Training in progress, epoch 14
Browse files- last-checkpoint/optimizer.pt +1 -1
- last-checkpoint/pytorch_model.bin +1 -1
- last-checkpoint/rng_state.pth +1 -1
- last-checkpoint/scaler.pt +1 -1
- last-checkpoint/scheduler.pt +1 -1
- last-checkpoint/trainer_state.json +846 -3
- pytorch_model.bin +1 -1
- runs/Feb22_09-35-52_ubuntu-2004/events.out.tfevents.1677026198.ubuntu-2004.1870487.0 +2 -2
last-checkpoint/optimizer.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 236470789
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:1767ad24a3383984ce8730d69f87ce5710e95d8d3ba7a78f51e031b4074e752b
|
| 3 |
size 236470789
|
last-checkpoint/pytorch_model.bin
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 118243218
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:5f558a9b6175c80a1c75dcb143fa35aa44a75cbdb0c9e7f8f52c48de2931aed8
|
| 3 |
size 118243218
|
last-checkpoint/rng_state.pth
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 15597
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:3dd94082d220dbe1afff71cc1b8a20fec6bc1ac2bb822cbc323c364b293dc63a
|
| 3 |
size 15597
|
last-checkpoint/scaler.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 557
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:a01837036b51fcb751768f243bbc8ef9cd258f274881e125d9a37bb0fe81a367
|
| 3 |
size 557
|
last-checkpoint/scheduler.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 627
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:dbabbf8f48167d120f969a2bc0bc43b8158630bffc734ccac6b6b36381684d2b
|
| 3 |
size 627
|
last-checkpoint/trainer_state.json
CHANGED
|
@@ -1,8 +1,8 @@
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
-
"epoch":
|
| 5 |
-
"global_step":
|
| 6 |
"is_hyper_param_search": false,
|
| 7 |
"is_local_process_zero": true,
|
| 8 |
"is_world_process_zero": true,
|
|
@@ -10959,11 +10959,854 @@
|
|
| 10959 |
"eval_samples_per_second": 416.857,
|
| 10960 |
"eval_steps_per_second": 26.054,
|
| 10961 |
"step": 903149
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 10962 |
}
|
| 10963 |
],
|
| 10964 |
"max_steps": 972622,
|
| 10965 |
"num_train_epochs": 14,
|
| 10966 |
-
"total_flos":
|
| 10967 |
"trial_name": null,
|
| 10968 |
"trial_params": null
|
| 10969 |
}
|
|
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 14.0,
|
| 5 |
+
"global_step": 972622,
|
| 6 |
"is_hyper_param_search": false,
|
| 7 |
"is_local_process_zero": true,
|
| 8 |
"is_world_process_zero": true,
|
|
|
|
| 10959 |
"eval_samples_per_second": 416.857,
|
| 10960 |
"eval_steps_per_second": 26.054,
|
| 10961 |
"step": 903149
|
| 10962 |
+
},
|
| 10963 |
+
{
|
| 10964 |
+
"epoch": 13.01,
|
| 10965 |
+
"learning_rate": 7.221244268585879e-06,
|
| 10966 |
+
"loss": 1.7015,
|
| 10967 |
+
"step": 903500
|
| 10968 |
+
},
|
| 10969 |
+
{
|
| 10970 |
+
"epoch": 13.01,
|
| 10971 |
+
"learning_rate": 7.1693175268331445e-06,
|
| 10972 |
+
"loss": 1.7008,
|
| 10973 |
+
"step": 904000
|
| 10974 |
+
},
|
| 10975 |
+
{
|
| 10976 |
+
"epoch": 13.02,
|
| 10977 |
+
"learning_rate": 7.1173907850804094e-06,
|
| 10978 |
+
"loss": 1.7008,
|
| 10979 |
+
"step": 904500
|
| 10980 |
+
},
|
| 10981 |
+
{
|
| 10982 |
+
"epoch": 13.03,
|
| 10983 |
+
"learning_rate": 7.065567896811179e-06,
|
| 10984 |
+
"loss": 1.7008,
|
| 10985 |
+
"step": 905000
|
| 10986 |
+
},
|
| 10987 |
+
{
|
| 10988 |
+
"epoch": 13.03,
|
| 10989 |
+
"learning_rate": 7.013641155058444e-06,
|
| 10990 |
+
"loss": 1.7018,
|
| 10991 |
+
"step": 905500
|
| 10992 |
+
},
|
| 10993 |
+
{
|
| 10994 |
+
"epoch": 13.04,
|
| 10995 |
+
"learning_rate": 6.961714413305709e-06,
|
| 10996 |
+
"loss": 1.7048,
|
| 10997 |
+
"step": 906000
|
| 10998 |
+
},
|
| 10999 |
+
{
|
| 11000 |
+
"epoch": 13.05,
|
| 11001 |
+
"learning_rate": 6.909787671552974e-06,
|
| 11002 |
+
"loss": 1.7036,
|
| 11003 |
+
"step": 906500
|
| 11004 |
+
},
|
| 11005 |
+
{
|
| 11006 |
+
"epoch": 13.06,
|
| 11007 |
+
"learning_rate": 6.857860929800238e-06,
|
| 11008 |
+
"loss": 1.7012,
|
| 11009 |
+
"step": 907000
|
| 11010 |
+
},
|
| 11011 |
+
{
|
| 11012 |
+
"epoch": 13.06,
|
| 11013 |
+
"learning_rate": 6.805934188047503e-06,
|
| 11014 |
+
"loss": 1.6996,
|
| 11015 |
+
"step": 907500
|
| 11016 |
+
},
|
| 11017 |
+
{
|
| 11018 |
+
"epoch": 13.07,
|
| 11019 |
+
"learning_rate": 6.754111299778273e-06,
|
| 11020 |
+
"loss": 1.701,
|
| 11021 |
+
"step": 908000
|
| 11022 |
+
},
|
| 11023 |
+
{
|
| 11024 |
+
"epoch": 13.08,
|
| 11025 |
+
"learning_rate": 6.702184558025538e-06,
|
| 11026 |
+
"loss": 1.7037,
|
| 11027 |
+
"step": 908500
|
| 11028 |
+
},
|
| 11029 |
+
{
|
| 11030 |
+
"epoch": 13.08,
|
| 11031 |
+
"learning_rate": 6.650257816272802e-06,
|
| 11032 |
+
"loss": 1.7042,
|
| 11033 |
+
"step": 909000
|
| 11034 |
+
},
|
| 11035 |
+
{
|
| 11036 |
+
"epoch": 13.09,
|
| 11037 |
+
"learning_rate": 6.598331074520067e-06,
|
| 11038 |
+
"loss": 1.704,
|
| 11039 |
+
"step": 909500
|
| 11040 |
+
},
|
| 11041 |
+
{
|
| 11042 |
+
"epoch": 13.1,
|
| 11043 |
+
"learning_rate": 6.546508186250838e-06,
|
| 11044 |
+
"loss": 1.7035,
|
| 11045 |
+
"step": 910000
|
| 11046 |
+
},
|
| 11047 |
+
{
|
| 11048 |
+
"epoch": 13.11,
|
| 11049 |
+
"learning_rate": 6.494581444498102e-06,
|
| 11050 |
+
"loss": 1.7007,
|
| 11051 |
+
"step": 910500
|
| 11052 |
+
},
|
| 11053 |
+
{
|
| 11054 |
+
"epoch": 13.11,
|
| 11055 |
+
"learning_rate": 6.4426547027453665e-06,
|
| 11056 |
+
"loss": 1.703,
|
| 11057 |
+
"step": 911000
|
| 11058 |
+
},
|
| 11059 |
+
{
|
| 11060 |
+
"epoch": 13.12,
|
| 11061 |
+
"learning_rate": 6.390727960992631e-06,
|
| 11062 |
+
"loss": 1.7008,
|
| 11063 |
+
"step": 911500
|
| 11064 |
+
},
|
| 11065 |
+
{
|
| 11066 |
+
"epoch": 13.13,
|
| 11067 |
+
"learning_rate": 6.338905072723401e-06,
|
| 11068 |
+
"loss": 1.7057,
|
| 11069 |
+
"step": 912000
|
| 11070 |
+
},
|
| 11071 |
+
{
|
| 11072 |
+
"epoch": 13.13,
|
| 11073 |
+
"learning_rate": 6.286978330970667e-06,
|
| 11074 |
+
"loss": 1.7022,
|
| 11075 |
+
"step": 912500
|
| 11076 |
+
},
|
| 11077 |
+
{
|
| 11078 |
+
"epoch": 13.14,
|
| 11079 |
+
"learning_rate": 6.235051589217932e-06,
|
| 11080 |
+
"loss": 1.7011,
|
| 11081 |
+
"step": 913000
|
| 11082 |
+
},
|
| 11083 |
+
{
|
| 11084 |
+
"epoch": 13.15,
|
| 11085 |
+
"learning_rate": 6.183124847465197e-06,
|
| 11086 |
+
"loss": 1.7006,
|
| 11087 |
+
"step": 913500
|
| 11088 |
+
},
|
| 11089 |
+
{
|
| 11090 |
+
"epoch": 13.16,
|
| 11091 |
+
"learning_rate": 6.131198105712461e-06,
|
| 11092 |
+
"loss": 1.6992,
|
| 11093 |
+
"step": 914000
|
| 11094 |
+
},
|
| 11095 |
+
{
|
| 11096 |
+
"epoch": 13.16,
|
| 11097 |
+
"learning_rate": 6.079375217443231e-06,
|
| 11098 |
+
"loss": 1.702,
|
| 11099 |
+
"step": 914500
|
| 11100 |
+
},
|
| 11101 |
+
{
|
| 11102 |
+
"epoch": 13.17,
|
| 11103 |
+
"learning_rate": 6.027448475690496e-06,
|
| 11104 |
+
"loss": 1.6989,
|
| 11105 |
+
"step": 915000
|
| 11106 |
+
},
|
| 11107 |
+
{
|
| 11108 |
+
"epoch": 13.18,
|
| 11109 |
+
"learning_rate": 5.975521733937761e-06,
|
| 11110 |
+
"loss": 1.7008,
|
| 11111 |
+
"step": 915500
|
| 11112 |
+
},
|
| 11113 |
+
{
|
| 11114 |
+
"epoch": 13.18,
|
| 11115 |
+
"learning_rate": 5.923594992185026e-06,
|
| 11116 |
+
"loss": 1.7023,
|
| 11117 |
+
"step": 916000
|
| 11118 |
+
},
|
| 11119 |
+
{
|
| 11120 |
+
"epoch": 13.19,
|
| 11121 |
+
"learning_rate": 5.871668250432291e-06,
|
| 11122 |
+
"loss": 1.7007,
|
| 11123 |
+
"step": 916500
|
| 11124 |
+
},
|
| 11125 |
+
{
|
| 11126 |
+
"epoch": 13.2,
|
| 11127 |
+
"learning_rate": 5.819845362163061e-06,
|
| 11128 |
+
"loss": 1.6995,
|
| 11129 |
+
"step": 917000
|
| 11130 |
+
},
|
| 11131 |
+
{
|
| 11132 |
+
"epoch": 13.21,
|
| 11133 |
+
"learning_rate": 5.767918620410325e-06,
|
| 11134 |
+
"loss": 1.7002,
|
| 11135 |
+
"step": 917500
|
| 11136 |
+
},
|
| 11137 |
+
{
|
| 11138 |
+
"epoch": 13.21,
|
| 11139 |
+
"learning_rate": 5.71599187865759e-06,
|
| 11140 |
+
"loss": 1.7031,
|
| 11141 |
+
"step": 918000
|
| 11142 |
+
},
|
| 11143 |
+
{
|
| 11144 |
+
"epoch": 13.22,
|
| 11145 |
+
"learning_rate": 5.664065136904855e-06,
|
| 11146 |
+
"loss": 1.6987,
|
| 11147 |
+
"step": 918500
|
| 11148 |
+
},
|
| 11149 |
+
{
|
| 11150 |
+
"epoch": 13.23,
|
| 11151 |
+
"learning_rate": 5.612138395152119e-06,
|
| 11152 |
+
"loss": 1.7016,
|
| 11153 |
+
"step": 919000
|
| 11154 |
+
},
|
| 11155 |
+
{
|
| 11156 |
+
"epoch": 13.24,
|
| 11157 |
+
"learning_rate": 5.56031550688289e-06,
|
| 11158 |
+
"loss": 1.6996,
|
| 11159 |
+
"step": 919500
|
| 11160 |
+
},
|
| 11161 |
+
{
|
| 11162 |
+
"epoch": 13.24,
|
| 11163 |
+
"learning_rate": 5.5083887651301545e-06,
|
| 11164 |
+
"loss": 1.7036,
|
| 11165 |
+
"step": 920000
|
| 11166 |
+
},
|
| 11167 |
+
{
|
| 11168 |
+
"epoch": 13.25,
|
| 11169 |
+
"learning_rate": 5.456462023377419e-06,
|
| 11170 |
+
"loss": 1.7015,
|
| 11171 |
+
"step": 920500
|
| 11172 |
+
},
|
| 11173 |
+
{
|
| 11174 |
+
"epoch": 13.26,
|
| 11175 |
+
"learning_rate": 5.404535281624684e-06,
|
| 11176 |
+
"loss": 1.7019,
|
| 11177 |
+
"step": 921000
|
| 11178 |
+
},
|
| 11179 |
+
{
|
| 11180 |
+
"epoch": 13.26,
|
| 11181 |
+
"learning_rate": 5.352712393355455e-06,
|
| 11182 |
+
"loss": 1.7019,
|
| 11183 |
+
"step": 921500
|
| 11184 |
+
},
|
| 11185 |
+
{
|
| 11186 |
+
"epoch": 13.27,
|
| 11187 |
+
"learning_rate": 5.30078565160272e-06,
|
| 11188 |
+
"loss": 1.7024,
|
| 11189 |
+
"step": 922000
|
| 11190 |
+
},
|
| 11191 |
+
{
|
| 11192 |
+
"epoch": 13.28,
|
| 11193 |
+
"learning_rate": 5.248858909849984e-06,
|
| 11194 |
+
"loss": 1.7003,
|
| 11195 |
+
"step": 922500
|
| 11196 |
+
},
|
| 11197 |
+
{
|
| 11198 |
+
"epoch": 13.29,
|
| 11199 |
+
"learning_rate": 5.196932168097249e-06,
|
| 11200 |
+
"loss": 1.6995,
|
| 11201 |
+
"step": 923000
|
| 11202 |
+
},
|
| 11203 |
+
{
|
| 11204 |
+
"epoch": 13.29,
|
| 11205 |
+
"learning_rate": 5.145109279828019e-06,
|
| 11206 |
+
"loss": 1.6998,
|
| 11207 |
+
"step": 923500
|
| 11208 |
+
},
|
| 11209 |
+
{
|
| 11210 |
+
"epoch": 13.3,
|
| 11211 |
+
"learning_rate": 5.093182538075283e-06,
|
| 11212 |
+
"loss": 1.6999,
|
| 11213 |
+
"step": 924000
|
| 11214 |
+
},
|
| 11215 |
+
{
|
| 11216 |
+
"epoch": 13.31,
|
| 11217 |
+
"learning_rate": 5.041255796322548e-06,
|
| 11218 |
+
"loss": 1.6995,
|
| 11219 |
+
"step": 924500
|
| 11220 |
+
},
|
| 11221 |
+
{
|
| 11222 |
+
"epoch": 13.31,
|
| 11223 |
+
"learning_rate": 4.989329054569814e-06,
|
| 11224 |
+
"loss": 1.6965,
|
| 11225 |
+
"step": 925000
|
| 11226 |
+
},
|
| 11227 |
+
{
|
| 11228 |
+
"epoch": 13.32,
|
| 11229 |
+
"learning_rate": 4.9375061663005835e-06,
|
| 11230 |
+
"loss": 1.7033,
|
| 11231 |
+
"step": 925500
|
| 11232 |
+
},
|
| 11233 |
+
{
|
| 11234 |
+
"epoch": 13.33,
|
| 11235 |
+
"learning_rate": 4.885579424547848e-06,
|
| 11236 |
+
"loss": 1.6986,
|
| 11237 |
+
"step": 926000
|
| 11238 |
+
},
|
| 11239 |
+
{
|
| 11240 |
+
"epoch": 13.34,
|
| 11241 |
+
"learning_rate": 4.833652682795113e-06,
|
| 11242 |
+
"loss": 1.6984,
|
| 11243 |
+
"step": 926500
|
| 11244 |
+
},
|
| 11245 |
+
{
|
| 11246 |
+
"epoch": 13.34,
|
| 11247 |
+
"learning_rate": 4.781725941042377e-06,
|
| 11248 |
+
"loss": 1.7015,
|
| 11249 |
+
"step": 927000
|
| 11250 |
+
},
|
| 11251 |
+
{
|
| 11252 |
+
"epoch": 13.35,
|
| 11253 |
+
"learning_rate": 4.729799199289642e-06,
|
| 11254 |
+
"loss": 1.6982,
|
| 11255 |
+
"step": 927500
|
| 11256 |
+
},
|
| 11257 |
+
{
|
| 11258 |
+
"epoch": 13.36,
|
| 11259 |
+
"learning_rate": 4.677976311020413e-06,
|
| 11260 |
+
"loss": 1.7004,
|
| 11261 |
+
"step": 928000
|
| 11262 |
+
},
|
| 11263 |
+
{
|
| 11264 |
+
"epoch": 13.36,
|
| 11265 |
+
"learning_rate": 4.626049569267677e-06,
|
| 11266 |
+
"loss": 1.702,
|
| 11267 |
+
"step": 928500
|
| 11268 |
+
},
|
| 11269 |
+
{
|
| 11270 |
+
"epoch": 13.37,
|
| 11271 |
+
"learning_rate": 4.5741228275149425e-06,
|
| 11272 |
+
"loss": 1.7002,
|
| 11273 |
+
"step": 929000
|
| 11274 |
+
},
|
| 11275 |
+
{
|
| 11276 |
+
"epoch": 13.38,
|
| 11277 |
+
"learning_rate": 4.5221960857622074e-06,
|
| 11278 |
+
"loss": 1.7009,
|
| 11279 |
+
"step": 929500
|
| 11280 |
+
},
|
| 11281 |
+
{
|
| 11282 |
+
"epoch": 13.39,
|
| 11283 |
+
"learning_rate": 4.4702693440094715e-06,
|
| 11284 |
+
"loss": 1.6987,
|
| 11285 |
+
"step": 930000
|
| 11286 |
+
},
|
| 11287 |
+
{
|
| 11288 |
+
"epoch": 13.39,
|
| 11289 |
+
"learning_rate": 4.418446455740242e-06,
|
| 11290 |
+
"loss": 1.7007,
|
| 11291 |
+
"step": 930500
|
| 11292 |
+
},
|
| 11293 |
+
{
|
| 11294 |
+
"epoch": 13.4,
|
| 11295 |
+
"learning_rate": 4.366519713987507e-06,
|
| 11296 |
+
"loss": 1.6964,
|
| 11297 |
+
"step": 931000
|
| 11298 |
+
},
|
| 11299 |
+
{
|
| 11300 |
+
"epoch": 13.41,
|
| 11301 |
+
"learning_rate": 4.314592972234771e-06,
|
| 11302 |
+
"loss": 1.6999,
|
| 11303 |
+
"step": 931500
|
| 11304 |
+
},
|
| 11305 |
+
{
|
| 11306 |
+
"epoch": 13.42,
|
| 11307 |
+
"learning_rate": 4.262666230482036e-06,
|
| 11308 |
+
"loss": 1.7002,
|
| 11309 |
+
"step": 932000
|
| 11310 |
+
},
|
| 11311 |
+
{
|
| 11312 |
+
"epoch": 13.42,
|
| 11313 |
+
"learning_rate": 4.210843342212806e-06,
|
| 11314 |
+
"loss": 1.6982,
|
| 11315 |
+
"step": 932500
|
| 11316 |
+
},
|
| 11317 |
+
{
|
| 11318 |
+
"epoch": 13.43,
|
| 11319 |
+
"learning_rate": 4.158916600460071e-06,
|
| 11320 |
+
"loss": 1.697,
|
| 11321 |
+
"step": 933000
|
| 11322 |
+
},
|
| 11323 |
+
{
|
| 11324 |
+
"epoch": 13.44,
|
| 11325 |
+
"learning_rate": 4.106989858707336e-06,
|
| 11326 |
+
"loss": 1.7032,
|
| 11327 |
+
"step": 933500
|
| 11328 |
+
},
|
| 11329 |
+
{
|
| 11330 |
+
"epoch": 13.44,
|
| 11331 |
+
"learning_rate": 4.055063116954601e-06,
|
| 11332 |
+
"loss": 1.7021,
|
| 11333 |
+
"step": 934000
|
| 11334 |
+
},
|
| 11335 |
+
{
|
| 11336 |
+
"epoch": 13.45,
|
| 11337 |
+
"learning_rate": 4.003136375201865e-06,
|
| 11338 |
+
"loss": 1.7011,
|
| 11339 |
+
"step": 934500
|
| 11340 |
+
},
|
| 11341 |
+
{
|
| 11342 |
+
"epoch": 13.46,
|
| 11343 |
+
"learning_rate": 3.9513134869326356e-06,
|
| 11344 |
+
"loss": 1.7007,
|
| 11345 |
+
"step": 935000
|
| 11346 |
+
},
|
| 11347 |
+
{
|
| 11348 |
+
"epoch": 13.47,
|
| 11349 |
+
"learning_rate": 3.8993867451799005e-06,
|
| 11350 |
+
"loss": 1.6997,
|
| 11351 |
+
"step": 935500
|
| 11352 |
+
},
|
| 11353 |
+
{
|
| 11354 |
+
"epoch": 13.47,
|
| 11355 |
+
"learning_rate": 3.847460003427165e-06,
|
| 11356 |
+
"loss": 1.7,
|
| 11357 |
+
"step": 936000
|
| 11358 |
+
},
|
| 11359 |
+
{
|
| 11360 |
+
"epoch": 13.48,
|
| 11361 |
+
"learning_rate": 3.79553326167443e-06,
|
| 11362 |
+
"loss": 1.7009,
|
| 11363 |
+
"step": 936500
|
| 11364 |
+
},
|
| 11365 |
+
{
|
| 11366 |
+
"epoch": 13.49,
|
| 11367 |
+
"learning_rate": 3.7436065199216947e-06,
|
| 11368 |
+
"loss": 1.698,
|
| 11369 |
+
"step": 937000
|
| 11370 |
+
},
|
| 11371 |
+
{
|
| 11372 |
+
"epoch": 13.49,
|
| 11373 |
+
"learning_rate": 3.6917836316524652e-06,
|
| 11374 |
+
"loss": 1.703,
|
| 11375 |
+
"step": 937500
|
| 11376 |
+
},
|
| 11377 |
+
{
|
| 11378 |
+
"epoch": 13.5,
|
| 11379 |
+
"learning_rate": 3.6398568898997293e-06,
|
| 11380 |
+
"loss": 1.6989,
|
| 11381 |
+
"step": 938000
|
| 11382 |
+
},
|
| 11383 |
+
{
|
| 11384 |
+
"epoch": 13.51,
|
| 11385 |
+
"learning_rate": 3.5879301481469946e-06,
|
| 11386 |
+
"loss": 1.6989,
|
| 11387 |
+
"step": 938500
|
| 11388 |
+
},
|
| 11389 |
+
{
|
| 11390 |
+
"epoch": 13.52,
|
| 11391 |
+
"learning_rate": 3.5360034063942595e-06,
|
| 11392 |
+
"loss": 1.6994,
|
| 11393 |
+
"step": 939000
|
| 11394 |
+
},
|
| 11395 |
+
{
|
| 11396 |
+
"epoch": 13.52,
|
| 11397 |
+
"learning_rate": 3.484180518125029e-06,
|
| 11398 |
+
"loss": 1.7032,
|
| 11399 |
+
"step": 939500
|
| 11400 |
+
},
|
| 11401 |
+
{
|
| 11402 |
+
"epoch": 13.53,
|
| 11403 |
+
"learning_rate": 3.432253776372294e-06,
|
| 11404 |
+
"loss": 1.7012,
|
| 11405 |
+
"step": 940000
|
| 11406 |
+
},
|
| 11407 |
+
{
|
| 11408 |
+
"epoch": 13.54,
|
| 11409 |
+
"learning_rate": 3.380327034619559e-06,
|
| 11410 |
+
"loss": 1.6949,
|
| 11411 |
+
"step": 940500
|
| 11412 |
+
},
|
| 11413 |
+
{
|
| 11414 |
+
"epoch": 13.54,
|
| 11415 |
+
"learning_rate": 3.3284002928668234e-06,
|
| 11416 |
+
"loss": 1.6991,
|
| 11417 |
+
"step": 941000
|
| 11418 |
+
},
|
| 11419 |
+
{
|
| 11420 |
+
"epoch": 13.55,
|
| 11421 |
+
"learning_rate": 3.2764735511140883e-06,
|
| 11422 |
+
"loss": 1.6962,
|
| 11423 |
+
"step": 941500
|
| 11424 |
+
},
|
| 11425 |
+
{
|
| 11426 |
+
"epoch": 13.56,
|
| 11427 |
+
"learning_rate": 3.224650662844859e-06,
|
| 11428 |
+
"loss": 1.6984,
|
| 11429 |
+
"step": 942000
|
| 11430 |
+
},
|
| 11431 |
+
{
|
| 11432 |
+
"epoch": 13.57,
|
| 11433 |
+
"learning_rate": 3.1727239210921233e-06,
|
| 11434 |
+
"loss": 1.6974,
|
| 11435 |
+
"step": 942500
|
| 11436 |
+
},
|
| 11437 |
+
{
|
| 11438 |
+
"epoch": 13.57,
|
| 11439 |
+
"learning_rate": 3.120797179339388e-06,
|
| 11440 |
+
"loss": 1.6989,
|
| 11441 |
+
"step": 943000
|
| 11442 |
+
},
|
| 11443 |
+
{
|
| 11444 |
+
"epoch": 13.58,
|
| 11445 |
+
"learning_rate": 3.0688704375866527e-06,
|
| 11446 |
+
"loss": 1.6986,
|
| 11447 |
+
"step": 943500
|
| 11448 |
+
},
|
| 11449 |
+
{
|
| 11450 |
+
"epoch": 13.59,
|
| 11451 |
+
"learning_rate": 3.017047549317423e-06,
|
| 11452 |
+
"loss": 1.6972,
|
| 11453 |
+
"step": 944000
|
| 11454 |
+
},
|
| 11455 |
+
{
|
| 11456 |
+
"epoch": 13.6,
|
| 11457 |
+
"learning_rate": 2.9651208075646876e-06,
|
| 11458 |
+
"loss": 1.6985,
|
| 11459 |
+
"step": 944500
|
| 11460 |
+
},
|
| 11461 |
+
{
|
| 11462 |
+
"epoch": 13.6,
|
| 11463 |
+
"learning_rate": 2.913194065811953e-06,
|
| 11464 |
+
"loss": 1.6983,
|
| 11465 |
+
"step": 945000
|
| 11466 |
+
},
|
| 11467 |
+
{
|
| 11468 |
+
"epoch": 13.61,
|
| 11469 |
+
"learning_rate": 2.8612673240592174e-06,
|
| 11470 |
+
"loss": 1.7019,
|
| 11471 |
+
"step": 945500
|
| 11472 |
+
},
|
| 11473 |
+
{
|
| 11474 |
+
"epoch": 13.62,
|
| 11475 |
+
"learning_rate": 2.809444435789988e-06,
|
| 11476 |
+
"loss": 1.6978,
|
| 11477 |
+
"step": 946000
|
| 11478 |
+
},
|
| 11479 |
+
{
|
| 11480 |
+
"epoch": 13.62,
|
| 11481 |
+
"learning_rate": 2.7575176940372524e-06,
|
| 11482 |
+
"loss": 1.6995,
|
| 11483 |
+
"step": 946500
|
| 11484 |
+
},
|
| 11485 |
+
{
|
| 11486 |
+
"epoch": 13.63,
|
| 11487 |
+
"learning_rate": 2.705590952284517e-06,
|
| 11488 |
+
"loss": 1.6951,
|
| 11489 |
+
"step": 947000
|
| 11490 |
+
},
|
| 11491 |
+
{
|
| 11492 |
+
"epoch": 13.64,
|
| 11493 |
+
"learning_rate": 2.6536642105317818e-06,
|
| 11494 |
+
"loss": 1.7,
|
| 11495 |
+
"step": 947500
|
| 11496 |
+
},
|
| 11497 |
+
{
|
| 11498 |
+
"epoch": 13.65,
|
| 11499 |
+
"learning_rate": 2.6017374687790467e-06,
|
| 11500 |
+
"loss": 1.6994,
|
| 11501 |
+
"step": 948000
|
| 11502 |
+
},
|
| 11503 |
+
{
|
| 11504 |
+
"epoch": 13.65,
|
| 11505 |
+
"learning_rate": 2.5499145805098167e-06,
|
| 11506 |
+
"loss": 1.6966,
|
| 11507 |
+
"step": 948500
|
| 11508 |
+
},
|
| 11509 |
+
{
|
| 11510 |
+
"epoch": 13.66,
|
| 11511 |
+
"learning_rate": 2.4979878387570816e-06,
|
| 11512 |
+
"loss": 1.6966,
|
| 11513 |
+
"step": 949000
|
| 11514 |
+
},
|
| 11515 |
+
{
|
| 11516 |
+
"epoch": 13.67,
|
| 11517 |
+
"learning_rate": 2.4460610970043465e-06,
|
| 11518 |
+
"loss": 1.697,
|
| 11519 |
+
"step": 949500
|
| 11520 |
+
},
|
| 11521 |
+
{
|
| 11522 |
+
"epoch": 13.67,
|
| 11523 |
+
"learning_rate": 2.394134355251611e-06,
|
| 11524 |
+
"loss": 1.6988,
|
| 11525 |
+
"step": 950000
|
| 11526 |
+
},
|
| 11527 |
+
{
|
| 11528 |
+
"epoch": 13.68,
|
| 11529 |
+
"learning_rate": 2.342207613498876e-06,
|
| 11530 |
+
"loss": 1.6992,
|
| 11531 |
+
"step": 950500
|
| 11532 |
+
},
|
| 11533 |
+
{
|
| 11534 |
+
"epoch": 13.69,
|
| 11535 |
+
"learning_rate": 2.290280871746141e-06,
|
| 11536 |
+
"loss": 1.6961,
|
| 11537 |
+
"step": 951000
|
| 11538 |
+
},
|
| 11539 |
+
{
|
| 11540 |
+
"epoch": 13.7,
|
| 11541 |
+
"learning_rate": 2.238457983476911e-06,
|
| 11542 |
+
"loss": 1.7023,
|
| 11543 |
+
"step": 951500
|
| 11544 |
+
},
|
| 11545 |
+
{
|
| 11546 |
+
"epoch": 13.7,
|
| 11547 |
+
"learning_rate": 2.1865312417241758e-06,
|
| 11548 |
+
"loss": 1.7008,
|
| 11549 |
+
"step": 952000
|
| 11550 |
+
},
|
| 11551 |
+
{
|
| 11552 |
+
"epoch": 13.71,
|
| 11553 |
+
"learning_rate": 2.1346044999714403e-06,
|
| 11554 |
+
"loss": 1.7013,
|
| 11555 |
+
"step": 952500
|
| 11556 |
+
},
|
| 11557 |
+
{
|
| 11558 |
+
"epoch": 13.72,
|
| 11559 |
+
"learning_rate": 2.082677758218705e-06,
|
| 11560 |
+
"loss": 1.6989,
|
| 11561 |
+
"step": 953000
|
| 11562 |
+
},
|
| 11563 |
+
{
|
| 11564 |
+
"epoch": 13.72,
|
| 11565 |
+
"learning_rate": 2.0307510164659696e-06,
|
| 11566 |
+
"loss": 1.6959,
|
| 11567 |
+
"step": 953500
|
| 11568 |
+
},
|
| 11569 |
+
{
|
| 11570 |
+
"epoch": 13.73,
|
| 11571 |
+
"learning_rate": 1.97892812819674e-06,
|
| 11572 |
+
"loss": 1.6983,
|
| 11573 |
+
"step": 954000
|
| 11574 |
+
},
|
| 11575 |
+
{
|
| 11576 |
+
"epoch": 13.74,
|
| 11577 |
+
"learning_rate": 1.9270013864440046e-06,
|
| 11578 |
+
"loss": 1.7009,
|
| 11579 |
+
"step": 954500
|
| 11580 |
+
},
|
| 11581 |
+
{
|
| 11582 |
+
"epoch": 13.75,
|
| 11583 |
+
"learning_rate": 1.8750746446912697e-06,
|
| 11584 |
+
"loss": 1.6965,
|
| 11585 |
+
"step": 955000
|
| 11586 |
+
},
|
| 11587 |
+
{
|
| 11588 |
+
"epoch": 13.75,
|
| 11589 |
+
"learning_rate": 1.8231479029385344e-06,
|
| 11590 |
+
"loss": 1.697,
|
| 11591 |
+
"step": 955500
|
| 11592 |
+
},
|
| 11593 |
+
{
|
| 11594 |
+
"epoch": 13.76,
|
| 11595 |
+
"learning_rate": 1.7713250146693047e-06,
|
| 11596 |
+
"loss": 1.6969,
|
| 11597 |
+
"step": 956000
|
| 11598 |
+
},
|
| 11599 |
+
{
|
| 11600 |
+
"epoch": 13.77,
|
| 11601 |
+
"learning_rate": 1.7193982729165694e-06,
|
| 11602 |
+
"loss": 1.6936,
|
| 11603 |
+
"step": 956500
|
| 11604 |
+
},
|
| 11605 |
+
{
|
| 11606 |
+
"epoch": 13.78,
|
| 11607 |
+
"learning_rate": 1.667471531163834e-06,
|
| 11608 |
+
"loss": 1.7003,
|
| 11609 |
+
"step": 957000
|
| 11610 |
+
},
|
| 11611 |
+
{
|
| 11612 |
+
"epoch": 13.78,
|
| 11613 |
+
"learning_rate": 1.6155447894110987e-06,
|
| 11614 |
+
"loss": 1.6998,
|
| 11615 |
+
"step": 957500
|
| 11616 |
+
},
|
| 11617 |
+
{
|
| 11618 |
+
"epoch": 13.79,
|
| 11619 |
+
"learning_rate": 1.563721901141869e-06,
|
| 11620 |
+
"loss": 1.6969,
|
| 11621 |
+
"step": 958000
|
| 11622 |
+
},
|
| 11623 |
+
{
|
| 11624 |
+
"epoch": 13.8,
|
| 11625 |
+
"learning_rate": 1.511795159389134e-06,
|
| 11626 |
+
"loss": 1.6965,
|
| 11627 |
+
"step": 958500
|
| 11628 |
+
},
|
| 11629 |
+
{
|
| 11630 |
+
"epoch": 13.8,
|
| 11631 |
+
"learning_rate": 1.4598684176363986e-06,
|
| 11632 |
+
"loss": 1.6967,
|
| 11633 |
+
"step": 959000
|
| 11634 |
+
},
|
| 11635 |
+
{
|
| 11636 |
+
"epoch": 13.81,
|
| 11637 |
+
"learning_rate": 1.4079416758836633e-06,
|
| 11638 |
+
"loss": 1.6951,
|
| 11639 |
+
"step": 959500
|
| 11640 |
+
},
|
| 11641 |
+
{
|
| 11642 |
+
"epoch": 13.82,
|
| 11643 |
+
"learning_rate": 1.3561187876144336e-06,
|
| 11644 |
+
"loss": 1.6965,
|
| 11645 |
+
"step": 960000
|
| 11646 |
+
},
|
| 11647 |
+
{
|
| 11648 |
+
"epoch": 13.83,
|
| 11649 |
+
"learning_rate": 1.3041920458616985e-06,
|
| 11650 |
+
"loss": 1.696,
|
| 11651 |
+
"step": 960500
|
| 11652 |
+
},
|
| 11653 |
+
{
|
| 11654 |
+
"epoch": 13.83,
|
| 11655 |
+
"learning_rate": 1.2522653041089632e-06,
|
| 11656 |
+
"loss": 1.6984,
|
| 11657 |
+
"step": 961000
|
| 11658 |
+
},
|
| 11659 |
+
{
|
| 11660 |
+
"epoch": 13.84,
|
| 11661 |
+
"learning_rate": 1.2003385623562278e-06,
|
| 11662 |
+
"loss": 1.6992,
|
| 11663 |
+
"step": 961500
|
| 11664 |
+
},
|
| 11665 |
+
{
|
| 11666 |
+
"epoch": 13.85,
|
| 11667 |
+
"learning_rate": 1.1484118206034927e-06,
|
| 11668 |
+
"loss": 1.6976,
|
| 11669 |
+
"step": 962000
|
| 11670 |
+
},
|
| 11671 |
+
{
|
| 11672 |
+
"epoch": 13.85,
|
| 11673 |
+
"learning_rate": 1.0965889323342628e-06,
|
| 11674 |
+
"loss": 1.6984,
|
| 11675 |
+
"step": 962500
|
| 11676 |
+
},
|
| 11677 |
+
{
|
| 11678 |
+
"epoch": 13.86,
|
| 11679 |
+
"learning_rate": 1.0446621905815277e-06,
|
| 11680 |
+
"loss": 1.6995,
|
| 11681 |
+
"step": 963000
|
| 11682 |
+
},
|
| 11683 |
+
{
|
| 11684 |
+
"epoch": 13.87,
|
| 11685 |
+
"learning_rate": 9.927354488287924e-07,
|
| 11686 |
+
"loss": 1.6993,
|
| 11687 |
+
"step": 963500
|
| 11688 |
+
},
|
| 11689 |
+
{
|
| 11690 |
+
"epoch": 13.88,
|
| 11691 |
+
"learning_rate": 9.408087070760572e-07,
|
| 11692 |
+
"loss": 1.6974,
|
| 11693 |
+
"step": 964000
|
| 11694 |
+
},
|
| 11695 |
+
{
|
| 11696 |
+
"epoch": 13.88,
|
| 11697 |
+
"learning_rate": 8.889858188068274e-07,
|
| 11698 |
+
"loss": 1.6952,
|
| 11699 |
+
"step": 964500
|
| 11700 |
+
},
|
| 11701 |
+
{
|
| 11702 |
+
"epoch": 13.89,
|
| 11703 |
+
"learning_rate": 8.370590770540923e-07,
|
| 11704 |
+
"loss": 1.6941,
|
| 11705 |
+
"step": 965000
|
| 11706 |
+
},
|
| 11707 |
+
{
|
| 11708 |
+
"epoch": 13.9,
|
| 11709 |
+
"learning_rate": 7.851323353013568e-07,
|
| 11710 |
+
"loss": 1.6948,
|
| 11711 |
+
"step": 965500
|
| 11712 |
+
},
|
| 11713 |
+
{
|
| 11714 |
+
"epoch": 13.9,
|
| 11715 |
+
"learning_rate": 7.332055935486216e-07,
|
| 11716 |
+
"loss": 1.6992,
|
| 11717 |
+
"step": 966000
|
| 11718 |
+
},
|
| 11719 |
+
{
|
| 11720 |
+
"epoch": 13.91,
|
| 11721 |
+
"learning_rate": 6.812788517958863e-07,
|
| 11722 |
+
"loss": 1.7009,
|
| 11723 |
+
"step": 966500
|
| 11724 |
+
},
|
| 11725 |
+
{
|
| 11726 |
+
"epoch": 13.92,
|
| 11727 |
+
"learning_rate": 6.294559635266566e-07,
|
| 11728 |
+
"loss": 1.6947,
|
| 11729 |
+
"step": 967000
|
| 11730 |
+
},
|
| 11731 |
+
{
|
| 11732 |
+
"epoch": 13.93,
|
| 11733 |
+
"learning_rate": 5.775292217739214e-07,
|
| 11734 |
+
"loss": 1.7006,
|
| 11735 |
+
"step": 967500
|
| 11736 |
+
},
|
| 11737 |
+
{
|
| 11738 |
+
"epoch": 13.93,
|
| 11739 |
+
"learning_rate": 5.256024800211861e-07,
|
| 11740 |
+
"loss": 1.7017,
|
| 11741 |
+
"step": 968000
|
| 11742 |
+
},
|
| 11743 |
+
{
|
| 11744 |
+
"epoch": 13.94,
|
| 11745 |
+
"learning_rate": 4.736757382684509e-07,
|
| 11746 |
+
"loss": 1.6951,
|
| 11747 |
+
"step": 968500
|
| 11748 |
+
},
|
| 11749 |
+
{
|
| 11750 |
+
"epoch": 13.95,
|
| 11751 |
+
"learning_rate": 4.218528499992211e-07,
|
| 11752 |
+
"loss": 1.6955,
|
| 11753 |
+
"step": 969000
|
| 11754 |
+
},
|
| 11755 |
+
{
|
| 11756 |
+
"epoch": 13.96,
|
| 11757 |
+
"learning_rate": 3.6992610824648584e-07,
|
| 11758 |
+
"loss": 1.6968,
|
| 11759 |
+
"step": 969500
|
| 11760 |
+
},
|
| 11761 |
+
{
|
| 11762 |
+
"epoch": 13.96,
|
| 11763 |
+
"learning_rate": 3.1799936649375064e-07,
|
| 11764 |
+
"loss": 1.6982,
|
| 11765 |
+
"step": 970000
|
| 11766 |
+
},
|
| 11767 |
+
{
|
| 11768 |
+
"epoch": 13.97,
|
| 11769 |
+
"learning_rate": 2.660726247410154e-07,
|
| 11770 |
+
"loss": 1.6923,
|
| 11771 |
+
"step": 970500
|
| 11772 |
+
},
|
| 11773 |
+
{
|
| 11774 |
+
"epoch": 13.98,
|
| 11775 |
+
"learning_rate": 2.1414588298828015e-07,
|
| 11776 |
+
"loss": 1.6925,
|
| 11777 |
+
"step": 971000
|
| 11778 |
+
},
|
| 11779 |
+
{
|
| 11780 |
+
"epoch": 13.98,
|
| 11781 |
+
"learning_rate": 1.6232299471905037e-07,
|
| 11782 |
+
"loss": 1.695,
|
| 11783 |
+
"step": 971500
|
| 11784 |
+
},
|
| 11785 |
+
{
|
| 11786 |
+
"epoch": 13.99,
|
| 11787 |
+
"learning_rate": 1.1039625296631512e-07,
|
| 11788 |
+
"loss": 1.6977,
|
| 11789 |
+
"step": 972000
|
| 11790 |
+
},
|
| 11791 |
+
{
|
| 11792 |
+
"epoch": 14.0,
|
| 11793 |
+
"learning_rate": 5.846951121357988e-08,
|
| 11794 |
+
"loss": 1.7,
|
| 11795 |
+
"step": 972500
|
| 11796 |
+
},
|
| 11797 |
+
{
|
| 11798 |
+
"epoch": 14.0,
|
| 11799 |
+
"eval_accuracy": 0.6690181349948218,
|
| 11800 |
+
"eval_loss": 1.5665007829666138,
|
| 11801 |
+
"eval_runtime": 1297.3439,
|
| 11802 |
+
"eval_samples_per_second": 415.416,
|
| 11803 |
+
"eval_steps_per_second": 25.964,
|
| 11804 |
+
"step": 972622
|
| 11805 |
}
|
| 11806 |
],
|
| 11807 |
"max_steps": 972622,
|
| 11808 |
"num_train_epochs": 14,
|
| 11809 |
+
"total_flos": 7.363280880004055e+18,
|
| 11810 |
"trial_name": null,
|
| 11811 |
"trial_params": null
|
| 11812 |
}
|
pytorch_model.bin
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 118243218
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:5f558a9b6175c80a1c75dcb143fa35aa44a75cbdb0c9e7f8f52c48de2931aed8
|
| 3 |
size 118243218
|
runs/Feb22_09-35-52_ubuntu-2004/events.out.tfevents.1677026198.ubuntu-2004.1870487.0
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
-
size
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:f1625624de4e425f7b80c826a541c861aee58acfb64ef0613b7acf8e2792d039
|
| 3 |
+
size 319466
|