": 0.00016263711174239914,
2233
+ "loss": 0.2808,
2234
+ "step": 371
2235
+ },
2236
+ {
2237
+ "epoch": 1.66,
2238
+ "learning_rate": 0.00016223348953675162,
2239
+ "loss": 0.2215,
2240
+ "step": 372
2241
+ },
2242
+ {
2243
+ "epoch": 1.66,
2244
+ "learning_rate": 0.00016182820577207842,
2245
+ "loss": 0.2374,
2246
+ "step": 373
2247
+ },
2248
+ {
2249
+ "epoch": 1.67,
2250
+ "learning_rate": 0.0001614212712689668,
2251
+ "loss": 0.1185,
2252
+ "step": 374
2253
+ },
2254
+ {
2255
+ "epoch": 1.67,
2256
+ "learning_rate": 0.00016101269689207655,
2257
+ "loss": 1.9245,
2258
+ "step": 375
2259
+ },
2260
+ {
2261
+ "epoch": 1.68,
2262
+ "learning_rate": 0.00016060249354985025,
2263
+ "loss": 0.0,
2264
+ "step": 376
2265
+ },
2266
+ {
2267
+ "epoch": 1.68,
2268
+ "learning_rate": 0.00016019067219422178,
2269
+ "loss": 0.0,
2270
+ "step": 377
2271
+ },
2272
+ {
2273
+ "epoch": 1.69,
2274
+ "learning_rate": 0.0001597772438203241,
2275
+ "loss": 0.1409,
2276
+ "step": 378
2277
+ },
2278
+ {
2279
+ "epoch": 1.69,
2280
+ "learning_rate": 0.0001593622194661956,
2281
+ "loss": 0.4018,
2282
+ "step": 379
2283
+ },
2284
+ {
2285
+ "epoch": 1.69,
2286
+ "learning_rate": 0.00015894561021248535,
2287
+ "loss": 0.0,
2288
+ "step": 380
2289
+ },
2290
+ {
2291
+ "epoch": 1.7,
2292
+ "learning_rate": 0.00015852742718215743,
2293
+ "loss": 0.8553,
2294
+ "step": 381
2295
+ },
2296
+ {
2297
+ "epoch": 1.7,
2298
+ "learning_rate": 0.00015810768154019385,
2299
+ "loss": 0.0,
2300
+ "step": 382
2301
+ },
2302
+ {
2303
+ "epoch": 1.71,
2304
+ "learning_rate": 0.0001576863844932963,
2305
+ "loss": 0.0,
2306
+ "step": 383
2307
+ },
2308
+ {
2309
+ "epoch": 1.71,
2310
+ "learning_rate": 0.00015726354728958736,
2311
+ "loss": 0.5503,
2312
+ "step": 384
2313
+ },
2314
+ {
2315
+ "epoch": 1.72,
2316
+ "learning_rate": 0.0001568391812183097,
2317
+ "loss": 0.0961,
2318
+ "step": 385
2319
+ },
2320
+ {
2321
+ "epoch": 1.72,
2322
+ "learning_rate": 0.00015641329760952513,
2323
+ "loss": 0.6591,
2324
+ "step": 386
2325
+ },
2326
+ {
2327
+ "epoch": 1.73,
2328
+ "learning_rate": 0.00015598590783381163,
2329
+ "loss": 0.0,
2330
+ "step": 387
2331
+ },
2332
+ {
2333
+ "epoch": 1.73,
2334
+ "learning_rate": 0.00015555702330196023,
2335
+ "loss": 0.122,
2336
+ "step": 388
2337
+ },
2338
+ {
2339
+ "epoch": 1.73,
2340
+ "learning_rate": 0.00015512665546467007,
2341
+ "loss": 0.0,
2342
+ "step": 389
2343
+ },
2344
+ {
2345
+ "epoch": 1.74,
2346
+ "learning_rate": 0.00015469481581224272,
2347
+ "loss": 0.0,
2348
+ "step": 390
2349
+ },
2350
+ {
2351
+ "epoch": 1.74,
2352
+ "learning_rate": 0.00015426151587427547,
2353
+ "loss": 0.0,
2354
+ "step": 391
2355
+ },
2356
+ {
2357
+ "epoch": 1.75,
2358
+ "learning_rate": 0.00015382676721935345,
2359
+ "loss": 0.3215,
2360
+ "step": 392
2361
+ },
2362
+ {
2363
+ "epoch": 1.75,
2364
+ "learning_rate": 0.00015339058145474085,
2365
+ "loss": 0.0,
2366
+ "step": 393
2367
+ },
2368
+ {
2369
+ "epoch": 1.76,
2370
+ "learning_rate": 0.00015295297022607088,
2371
+ "loss": 0.2013,
2372
+ "step": 394
2373
+ },
2374
+ {
2375
+ "epoch": 1.76,
2376
+ "learning_rate": 0.00015251394521703494,
2377
+ "loss": 0.0,
2378
+ "step": 395
2379
+ },
2380
+ {
2381
+ "epoch": 1.77,
2382
+ "learning_rate": 0.00015207351814907068,
2383
+ "loss": 0.3587,
2384
+ "step": 396
2385
+ },
2386
+ {
2387
+ "epoch": 1.77,
2388
+ "learning_rate": 0.000151631700781049,
2389
+ "loss": 0.0,
2390
+ "step": 397
2391
+ },
2392
+ {
2393
+ "epoch": 1.77,
2394
+ "learning_rate": 0.00015118850490896012,
2395
+ "loss": 0.7218,
2396
+ "step": 398
2397
+ },
2398
+ {
2399
+ "epoch": 1.78,
2400
+ "learning_rate": 0.0001507439423655987,
2401
+ "loss": 1.1481,
2402
+ "step": 399
2403
+ },
2404
+ {
2405
+ "epoch": 1.78,
2406
+ "learning_rate": 0.00015029802502024788,
2407
+ "loss": 0.4194,
2408
+ "step": 400
2409
+ },
2410
+ {
2411
+ "epoch": 1.79,
2412
+ "learning_rate": 0.0001498507647783623,
2413
+ "loss": 0.0,
2414
+ "step": 401
2415
+ },
2416
+ {
2417
+ "epoch": 1.79,
2418
+ "learning_rate": 0.00014940217358125042,
2419
+ "loss": 0.0,
2420
+ "step": 402
2421
+ },
2422
+ {
2423
+ "epoch": 1.8,
2424
+ "learning_rate": 0.0001489522634057555,
2425
+ "loss": 2.2446,
2426
+ "step": 403
2427
+ },
2428
+ {
2429
+ "epoch": 1.8,
2430
+ "learning_rate": 0.00014850104626393598,
2431
+ "loss": 0.3913,
2432
+ "step": 404
2433
+ },
2434
+ {
2435
+ "epoch": 1.81,
2436
+ "learning_rate": 0.00014804853420274472,
2437
+ "loss": 0.1975,
2438
+ "step": 405
2439
+ },
2440
+ {
2441
+ "epoch": 1.81,
2442
+ "learning_rate": 0.00014759473930370736,
2443
+ "loss": 0.1008,
2444
+ "step": 406
2445
+ },
2446
+ {
2447
+ "epoch": 1.81,
2448
+ "learning_rate": 0.0001471396736825998,
2449
+ "loss": 0.3,
2450
+ "step": 407
2451
+ },
2452
+ {
2453
+ "epoch": 1.82,
2454
+ "learning_rate": 0.00014668334948912453,
2455
+ "loss": 0.0,
2456
+ "step": 408
2457
+ },
2458
+ {
2459
+ "epoch": 1.82,
2460
+ "learning_rate": 0.00014622577890658665,
2461
+ "loss": 0.3837,
2462
+ "step": 409
2463
+ },
2464
+ {
2465
+ "epoch": 1.83,
2466
+ "learning_rate": 0.00014576697415156817,
2467
+ "loss": 0.0,
2468
+ "step": 410
2469
+ },
2470
+ {
2471
+ "epoch": 1.83,
2472
+ "learning_rate": 0.00014530694747360204,
2473
+ "loss": 0.0,
2474
+ "step": 411
2475
+ },
2476
+ {
2477
+ "epoch": 1.84,
2478
+ "learning_rate": 0.00014484571115484508,
2479
+ "loss": 0.0,
2480
+ "step": 412
2481
+ },
2482
+ {
2483
+ "epoch": 1.84,
2484
+ "learning_rate": 0.0001443832775097501,
2485
+ "loss": 0.7469,
2486
+ "step": 413
2487
+ },
2488
+ {
2489
+ "epoch": 1.85,
2490
+ "learning_rate": 0.00014391965888473703,
2491
+ "loss": 0.0,
2492
+ "step": 414
2493
+ },
2494
+ {
2495
+ "epoch": 1.85,
2496
+ "learning_rate": 0.0001434548676578634,
2497
+ "loss": 0.0,
2498
+ "step": 415
2499
+ },
2500
+ {
2501
+ "epoch": 1.86,
2502
+ "learning_rate": 0.0001429889162384937,
2503
+ "loss": 0.0,
2504
+ "step": 416
2505
+ },
2506
+ {
2507
+ "epoch": 1.86,
2508
+ "learning_rate": 0.00014252181706696817,
2509
+ "loss": 0.7078,
2510
+ "step": 417
2511
+ },
2512
+ {
2513
+ "epoch": 1.86,
2514
+ "learning_rate": 0.00014205358261427074,
2515
+ "loss": 0.308,
2516
+ "step": 418
2517
+ },
2518
+ {
2519
+ "epoch": 1.87,
2520
+ "learning_rate": 0.00014158422538169596,
2521
+ "loss": 0.6541,
2522
+ "step": 419
2523
+ },
2524
+ {
2525
+ "epoch": 1.87,
2526
+ "learning_rate": 0.0001411137579005151,
2527
+ "loss": 1.7852,
2528
+ "step": 420
2529
+ },
2530
+ {
2531
+ "epoch": 1.88,
2532
+ "learning_rate": 0.0001406421927316419,
2533
+ "loss": 0.4482,
2534
+ "step": 421
2535
+ },
2536
+ {
2537
+ "epoch": 1.88,
2538
+ "learning_rate": 0.00014016954246529696,
2539
+ "loss": 0.0,
2540
+ "step": 422
2541
+ },
2542
+ {
2543
+ "epoch": 1.89,
2544
+ "learning_rate": 0.00013969581972067164,
2545
+ "loss": 0.0,
2546
+ "step": 423
2547
+ },
2548
+ {
2549
+ "epoch": 1.89,
2550
+ "learning_rate": 0.0001392210371455913,
2551
+ "loss": 0.4676,
2552
+ "step": 424
2553
+ },
2554
+ {
2555
+ "epoch": 1.9,
2556
+ "learning_rate": 0.00013874520741617735,
2557
+ "loss": 0.0,
2558
+ "step": 425
2559
+ },
2560
+ {
2561
+ "epoch": 1.9,
2562
+ "learning_rate": 0.000138268343236509,
2563
+ "loss": 1.6066,
2564
+ "step": 426
2565
+ },
2566
+ {
2567
+ "epoch": 1.9,
2568
+ "learning_rate": 0.00013779045733828407,
2569
+ "loss": 0.617,
2570
+ "step": 427
2571
+ },
2572
+ {
2573
+ "epoch": 1.91,
2574
+ "learning_rate": 0.00013731156248047904,
2575
+ "loss": 0.2663,
2576
+ "step": 428
2577
+ },
2578
+ {
2579
+ "epoch": 1.91,
2580
+ "learning_rate": 0.00013683167144900834,
2581
+ "loss": 0.1382,
2582
+ "step": 429
2583
+ },
2584
+ {
2585
+ "epoch": 1.92,
2586
+ "learning_rate": 0.00013635079705638298,
2587
+ "loss": 0.2504,
2588
+ "step": 430
2589
+ },
2590
+ {
2591
+ "epoch": 1.92,
2592
+ "learning_rate": 0.00013586895214136874,
2593
+ "loss": 0.294,
2594
+ "step": 431
2595
+ },
2596
+ {
2597
+ "epoch": 1.93,
2598
+ "learning_rate": 0.00013538614956864296,
2599
+ "loss": 0.0,
2600
+ "step": 432
2601
+ },
2602
+ {
2603
+ "epoch": 1.93,
2604
+ "learning_rate": 0.0001349024022284514,
2605
+ "loss": 0.0,
2606
+ "step": 433
2607
+ },
2608
+ {
2609
+ "epoch": 1.94,
2610
+ "learning_rate": 0.00013441772303626387,
2611
+ "loss": 0.7406,
2612
+ "step": 434
2613
+ },
2614
+ {
2615
+ "epoch": 1.94,
2616
+ "learning_rate": 0.00013393212493242963,
2617
+ "loss": 0.5646,
2618
+ "step": 435
2619
+ },
2620
+ {
2621
+ "epoch": 1.94,
2622
+ "learning_rate": 0.00013344562088183165,
2623
+ "loss": 0.2678,
2624
+ "step": 436
2625
+ },
2626
+ {
2627
+ "epoch": 1.95,
2628
+ "learning_rate": 0.00013295822387354071,
2629
+ "loss": 0.9045,
2630
+ "step": 437
2631
+ },
2632
+ {
2633
+ "epoch": 1.95,
2634
+ "learning_rate": 0.00013246994692046836,
2635
+ "loss": 2.233,
2636
+ "step": 438
2637
+ },
2638
+ {
2639
+ "epoch": 1.96,
2640
+ "learning_rate": 0.0001319808030590197,
2641
+ "loss": 0.0,
2642
+ "step": 439
2643
+ },
2644
+ {
2645
+ "epoch": 1.96,
2646
+ "learning_rate": 0.0001314908053487452,
2647
+ "loss": 0.0,
2648
+ "step": 440
2649
+ },
2650
+ {
2651
+ "epoch": 1.97,
2652
+ "learning_rate": 0.00013099996687199203,
2653
+ "loss": 0.7132,
2654
+ "step": 441
2655
+ },
2656
+ {
2657
+ "epoch": 1.97,
2658
+ "learning_rate": 0.00013050830073355488,
2659
+ "loss": 0.0,
2660
+ "step": 442
2661
+ },
2662
+ {
2663
+ "epoch": 1.98,
2664
+ "learning_rate": 0.000130015820060326,
2665
+ "loss": 0.0,
2666
+ "step": 443
2667
+ },
2668
+ {
2669
+ "epoch": 1.98,
2670
+ "learning_rate": 0.00012952253800094468,
2671
+ "loss": 0.0,
2672
+ "step": 444
2673
+ },
2674
+ {
2675
+ "epoch": 1.98,
2676
+ "learning_rate": 0.00012902846772544624,
2677
+ "loss": 0.0,
2678
+ "step": 445
2679
+ },
2680
+ {
2681
+ "epoch": 1.99,
2682
+ "learning_rate": 0.00012853362242491053,
2683
+ "loss": 0.2216,
2684
+ "step": 446
2685
+ },
2686
+ {
2687
+ "epoch": 1.99,
2688
+ "learning_rate": 0.00012803801531110955,
2689
+ "loss": 1.9331,
2690
+ "step": 447
2691
+ },
2692
+ {
2693
+ "epoch": 2.0,
2694
+ "learning_rate": 0.0001275416596161548,
2695
+ "loss": 0.328,
2696
+ "step": 448
2697
+ },
2698
+ {
2699
+ "epoch": 2.0,
2700
+ "learning_rate": 0.00012704456859214397,
2701
+ "loss": 0.0,
2702
+ "step": 449
2703
+ },
2704
+ {
2705
+ "epoch": 2.01,
2706
+ "learning_rate": 0.00012654675551080724,
2707
+ "loss": 0.5864,
2708
+ "step": 450
2709
+ },
2710
+ {
2711
+ "epoch": 2.01,
2712
+ "learning_rate": 0.00012604823366315273,
2713
+ "loss": 0.0,
2714
+ "step": 451
2715
+ },
2716
+ {
2717
+ "epoch": 2.02,
2718
+ "learning_rate": 0.00012554901635911187,
2719
+ "loss": 0.0,
2720
+ "step": 452
2721
+ },
2722
+ {
2723
+ "epoch": 2.02,
2724
+ "learning_rate": 0.00012504911692718385,
2725
+ "loss": 0.1276,
2726
+ "step": 453
2727
+ },
2728
+ {
2729
+ "epoch": 2.02,
2730
+ "learning_rate": 0.00012454854871407994,
2731
+ "loss": 0.0,
2732
+ "step": 454
2733
+ },
2734
+ {
2735
+ "epoch": 2.03,
2736
+ "learning_rate": 0.0001240473250843669,
2737
+ "loss": 0.6042,
2738
+ "step": 455
2739
+ },
2740
+ {
2741
+ "epoch": 2.03,
2742
+ "learning_rate": 0.00012354545942011057,
2743
+ "loss": 0.2888,
2744
+ "step": 456
2745
+ },
2746
+ {
2747
+ "epoch": 2.04,
2748
+ "learning_rate": 0.00012304296512051814,
2749
+ "loss": 0.6748,
2750
+ "step": 457
2751
+ },
2752
+ {
2753
+ "epoch": 2.04,
2754
+ "learning_rate": 0.00012253985560158062,
2755
+ "loss": 0.1189,
2756
+ "step": 458
2757
+ },
2758
+ {
2759
+ "epoch": 2.05,
2760
+ "learning_rate": 0.00012203614429571475,
2761
+ "loss": 0.0,
2762
+ "step": 459
2763
+ },
2764
+ {
2765
+ "epoch": 2.05,
2766
+ "learning_rate": 0.00012153184465140413,
2767
+ "loss": 0.0,
2768
+ "step": 460
2769
+ },
2770
+ {
2771
+ "epoch": 2.06,
2772
+ "learning_rate": 0.00012102697013284034,
2773
+ "loss": 0.0,
2774
+ "step": 461
2775
+ },
2776
+ {
2777
+ "epoch": 2.06,
2778
+ "learning_rate": 0.00012052153421956342,
2779
+ "loss": 0.3809,
2780
+ "step": 462
2781
+ },
2782
+ {
2783
+ "epoch": 2.06,
2784
+ "learning_rate": 0.00012001555040610197,
2785
+ "loss": 0.0,
2786
+ "step": 463
2787
+ },
2788
+ {
2789
+ "epoch": 2.07,
2790
+ "learning_rate": 0.00011950903220161285,
2791
+ "loss": 1.0458,
2792
+ "step": 464
2793
+ },
2794
+ {
2795
+ "epoch": 2.07,
2796
+ "learning_rate": 0.00011900199312952047,
2797
+ "loss": 0.7787,
2798
+ "step": 465
2799
+ },
2800
+ {
2801
+ "epoch": 2.08,
2802
+ "learning_rate": 0.00011849444672715586,
2803
+ "loss": 1.0734,
2804
+ "step": 466
2805
+ },
2806
+ {
2807
+ "epoch": 2.08,
2808
+ "learning_rate": 0.0001179864065453951,
2809
+ "loss": 0.7641,
2810
+ "step": 467
2811
+ },
2812
+ {
2813
+ "epoch": 2.09,
2814
+ "learning_rate": 0.00011747788614829758,
2815
+ "loss": 0.1514,
2816
+ "step": 468
2817
+ },
2818
+ {
2819
+ "epoch": 2.09,
2820
+ "learning_rate": 0.00011696889911274393,
2821
+ "loss": 0.0,
2822
+ "step": 469
2823
+ },
2824
+ {
2825
+ "epoch": 2.1,
2826
+ "learning_rate": 0.00011645945902807341,
2827
+ "loss": 1.34,
2828
+ "step": 470
2829
+ },
2830
+ {
2831
+ "epoch": 2.1,
2832
+ "learning_rate": 0.0001159495794957211,
2833
+ "loss": 0.9174,
2834
+ "step": 471
2835
+ },
2836
+ {
2837
+ "epoch": 2.1,
2838
+ "learning_rate": 0.00011543927412885489,
2839
+ "loss": 1.1492,
2840
+ "step": 472
2841
+ },
2842
+ {
2843
+ "epoch": 2.11,
2844
+ "learning_rate": 0.0001149285565520119,
2845
+ "loss": 0.2276,
2846
+ "step": 473
2847
+ },
2848
+ {
2849
+ "epoch": 2.11,
2850
+ "learning_rate": 0.00011441744040073468,
2851
+ "loss": 0.0,
2852
+ "step": 474
2853
+ },
2854
+ {
2855
+ "epoch": 2.12,
2856
+ "learning_rate": 0.0001139059393212074,
2857
+ "loss": 0.0,
2858
+ "step": 475
2859
+ },
2860
+ {
2861
+ "epoch": 2.12,
2862
+ "learning_rate": 0.00011339406696989128,
2863
+ "loss": 0.0,
2864
+ "step": 476
2865
+ },
2866
+ {
2867
+ "epoch": 2.13,
2868
+ "learning_rate": 0.00011288183701315995,
2869
+ "loss": 0.13,
2870
+ "step": 477
2871
+ },
2872
+ {
2873
+ "epoch": 2.13,
2874
+ "learning_rate": 0.00011236926312693479,
2875
+ "loss": 0.7612,
2876
+ "step": 478
2877
+ },
2878
+ {
2879
+ "epoch": 2.14,
2880
+ "learning_rate": 0.00011185635899631963,
2881
+ "loss": 0.0,
2882
+ "step": 479
2883
+ },
2884
+ {
2885
+ "epoch": 2.14,
2886
+ "learning_rate": 0.00011134313831523547,
2887
+ "loss": 0.0,
2888
+ "step": 480
2889
+ },
2890
+ {
2891
+ "epoch": 2.14,
2892
+ "learning_rate": 0.00011082961478605475,
2893
+ "loss": 0.0,
2894
+ "step": 481
2895
+ },
2896
+ {
2897
+ "epoch": 2.15,
2898
+ "learning_rate": 0.00011031580211923571,
2899
+ "loss": 0.0,
2900
+ "step": 482
2901
+ },
2902
+ {
2903
+ "epoch": 2.15,
2904
+ "learning_rate": 0.0001098017140329561,
2905
+ "loss": 1.1886,
2906
+ "step": 483
2907
+ },
2908
+ {
2909
+ "epoch": 2.16,
2910
+ "learning_rate": 0.00010928736425274701,
2911
+ "loss": 1.2054,
2912
+ "step": 484
2913
+ },
2914
+ {
2915
+ "epoch": 2.16,
2916
+ "learning_rate": 0.00010877276651112662,
2917
+ "loss": 0.0,
2918
+ "step": 485
2919
+ },
2920
+ {
2921
+ "epoch": 2.17,
2922
+ "learning_rate": 0.00010825793454723325,
2923
+ "loss": 0.0,
2924
+ "step": 486
2925
+ },
2926
+ {
2927
+ "epoch": 2.17,
2928
+ "learning_rate": 0.00010774288210645862,
2929
+ "loss": 0.4863,
2930
+ "step": 487
2931
+ },
2932
+ {
2933
+ "epoch": 2.18,
2934
+ "learning_rate": 0.00010722762294008106,
2935
+ "loss": 0.0,
2936
+ "step": 488
2937
+ },
2938
+ {
2939
+ "epoch": 2.18,
2940
+ "learning_rate": 0.00010671217080489814,
2941
+ "loss": 0.0,
2942
+ "step": 489
2943
+ },
2944
+ {
2945
+ "epoch": 2.19,
2946
+ "learning_rate": 0.00010619653946285947,
2947
+ "loss": 0.929,
2948
+ "step": 490
2949
+ },
2950
+ {
2951
+ "epoch": 2.19,
2952
+ "learning_rate": 0.00010568074268069928,
2953
+ "loss": 0.262,
2954
+ "step": 491
2955
+ },
2956
+ {
2957
+ "epoch": 2.19,
2958
+ "learning_rate": 0.00010516479422956882,
2959
+ "loss": 0.7501,
2960
+ "step": 492
2961
+ },
2962
+ {
2963
+ "epoch": 2.2,
2964
+ "learning_rate": 0.00010464870788466873,
2965
+ "loss": 0.3049,
2966
+ "step": 493
2967
+ },
2968
+ {
2969
+ "epoch": 2.2,
2970
+ "learning_rate": 0.00010413249742488131,
2971
+ "loss": 0.8594,
2972
+ "step": 494
2973
+ },
2974
+ {
2975
+ "epoch": 2.21,
2976
+ "learning_rate": 0.00010361617663240253,
2977
+ "loss": 0.0,
2978
+ "step": 495
2979
+ },
2980
+ {
2981
+ "epoch": 2.21,
2982
+ "learning_rate": 0.00010309975929237408,
2983
+ "loss": 0.6285,
2984
+ "step": 496
2985
+ },
2986
+ {
2987
+ "epoch": 2.22,
2988
+ "learning_rate": 0.00010258325919251537,
2989
+ "loss": 1.0879,
2990
+ "step": 497
2991
+ },
2992
+ {
2993
+ "epoch": 2.22,
2994
+ "learning_rate": 0.00010206669012275545,
2995
+ "loss": 0.6124,
2996
+ "step": 498
2997
+ },
2998
+ {
2999
+ "epoch": 2.23,
3000
+ "learning_rate": 0.00010155006587486469,
3001
+ "loss": 0.0,
3002
+ "step": 499
3003
+ },
3004
+ {
3005
+ "epoch": 2.23,
3006
+ "learning_rate": 0.00010103340024208674,
3007
+ "loss": 0.4775,
3008
+ "step": 500
3009
+ }
3010
+ ],
3011
+ "max_steps": 640,
3012
+ "num_train_epochs": 3,
3013
+ "total_flos": 1.6543227452424192e+16,
3014
+ "trial_name": null,
3015
+ "trial_params": null
3016
+ }
train/checkpoint-500/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:e649dc040f39b15d9224b5ef2e3b9ef6607cf2b505e2004513b0df4866eba7cf
3
+ size 4344
train/runs/Dec09_21-43-47_1252a6b6660f/1702158227.8969784/events.out.tfevents.1702158227.1252a6b6660f ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:6814281b8a77cf3c666b474ec3b6e9b5ecfcb38cf2282da36440d449041754d7
3
+ size 5600
train/runs/Dec09_21-43-47_1252a6b6660f/events.out.tfevents.1702158227.1252a6b6660f ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:e106971f580923aa82c3bc147027dd825669edcf8e83e1bcb302dc67cd63c4cb
3
+ size 4868
train/runs/Dec09_21-44-36_1252a6b6660f/1702158276.6846008/events.out.tfevents.1702158276.1252a6b6660f ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ccfe8c37dadd9ba69db7e20909cffab9efa3a30473fc454d42df272676cc63d9
3
+ size 5600
train/runs/Dec09_21-44-36_1252a6b6660f/events.out.tfevents.1702158276.1252a6b6660f ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:2c6f0e69359afcd2efe5c06a24601f6a2060c588a0c08674860f5c44dda872d0
3
+ size 5330
train/runs/Dec09_21-45-42_1252a6b6660f/1702158343.1430557/events.out.tfevents.1702158343.1252a6b6660f.4035.1 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:6789ef4a019fe08c0c73ead1fa2dd1ed7b99cc3d015fccc75eddedfe4d8fa2fd
3
+ size 5900
train/runs/Dec09_21-45-42_1252a6b6660f/events.out.tfevents.1702158343.1252a6b6660f.4035.0 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d464dc3c7a9c8fcee3c68ac058e2cbd2eab66460876b4616cdaf161121a95431
3
+ size 4454
train/runs/Dec09_21-45-55_1252a6b6660f/1702158355.5142772/events.out.tfevents.1702158355.1252a6b6660f.4035.3 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:0b6f0765cfa11426f105ceb631aa63ffc5af0dbcbd873c4a169abd6ce6af20ad
3
+ size 5900
train/runs/Dec09_21-45-55_1252a6b6660f/events.out.tfevents.1702158355.1252a6b6660f.4035.2 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:147efbd6be226d9b04e553dc686c085fd4df65fde1243623288dbb55dcb09d5f
3
+ size 5687
train/runs/Dec09_21-46-30_1252a6b6660f/1702158390.361645/events.out.tfevents.1702158390.1252a6b6660f.4035.5 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:2c48338994ee1445547f7777bf6f91857ae22980e6134fac3c18a0b87470a1ab
3
+ size 5900
train/runs/Dec09_21-46-30_1252a6b6660f/events.out.tfevents.1702158390.1252a6b6660f.4035.4 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:05075707eb5625f8dfa46d8bf9246095a813d4e605d9b08f4fd2f4043066a15c
3
+ size 54668
train/runs/Dec09_22-20-54_1252a6b6660f/1702160454.3142843/events.out.tfevents.1702160454.1252a6b6660f.4035.7 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:72b1ddccd852ecfc86add1638d0c34b397d910232c4147342475712ec844493f
3
+ size 5900
train/runs/Dec09_22-20-54_1252a6b6660f/events.out.tfevents.1702160454.1252a6b6660f.4035.6 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f20ce573481bae89d0d286fc41cfc93070d80a1c848159c68fa715624debdc94
3
+ size 4455
train/runs/Dec09_22-21-11_1252a6b6660f/1702160471.8232346/events.out.tfevents.1702160471.1252a6b6660f.4035.9 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:4dc27fcc4bad2214f2ec814041bff4ad32f7fdd656b968c3f7006b0d6ac59dbe
3
+ size 5900
train/runs/Dec09_22-21-11_1252a6b6660f/events.out.tfevents.1702160471.1252a6b6660f.4035.8 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ccc856cc37f3681e4f96b263c9c4debae325eab72e7f2e6bc7d50a8673baa504
3
+ size 4455
train/runs/Dec09_22-22-57_1252a6b6660f/1702160577.833262/events.out.tfevents.1702160577.1252a6b6660f.16555.1 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:10c0215e3c2ee1850894eb9c4a18c8e7d8d2af8e6f90bebde87a7a770f8c29ed
3
+ size 5900
train/runs/Dec09_22-22-57_1252a6b6660f/events.out.tfevents.1702160577.1252a6b6660f.16555.0 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:10f6a5135f551e9a7a94e54f53890081cafbfc4757d6dca068ce8acdc08925eb
3
+ size 5995
train/runs/Dec09_22-23-50_1252a6b6660f/1702160631.116859/events.out.tfevents.1702160631.1252a6b6660f.16555.3 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:af6a6faaad363d4089105220b6402211b9c3e6e930e2253b80343306f721d09b
3
+ size 5900
train/runs/Dec09_22-23-50_1252a6b6660f/events.out.tfevents.1702160631.1252a6b6660f.16555.2 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9225846b7f7de7f3a338eb922b856e08e84e4ccc0fb31a843a0e503bcf125b2a
3
+ size 104908
train/runs/Dec09_23-07-43_1252a6b6660f/1702163264.08709/events.out.tfevents.1702163264.1252a6b6660f.29008.1 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9d16cdccb29f487e2b860d08c642444cb99ec0c9fcf26c4da1bdf6559c4c0e94
3
+ size 5900
train/runs/Dec09_23-07-43_1252a6b6660f/events.out.tfevents.1702163264.1252a6b6660f.29008.0 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:2186f0a27ec8c53576402055c122386351aa311e363b61e7c1dedce4c4badc4b
3
+ size 29548
train/runs/Dec10_10-09-29_2b7f3ae7a72d/1702202970.7887614/events.out.tfevents.1702202970.2b7f3ae7a72d.3821.1 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:06a7ffc7d2c133c12b04255de91d01fd70cedc8a24e55e5c540e83409aab2171
3
+ size 5900
train/runs/Dec10_10-09-29_2b7f3ae7a72d/events.out.tfevents.1702202970.2b7f3ae7a72d.3821.0 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:8a61b3bb732f83d45e4299ae5f9086aab96c8036da0e1e8fe2ed2fa9a374c3d9
3
+ size 29563