Training in progress, epoch 13
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 sha256:3cf4bfff09a16707370556956675459f9d75876bfb14a22ccca62770b5a95bbc
|
| 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 sha256:612de748233d159437078bb7f4ee519273e837c5c8f59ae0d256c06a48bde03c
|
| 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:bfa02c10b9cd4c99264361fef392b6d9708225ffc3e69c7351c226b900a6c196
|
| 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 sha256:8ba5415d56f47c80da2bd5540c9633f62d6ea0cd4624ff0eb7758d31be968154
|
| 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 sha256:52058bdd1f404d63cb045dcfb79f7694db4c45d09eda931f288ba1a544e087e7
|
| 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,
|
|
@@ -10116,11 +10116,854 @@
|
|
| 10116 |
"eval_samples_per_second": 413.518,
|
| 10117 |
"eval_steps_per_second": 25.845,
|
| 10118 |
"step": 833676
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 10119 |
}
|
| 10120 |
],
|
| 10121 |
"max_steps": 972622,
|
| 10122 |
"num_train_epochs": 14,
|
| 10123 |
-
"total_flos": 6.
|
| 10124 |
"trial_name": null,
|
| 10125 |
"trial_params": null
|
| 10126 |
}
|
|
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 13.0,
|
| 5 |
+
"global_step": 903149,
|
| 6 |
"is_hyper_param_search": false,
|
| 7 |
"is_local_process_zero": true,
|
| 8 |
"is_world_process_zero": true,
|
|
|
|
| 10116 |
"eval_samples_per_second": 413.518,
|
| 10117 |
"eval_steps_per_second": 25.845,
|
| 10118 |
"step": 833676
|
| 10119 |
+
},
|
| 10120 |
+
{
|
| 10121 |
+
"epoch": 12.0,
|
| 10122 |
+
"learning_rate": 1.4435945767710915e-05,
|
| 10123 |
+
"loss": 1.7129,
|
| 10124 |
+
"step": 834000
|
| 10125 |
+
},
|
| 10126 |
+
{
|
| 10127 |
+
"epoch": 12.01,
|
| 10128 |
+
"learning_rate": 1.4384019025958179e-05,
|
| 10129 |
+
"loss": 1.7078,
|
| 10130 |
+
"step": 834500
|
| 10131 |
+
},
|
| 10132 |
+
{
|
| 10133 |
+
"epoch": 12.02,
|
| 10134 |
+
"learning_rate": 1.4332092284205445e-05,
|
| 10135 |
+
"loss": 1.7095,
|
| 10136 |
+
"step": 835000
|
| 10137 |
+
},
|
| 10138 |
+
{
|
| 10139 |
+
"epoch": 12.03,
|
| 10140 |
+
"learning_rate": 1.4280165542452708e-05,
|
| 10141 |
+
"loss": 1.7136,
|
| 10142 |
+
"step": 835500
|
| 10143 |
+
},
|
| 10144 |
+
{
|
| 10145 |
+
"epoch": 12.03,
|
| 10146 |
+
"learning_rate": 1.422834265418348e-05,
|
| 10147 |
+
"loss": 1.7119,
|
| 10148 |
+
"step": 836000
|
| 10149 |
+
},
|
| 10150 |
+
{
|
| 10151 |
+
"epoch": 12.04,
|
| 10152 |
+
"learning_rate": 1.4176415912430743e-05,
|
| 10153 |
+
"loss": 1.7085,
|
| 10154 |
+
"step": 836500
|
| 10155 |
+
},
|
| 10156 |
+
{
|
| 10157 |
+
"epoch": 12.05,
|
| 10158 |
+
"learning_rate": 1.4124489170678006e-05,
|
| 10159 |
+
"loss": 1.7114,
|
| 10160 |
+
"step": 837000
|
| 10161 |
+
},
|
| 10162 |
+
{
|
| 10163 |
+
"epoch": 12.06,
|
| 10164 |
+
"learning_rate": 1.4072562428925273e-05,
|
| 10165 |
+
"loss": 1.7107,
|
| 10166 |
+
"step": 837500
|
| 10167 |
+
},
|
| 10168 |
+
{
|
| 10169 |
+
"epoch": 12.06,
|
| 10170 |
+
"learning_rate": 1.4020739540656044e-05,
|
| 10171 |
+
"loss": 1.7106,
|
| 10172 |
+
"step": 838000
|
| 10173 |
+
},
|
| 10174 |
+
{
|
| 10175 |
+
"epoch": 12.07,
|
| 10176 |
+
"learning_rate": 1.3968812798903307e-05,
|
| 10177 |
+
"loss": 1.7099,
|
| 10178 |
+
"step": 838500
|
| 10179 |
+
},
|
| 10180 |
+
{
|
| 10181 |
+
"epoch": 12.08,
|
| 10182 |
+
"learning_rate": 1.3916886057150572e-05,
|
| 10183 |
+
"loss": 1.7152,
|
| 10184 |
+
"step": 839000
|
| 10185 |
+
},
|
| 10186 |
+
{
|
| 10187 |
+
"epoch": 12.08,
|
| 10188 |
+
"learning_rate": 1.3864959315397839e-05,
|
| 10189 |
+
"loss": 1.7166,
|
| 10190 |
+
"step": 839500
|
| 10191 |
+
},
|
| 10192 |
+
{
|
| 10193 |
+
"epoch": 12.09,
|
| 10194 |
+
"learning_rate": 1.3813136427128608e-05,
|
| 10195 |
+
"loss": 1.7123,
|
| 10196 |
+
"step": 840000
|
| 10197 |
+
},
|
| 10198 |
+
{
|
| 10199 |
+
"epoch": 12.1,
|
| 10200 |
+
"learning_rate": 1.3761209685375873e-05,
|
| 10201 |
+
"loss": 1.7127,
|
| 10202 |
+
"step": 840500
|
| 10203 |
+
},
|
| 10204 |
+
{
|
| 10205 |
+
"epoch": 12.11,
|
| 10206 |
+
"learning_rate": 1.3709282943623137e-05,
|
| 10207 |
+
"loss": 1.7144,
|
| 10208 |
+
"step": 841000
|
| 10209 |
+
},
|
| 10210 |
+
{
|
| 10211 |
+
"epoch": 12.11,
|
| 10212 |
+
"learning_rate": 1.3657356201870403e-05,
|
| 10213 |
+
"loss": 1.7121,
|
| 10214 |
+
"step": 841500
|
| 10215 |
+
},
|
| 10216 |
+
{
|
| 10217 |
+
"epoch": 12.12,
|
| 10218 |
+
"learning_rate": 1.3605429460117666e-05,
|
| 10219 |
+
"loss": 1.714,
|
| 10220 |
+
"step": 842000
|
| 10221 |
+
},
|
| 10222 |
+
{
|
| 10223 |
+
"epoch": 12.13,
|
| 10224 |
+
"learning_rate": 1.3553502718364933e-05,
|
| 10225 |
+
"loss": 1.7112,
|
| 10226 |
+
"step": 842500
|
| 10227 |
+
},
|
| 10228 |
+
{
|
| 10229 |
+
"epoch": 12.13,
|
| 10230 |
+
"learning_rate": 1.3501679830095701e-05,
|
| 10231 |
+
"loss": 1.7121,
|
| 10232 |
+
"step": 843000
|
| 10233 |
+
},
|
| 10234 |
+
{
|
| 10235 |
+
"epoch": 12.14,
|
| 10236 |
+
"learning_rate": 1.3449753088342967e-05,
|
| 10237 |
+
"loss": 1.706,
|
| 10238 |
+
"step": 843500
|
| 10239 |
+
},
|
| 10240 |
+
{
|
| 10241 |
+
"epoch": 12.15,
|
| 10242 |
+
"learning_rate": 1.339782634659023e-05,
|
| 10243 |
+
"loss": 1.7121,
|
| 10244 |
+
"step": 844000
|
| 10245 |
+
},
|
| 10246 |
+
{
|
| 10247 |
+
"epoch": 12.16,
|
| 10248 |
+
"learning_rate": 1.3345899604837494e-05,
|
| 10249 |
+
"loss": 1.7075,
|
| 10250 |
+
"step": 844500
|
| 10251 |
+
},
|
| 10252 |
+
{
|
| 10253 |
+
"epoch": 12.16,
|
| 10254 |
+
"learning_rate": 1.3294076716568265e-05,
|
| 10255 |
+
"loss": 1.7105,
|
| 10256 |
+
"step": 845000
|
| 10257 |
+
},
|
| 10258 |
+
{
|
| 10259 |
+
"epoch": 12.17,
|
| 10260 |
+
"learning_rate": 1.3242149974815532e-05,
|
| 10261 |
+
"loss": 1.7086,
|
| 10262 |
+
"step": 845500
|
| 10263 |
+
},
|
| 10264 |
+
{
|
| 10265 |
+
"epoch": 12.18,
|
| 10266 |
+
"learning_rate": 1.3190223233062795e-05,
|
| 10267 |
+
"loss": 1.7103,
|
| 10268 |
+
"step": 846000
|
| 10269 |
+
},
|
| 10270 |
+
{
|
| 10271 |
+
"epoch": 12.18,
|
| 10272 |
+
"learning_rate": 1.313829649131006e-05,
|
| 10273 |
+
"loss": 1.7082,
|
| 10274 |
+
"step": 846500
|
| 10275 |
+
},
|
| 10276 |
+
{
|
| 10277 |
+
"epoch": 12.19,
|
| 10278 |
+
"learning_rate": 1.3086369749557325e-05,
|
| 10279 |
+
"loss": 1.7101,
|
| 10280 |
+
"step": 847000
|
| 10281 |
+
},
|
| 10282 |
+
{
|
| 10283 |
+
"epoch": 12.2,
|
| 10284 |
+
"learning_rate": 1.3034546861288096e-05,
|
| 10285 |
+
"loss": 1.7145,
|
| 10286 |
+
"step": 847500
|
| 10287 |
+
},
|
| 10288 |
+
{
|
| 10289 |
+
"epoch": 12.21,
|
| 10290 |
+
"learning_rate": 1.2982620119535361e-05,
|
| 10291 |
+
"loss": 1.7118,
|
| 10292 |
+
"step": 848000
|
| 10293 |
+
},
|
| 10294 |
+
{
|
| 10295 |
+
"epoch": 12.21,
|
| 10296 |
+
"learning_rate": 1.2930693377782624e-05,
|
| 10297 |
+
"loss": 1.7077,
|
| 10298 |
+
"step": 848500
|
| 10299 |
+
},
|
| 10300 |
+
{
|
| 10301 |
+
"epoch": 12.22,
|
| 10302 |
+
"learning_rate": 1.2878766636029891e-05,
|
| 10303 |
+
"loss": 1.7097,
|
| 10304 |
+
"step": 849000
|
| 10305 |
+
},
|
| 10306 |
+
{
|
| 10307 |
+
"epoch": 12.23,
|
| 10308 |
+
"learning_rate": 1.2826943747760659e-05,
|
| 10309 |
+
"loss": 1.7084,
|
| 10310 |
+
"step": 849500
|
| 10311 |
+
},
|
| 10312 |
+
{
|
| 10313 |
+
"epoch": 12.23,
|
| 10314 |
+
"learning_rate": 1.2775017006007925e-05,
|
| 10315 |
+
"loss": 1.7086,
|
| 10316 |
+
"step": 850000
|
| 10317 |
+
},
|
| 10318 |
+
{
|
| 10319 |
+
"epoch": 12.24,
|
| 10320 |
+
"learning_rate": 1.2723090264255189e-05,
|
| 10321 |
+
"loss": 1.7108,
|
| 10322 |
+
"step": 850500
|
| 10323 |
+
},
|
| 10324 |
+
{
|
| 10325 |
+
"epoch": 12.25,
|
| 10326 |
+
"learning_rate": 1.2671163522502455e-05,
|
| 10327 |
+
"loss": 1.7097,
|
| 10328 |
+
"step": 851000
|
| 10329 |
+
},
|
| 10330 |
+
{
|
| 10331 |
+
"epoch": 12.26,
|
| 10332 |
+
"learning_rate": 1.2619340634233223e-05,
|
| 10333 |
+
"loss": 1.7124,
|
| 10334 |
+
"step": 851500
|
| 10335 |
+
},
|
| 10336 |
+
{
|
| 10337 |
+
"epoch": 12.26,
|
| 10338 |
+
"learning_rate": 1.256741389248049e-05,
|
| 10339 |
+
"loss": 1.7127,
|
| 10340 |
+
"step": 852000
|
| 10341 |
+
},
|
| 10342 |
+
{
|
| 10343 |
+
"epoch": 12.27,
|
| 10344 |
+
"learning_rate": 1.2515487150727753e-05,
|
| 10345 |
+
"loss": 1.7104,
|
| 10346 |
+
"step": 852500
|
| 10347 |
+
},
|
| 10348 |
+
{
|
| 10349 |
+
"epoch": 12.28,
|
| 10350 |
+
"learning_rate": 1.2463560408975018e-05,
|
| 10351 |
+
"loss": 1.7132,
|
| 10352 |
+
"step": 853000
|
| 10353 |
+
},
|
| 10354 |
+
{
|
| 10355 |
+
"epoch": 12.29,
|
| 10356 |
+
"learning_rate": 1.2411633667222283e-05,
|
| 10357 |
+
"loss": 1.7116,
|
| 10358 |
+
"step": 853500
|
| 10359 |
+
},
|
| 10360 |
+
{
|
| 10361 |
+
"epoch": 12.29,
|
| 10362 |
+
"learning_rate": 1.2359810778953054e-05,
|
| 10363 |
+
"loss": 1.7118,
|
| 10364 |
+
"step": 854000
|
| 10365 |
+
},
|
| 10366 |
+
{
|
| 10367 |
+
"epoch": 12.3,
|
| 10368 |
+
"learning_rate": 1.2307884037200317e-05,
|
| 10369 |
+
"loss": 1.7152,
|
| 10370 |
+
"step": 854500
|
| 10371 |
+
},
|
| 10372 |
+
{
|
| 10373 |
+
"epoch": 12.31,
|
| 10374 |
+
"learning_rate": 1.2255957295447582e-05,
|
| 10375 |
+
"loss": 1.7078,
|
| 10376 |
+
"step": 855000
|
| 10377 |
+
},
|
| 10378 |
+
{
|
| 10379 |
+
"epoch": 12.31,
|
| 10380 |
+
"learning_rate": 1.2204030553694847e-05,
|
| 10381 |
+
"loss": 1.7121,
|
| 10382 |
+
"step": 855500
|
| 10383 |
+
},
|
| 10384 |
+
{
|
| 10385 |
+
"epoch": 12.32,
|
| 10386 |
+
"learning_rate": 1.2152207665425617e-05,
|
| 10387 |
+
"loss": 1.7093,
|
| 10388 |
+
"step": 856000
|
| 10389 |
+
},
|
| 10390 |
+
{
|
| 10391 |
+
"epoch": 12.33,
|
| 10392 |
+
"learning_rate": 1.2100280923672883e-05,
|
| 10393 |
+
"loss": 1.708,
|
| 10394 |
+
"step": 856500
|
| 10395 |
+
},
|
| 10396 |
+
{
|
| 10397 |
+
"epoch": 12.34,
|
| 10398 |
+
"learning_rate": 1.2048354181920148e-05,
|
| 10399 |
+
"loss": 1.7122,
|
| 10400 |
+
"step": 857000
|
| 10401 |
+
},
|
| 10402 |
+
{
|
| 10403 |
+
"epoch": 12.34,
|
| 10404 |
+
"learning_rate": 1.1996427440167413e-05,
|
| 10405 |
+
"loss": 1.7071,
|
| 10406 |
+
"step": 857500
|
| 10407 |
+
},
|
| 10408 |
+
{
|
| 10409 |
+
"epoch": 12.35,
|
| 10410 |
+
"learning_rate": 1.1944604551898183e-05,
|
| 10411 |
+
"loss": 1.7148,
|
| 10412 |
+
"step": 858000
|
| 10413 |
+
},
|
| 10414 |
+
{
|
| 10415 |
+
"epoch": 12.36,
|
| 10416 |
+
"learning_rate": 1.1892677810145448e-05,
|
| 10417 |
+
"loss": 1.7069,
|
| 10418 |
+
"step": 858500
|
| 10419 |
+
},
|
| 10420 |
+
{
|
| 10421 |
+
"epoch": 12.36,
|
| 10422 |
+
"learning_rate": 1.1840751068392713e-05,
|
| 10423 |
+
"loss": 1.7117,
|
| 10424 |
+
"step": 859000
|
| 10425 |
+
},
|
| 10426 |
+
{
|
| 10427 |
+
"epoch": 12.37,
|
| 10428 |
+
"learning_rate": 1.1788824326639977e-05,
|
| 10429 |
+
"loss": 1.7114,
|
| 10430 |
+
"step": 859500
|
| 10431 |
+
},
|
| 10432 |
+
{
|
| 10433 |
+
"epoch": 12.38,
|
| 10434 |
+
"learning_rate": 1.1737001438370747e-05,
|
| 10435 |
+
"loss": 1.7095,
|
| 10436 |
+
"step": 860000
|
| 10437 |
+
},
|
| 10438 |
+
{
|
| 10439 |
+
"epoch": 12.39,
|
| 10440 |
+
"learning_rate": 1.1685074696618012e-05,
|
| 10441 |
+
"loss": 1.7103,
|
| 10442 |
+
"step": 860500
|
| 10443 |
+
},
|
| 10444 |
+
{
|
| 10445 |
+
"epoch": 12.39,
|
| 10446 |
+
"learning_rate": 1.1633147954865277e-05,
|
| 10447 |
+
"loss": 1.7089,
|
| 10448 |
+
"step": 861000
|
| 10449 |
+
},
|
| 10450 |
+
{
|
| 10451 |
+
"epoch": 12.4,
|
| 10452 |
+
"learning_rate": 1.1581221213112542e-05,
|
| 10453 |
+
"loss": 1.709,
|
| 10454 |
+
"step": 861500
|
| 10455 |
+
},
|
| 10456 |
+
{
|
| 10457 |
+
"epoch": 12.41,
|
| 10458 |
+
"learning_rate": 1.1529294471359805e-05,
|
| 10459 |
+
"loss": 1.7089,
|
| 10460 |
+
"step": 862000
|
| 10461 |
+
},
|
| 10462 |
+
{
|
| 10463 |
+
"epoch": 12.41,
|
| 10464 |
+
"learning_rate": 1.147736772960707e-05,
|
| 10465 |
+
"loss": 1.7028,
|
| 10466 |
+
"step": 862500
|
| 10467 |
+
},
|
| 10468 |
+
{
|
| 10469 |
+
"epoch": 12.42,
|
| 10470 |
+
"learning_rate": 1.1425544841337841e-05,
|
| 10471 |
+
"loss": 1.7084,
|
| 10472 |
+
"step": 863000
|
| 10473 |
+
},
|
| 10474 |
+
{
|
| 10475 |
+
"epoch": 12.43,
|
| 10476 |
+
"learning_rate": 1.1373618099585104e-05,
|
| 10477 |
+
"loss": 1.7113,
|
| 10478 |
+
"step": 863500
|
| 10479 |
+
},
|
| 10480 |
+
{
|
| 10481 |
+
"epoch": 12.44,
|
| 10482 |
+
"learning_rate": 1.1321691357832371e-05,
|
| 10483 |
+
"loss": 1.7077,
|
| 10484 |
+
"step": 864000
|
| 10485 |
+
},
|
| 10486 |
+
{
|
| 10487 |
+
"epoch": 12.44,
|
| 10488 |
+
"learning_rate": 1.1269764616079636e-05,
|
| 10489 |
+
"loss": 1.7084,
|
| 10490 |
+
"step": 864500
|
| 10491 |
+
},
|
| 10492 |
+
{
|
| 10493 |
+
"epoch": 12.45,
|
| 10494 |
+
"learning_rate": 1.1217837874326901e-05,
|
| 10495 |
+
"loss": 1.7084,
|
| 10496 |
+
"step": 865000
|
| 10497 |
+
},
|
| 10498 |
+
{
|
| 10499 |
+
"epoch": 12.46,
|
| 10500 |
+
"learning_rate": 1.116601498605767e-05,
|
| 10501 |
+
"loss": 1.7087,
|
| 10502 |
+
"step": 865500
|
| 10503 |
+
},
|
| 10504 |
+
{
|
| 10505 |
+
"epoch": 12.47,
|
| 10506 |
+
"learning_rate": 1.1114088244304935e-05,
|
| 10507 |
+
"loss": 1.7086,
|
| 10508 |
+
"step": 866000
|
| 10509 |
+
},
|
| 10510 |
+
{
|
| 10511 |
+
"epoch": 12.47,
|
| 10512 |
+
"learning_rate": 1.10621615025522e-05,
|
| 10513 |
+
"loss": 1.7121,
|
| 10514 |
+
"step": 866500
|
| 10515 |
+
},
|
| 10516 |
+
{
|
| 10517 |
+
"epoch": 12.48,
|
| 10518 |
+
"learning_rate": 1.1010234760799465e-05,
|
| 10519 |
+
"loss": 1.7092,
|
| 10520 |
+
"step": 867000
|
| 10521 |
+
},
|
| 10522 |
+
{
|
| 10523 |
+
"epoch": 12.49,
|
| 10524 |
+
"learning_rate": 1.0958411872530235e-05,
|
| 10525 |
+
"loss": 1.7075,
|
| 10526 |
+
"step": 867500
|
| 10527 |
+
},
|
| 10528 |
+
{
|
| 10529 |
+
"epoch": 12.49,
|
| 10530 |
+
"learning_rate": 1.09064851307775e-05,
|
| 10531 |
+
"loss": 1.7053,
|
| 10532 |
+
"step": 868000
|
| 10533 |
+
},
|
| 10534 |
+
{
|
| 10535 |
+
"epoch": 12.5,
|
| 10536 |
+
"learning_rate": 1.0854558389024765e-05,
|
| 10537 |
+
"loss": 1.7118,
|
| 10538 |
+
"step": 868500
|
| 10539 |
+
},
|
| 10540 |
+
{
|
| 10541 |
+
"epoch": 12.51,
|
| 10542 |
+
"learning_rate": 1.080263164727203e-05,
|
| 10543 |
+
"loss": 1.7089,
|
| 10544 |
+
"step": 869000
|
| 10545 |
+
},
|
| 10546 |
+
{
|
| 10547 |
+
"epoch": 12.52,
|
| 10548 |
+
"learning_rate": 1.07508087590028e-05,
|
| 10549 |
+
"loss": 1.7085,
|
| 10550 |
+
"step": 869500
|
| 10551 |
+
},
|
| 10552 |
+
{
|
| 10553 |
+
"epoch": 12.52,
|
| 10554 |
+
"learning_rate": 1.0698882017250064e-05,
|
| 10555 |
+
"loss": 1.7098,
|
| 10556 |
+
"step": 870000
|
| 10557 |
+
},
|
| 10558 |
+
{
|
| 10559 |
+
"epoch": 12.53,
|
| 10560 |
+
"learning_rate": 1.0646955275497329e-05,
|
| 10561 |
+
"loss": 1.7075,
|
| 10562 |
+
"step": 870500
|
| 10563 |
+
},
|
| 10564 |
+
{
|
| 10565 |
+
"epoch": 12.54,
|
| 10566 |
+
"learning_rate": 1.0595028533744594e-05,
|
| 10567 |
+
"loss": 1.7067,
|
| 10568 |
+
"step": 871000
|
| 10569 |
+
},
|
| 10570 |
+
{
|
| 10571 |
+
"epoch": 12.54,
|
| 10572 |
+
"learning_rate": 1.0543101791991857e-05,
|
| 10573 |
+
"loss": 1.7055,
|
| 10574 |
+
"step": 871500
|
| 10575 |
+
},
|
| 10576 |
+
{
|
| 10577 |
+
"epoch": 12.55,
|
| 10578 |
+
"learning_rate": 1.0491175050239124e-05,
|
| 10579 |
+
"loss": 1.7062,
|
| 10580 |
+
"step": 872000
|
| 10581 |
+
},
|
| 10582 |
+
{
|
| 10583 |
+
"epoch": 12.56,
|
| 10584 |
+
"learning_rate": 1.0439352161969893e-05,
|
| 10585 |
+
"loss": 1.708,
|
| 10586 |
+
"step": 872500
|
| 10587 |
+
},
|
| 10588 |
+
{
|
| 10589 |
+
"epoch": 12.57,
|
| 10590 |
+
"learning_rate": 1.0387425420217158e-05,
|
| 10591 |
+
"loss": 1.7085,
|
| 10592 |
+
"step": 873000
|
| 10593 |
+
},
|
| 10594 |
+
{
|
| 10595 |
+
"epoch": 12.57,
|
| 10596 |
+
"learning_rate": 1.0335498678464423e-05,
|
| 10597 |
+
"loss": 1.7083,
|
| 10598 |
+
"step": 873500
|
| 10599 |
+
},
|
| 10600 |
+
{
|
| 10601 |
+
"epoch": 12.58,
|
| 10602 |
+
"learning_rate": 1.0283571936711688e-05,
|
| 10603 |
+
"loss": 1.7091,
|
| 10604 |
+
"step": 874000
|
| 10605 |
+
},
|
| 10606 |
+
{
|
| 10607 |
+
"epoch": 12.59,
|
| 10608 |
+
"learning_rate": 1.0231645194958953e-05,
|
| 10609 |
+
"loss": 1.707,
|
| 10610 |
+
"step": 874500
|
| 10611 |
+
},
|
| 10612 |
+
{
|
| 10613 |
+
"epoch": 12.59,
|
| 10614 |
+
"learning_rate": 1.0179822306689723e-05,
|
| 10615 |
+
"loss": 1.7085,
|
| 10616 |
+
"step": 875000
|
| 10617 |
+
},
|
| 10618 |
+
{
|
| 10619 |
+
"epoch": 12.6,
|
| 10620 |
+
"learning_rate": 1.0127895564936987e-05,
|
| 10621 |
+
"loss": 1.7052,
|
| 10622 |
+
"step": 875500
|
| 10623 |
+
},
|
| 10624 |
+
{
|
| 10625 |
+
"epoch": 12.61,
|
| 10626 |
+
"learning_rate": 1.0075968823184252e-05,
|
| 10627 |
+
"loss": 1.7092,
|
| 10628 |
+
"step": 876000
|
| 10629 |
+
},
|
| 10630 |
+
{
|
| 10631 |
+
"epoch": 12.62,
|
| 10632 |
+
"learning_rate": 1.0024042081431517e-05,
|
| 10633 |
+
"loss": 1.705,
|
| 10634 |
+
"step": 876500
|
| 10635 |
+
},
|
| 10636 |
+
{
|
| 10637 |
+
"epoch": 12.62,
|
| 10638 |
+
"learning_rate": 9.972219193162287e-06,
|
| 10639 |
+
"loss": 1.7111,
|
| 10640 |
+
"step": 877000
|
| 10641 |
+
},
|
| 10642 |
+
{
|
| 10643 |
+
"epoch": 12.63,
|
| 10644 |
+
"learning_rate": 9.920292451409552e-06,
|
| 10645 |
+
"loss": 1.712,
|
| 10646 |
+
"step": 877500
|
| 10647 |
+
},
|
| 10648 |
+
{
|
| 10649 |
+
"epoch": 12.64,
|
| 10650 |
+
"learning_rate": 9.868365709656817e-06,
|
| 10651 |
+
"loss": 1.7061,
|
| 10652 |
+
"step": 878000
|
| 10653 |
+
},
|
| 10654 |
+
{
|
| 10655 |
+
"epoch": 12.65,
|
| 10656 |
+
"learning_rate": 9.816438967904082e-06,
|
| 10657 |
+
"loss": 1.7059,
|
| 10658 |
+
"step": 878500
|
| 10659 |
+
},
|
| 10660 |
+
{
|
| 10661 |
+
"epoch": 12.65,
|
| 10662 |
+
"learning_rate": 9.764616079634851e-06,
|
| 10663 |
+
"loss": 1.7086,
|
| 10664 |
+
"step": 879000
|
| 10665 |
+
},
|
| 10666 |
+
{
|
| 10667 |
+
"epoch": 12.66,
|
| 10668 |
+
"learning_rate": 9.712689337882116e-06,
|
| 10669 |
+
"loss": 1.7065,
|
| 10670 |
+
"step": 879500
|
| 10671 |
+
},
|
| 10672 |
+
{
|
| 10673 |
+
"epoch": 12.67,
|
| 10674 |
+
"learning_rate": 9.660762596129381e-06,
|
| 10675 |
+
"loss": 1.703,
|
| 10676 |
+
"step": 880000
|
| 10677 |
+
},
|
| 10678 |
+
{
|
| 10679 |
+
"epoch": 12.67,
|
| 10680 |
+
"learning_rate": 9.608835854376646e-06,
|
| 10681 |
+
"loss": 1.7063,
|
| 10682 |
+
"step": 880500
|
| 10683 |
+
},
|
| 10684 |
+
{
|
| 10685 |
+
"epoch": 12.68,
|
| 10686 |
+
"learning_rate": 9.556909112623911e-06,
|
| 10687 |
+
"loss": 1.7087,
|
| 10688 |
+
"step": 881000
|
| 10689 |
+
},
|
| 10690 |
+
{
|
| 10691 |
+
"epoch": 12.69,
|
| 10692 |
+
"learning_rate": 9.504982370871176e-06,
|
| 10693 |
+
"loss": 1.7095,
|
| 10694 |
+
"step": 881500
|
| 10695 |
+
},
|
| 10696 |
+
{
|
| 10697 |
+
"epoch": 12.7,
|
| 10698 |
+
"learning_rate": 9.45305562911844e-06,
|
| 10699 |
+
"loss": 1.7099,
|
| 10700 |
+
"step": 882000
|
| 10701 |
+
},
|
| 10702 |
+
{
|
| 10703 |
+
"epoch": 12.7,
|
| 10704 |
+
"learning_rate": 9.40123274084921e-06,
|
| 10705 |
+
"loss": 1.709,
|
| 10706 |
+
"step": 882500
|
| 10707 |
+
},
|
| 10708 |
+
{
|
| 10709 |
+
"epoch": 12.71,
|
| 10710 |
+
"learning_rate": 9.349305999096475e-06,
|
| 10711 |
+
"loss": 1.7032,
|
| 10712 |
+
"step": 883000
|
| 10713 |
+
},
|
| 10714 |
+
{
|
| 10715 |
+
"epoch": 12.72,
|
| 10716 |
+
"learning_rate": 9.29737925734374e-06,
|
| 10717 |
+
"loss": 1.7104,
|
| 10718 |
+
"step": 883500
|
| 10719 |
+
},
|
| 10720 |
+
{
|
| 10721 |
+
"epoch": 12.72,
|
| 10722 |
+
"learning_rate": 9.245452515591005e-06,
|
| 10723 |
+
"loss": 1.706,
|
| 10724 |
+
"step": 884000
|
| 10725 |
+
},
|
| 10726 |
+
{
|
| 10727 |
+
"epoch": 12.73,
|
| 10728 |
+
"learning_rate": 9.19352577383827e-06,
|
| 10729 |
+
"loss": 1.7081,
|
| 10730 |
+
"step": 884500
|
| 10731 |
+
},
|
| 10732 |
+
{
|
| 10733 |
+
"epoch": 12.74,
|
| 10734 |
+
"learning_rate": 9.14170288556904e-06,
|
| 10735 |
+
"loss": 1.7088,
|
| 10736 |
+
"step": 885000
|
| 10737 |
+
},
|
| 10738 |
+
{
|
| 10739 |
+
"epoch": 12.75,
|
| 10740 |
+
"learning_rate": 9.089776143816304e-06,
|
| 10741 |
+
"loss": 1.7035,
|
| 10742 |
+
"step": 885500
|
| 10743 |
+
},
|
| 10744 |
+
{
|
| 10745 |
+
"epoch": 12.75,
|
| 10746 |
+
"learning_rate": 9.03784940206357e-06,
|
| 10747 |
+
"loss": 1.7031,
|
| 10748 |
+
"step": 886000
|
| 10749 |
+
},
|
| 10750 |
+
{
|
| 10751 |
+
"epoch": 12.76,
|
| 10752 |
+
"learning_rate": 8.985922660310834e-06,
|
| 10753 |
+
"loss": 1.7079,
|
| 10754 |
+
"step": 886500
|
| 10755 |
+
},
|
| 10756 |
+
{
|
| 10757 |
+
"epoch": 12.77,
|
| 10758 |
+
"learning_rate": 8.933995918558097e-06,
|
| 10759 |
+
"loss": 1.7058,
|
| 10760 |
+
"step": 887000
|
| 10761 |
+
},
|
| 10762 |
+
{
|
| 10763 |
+
"epoch": 12.77,
|
| 10764 |
+
"learning_rate": 8.882173030288869e-06,
|
| 10765 |
+
"loss": 1.7032,
|
| 10766 |
+
"step": 887500
|
| 10767 |
+
},
|
| 10768 |
+
{
|
| 10769 |
+
"epoch": 12.78,
|
| 10770 |
+
"learning_rate": 8.830246288536134e-06,
|
| 10771 |
+
"loss": 1.7037,
|
| 10772 |
+
"step": 888000
|
| 10773 |
+
},
|
| 10774 |
+
{
|
| 10775 |
+
"epoch": 12.79,
|
| 10776 |
+
"learning_rate": 8.778319546783399e-06,
|
| 10777 |
+
"loss": 1.703,
|
| 10778 |
+
"step": 888500
|
| 10779 |
+
},
|
| 10780 |
+
{
|
| 10781 |
+
"epoch": 12.8,
|
| 10782 |
+
"learning_rate": 8.726392805030663e-06,
|
| 10783 |
+
"loss": 1.7074,
|
| 10784 |
+
"step": 889000
|
| 10785 |
+
},
|
| 10786 |
+
{
|
| 10787 |
+
"epoch": 12.8,
|
| 10788 |
+
"learning_rate": 8.674466063277928e-06,
|
| 10789 |
+
"loss": 1.7061,
|
| 10790 |
+
"step": 889500
|
| 10791 |
+
},
|
| 10792 |
+
{
|
| 10793 |
+
"epoch": 12.81,
|
| 10794 |
+
"learning_rate": 8.622643175008698e-06,
|
| 10795 |
+
"loss": 1.7065,
|
| 10796 |
+
"step": 890000
|
| 10797 |
+
},
|
| 10798 |
+
{
|
| 10799 |
+
"epoch": 12.82,
|
| 10800 |
+
"learning_rate": 8.570716433255963e-06,
|
| 10801 |
+
"loss": 1.7058,
|
| 10802 |
+
"step": 890500
|
| 10803 |
+
},
|
| 10804 |
+
{
|
| 10805 |
+
"epoch": 12.83,
|
| 10806 |
+
"learning_rate": 8.518789691503228e-06,
|
| 10807 |
+
"loss": 1.7054,
|
| 10808 |
+
"step": 891000
|
| 10809 |
+
},
|
| 10810 |
+
{
|
| 10811 |
+
"epoch": 12.83,
|
| 10812 |
+
"learning_rate": 8.466862949750493e-06,
|
| 10813 |
+
"loss": 1.7071,
|
| 10814 |
+
"step": 891500
|
| 10815 |
+
},
|
| 10816 |
+
{
|
| 10817 |
+
"epoch": 12.84,
|
| 10818 |
+
"learning_rate": 8.415040061481262e-06,
|
| 10819 |
+
"loss": 1.703,
|
| 10820 |
+
"step": 892000
|
| 10821 |
+
},
|
| 10822 |
+
{
|
| 10823 |
+
"epoch": 12.85,
|
| 10824 |
+
"learning_rate": 8.363113319728527e-06,
|
| 10825 |
+
"loss": 1.7054,
|
| 10826 |
+
"step": 892500
|
| 10827 |
+
},
|
| 10828 |
+
{
|
| 10829 |
+
"epoch": 12.85,
|
| 10830 |
+
"learning_rate": 8.311186577975792e-06,
|
| 10831 |
+
"loss": 1.7059,
|
| 10832 |
+
"step": 893000
|
| 10833 |
+
},
|
| 10834 |
+
{
|
| 10835 |
+
"epoch": 12.86,
|
| 10836 |
+
"learning_rate": 8.259259836223057e-06,
|
| 10837 |
+
"loss": 1.7043,
|
| 10838 |
+
"step": 893500
|
| 10839 |
+
},
|
| 10840 |
+
{
|
| 10841 |
+
"epoch": 12.87,
|
| 10842 |
+
"learning_rate": 8.207436947953827e-06,
|
| 10843 |
+
"loss": 1.6995,
|
| 10844 |
+
"step": 894000
|
| 10845 |
+
},
|
| 10846 |
+
{
|
| 10847 |
+
"epoch": 12.88,
|
| 10848 |
+
"learning_rate": 8.155510206201092e-06,
|
| 10849 |
+
"loss": 1.7058,
|
| 10850 |
+
"step": 894500
|
| 10851 |
+
},
|
| 10852 |
+
{
|
| 10853 |
+
"epoch": 12.88,
|
| 10854 |
+
"learning_rate": 8.103583464448357e-06,
|
| 10855 |
+
"loss": 1.7018,
|
| 10856 |
+
"step": 895000
|
| 10857 |
+
},
|
| 10858 |
+
{
|
| 10859 |
+
"epoch": 12.89,
|
| 10860 |
+
"learning_rate": 8.051656722695621e-06,
|
| 10861 |
+
"loss": 1.7065,
|
| 10862 |
+
"step": 895500
|
| 10863 |
+
},
|
| 10864 |
+
{
|
| 10865 |
+
"epoch": 12.9,
|
| 10866 |
+
"learning_rate": 7.999729980942886e-06,
|
| 10867 |
+
"loss": 1.7049,
|
| 10868 |
+
"step": 896000
|
| 10869 |
+
},
|
| 10870 |
+
{
|
| 10871 |
+
"epoch": 12.9,
|
| 10872 |
+
"learning_rate": 7.947907092673656e-06,
|
| 10873 |
+
"loss": 1.7068,
|
| 10874 |
+
"step": 896500
|
| 10875 |
+
},
|
| 10876 |
+
{
|
| 10877 |
+
"epoch": 12.91,
|
| 10878 |
+
"learning_rate": 7.895980350920921e-06,
|
| 10879 |
+
"loss": 1.7001,
|
| 10880 |
+
"step": 897000
|
| 10881 |
+
},
|
| 10882 |
+
{
|
| 10883 |
+
"epoch": 12.92,
|
| 10884 |
+
"learning_rate": 7.844053609168186e-06,
|
| 10885 |
+
"loss": 1.7098,
|
| 10886 |
+
"step": 897500
|
| 10887 |
+
},
|
| 10888 |
+
{
|
| 10889 |
+
"epoch": 12.93,
|
| 10890 |
+
"learning_rate": 7.79212686741545e-06,
|
| 10891 |
+
"loss": 1.7014,
|
| 10892 |
+
"step": 898000
|
| 10893 |
+
},
|
| 10894 |
+
{
|
| 10895 |
+
"epoch": 12.93,
|
| 10896 |
+
"learning_rate": 7.740303979146222e-06,
|
| 10897 |
+
"loss": 1.7061,
|
| 10898 |
+
"step": 898500
|
| 10899 |
+
},
|
| 10900 |
+
{
|
| 10901 |
+
"epoch": 12.94,
|
| 10902 |
+
"learning_rate": 7.688377237393487e-06,
|
| 10903 |
+
"loss": 1.705,
|
| 10904 |
+
"step": 899000
|
| 10905 |
+
},
|
| 10906 |
+
{
|
| 10907 |
+
"epoch": 12.95,
|
| 10908 |
+
"learning_rate": 7.63645049564075e-06,
|
| 10909 |
+
"loss": 1.7037,
|
| 10910 |
+
"step": 899500
|
| 10911 |
+
},
|
| 10912 |
+
{
|
| 10913 |
+
"epoch": 12.95,
|
| 10914 |
+
"learning_rate": 7.584523753888015e-06,
|
| 10915 |
+
"loss": 1.7083,
|
| 10916 |
+
"step": 900000
|
| 10917 |
+
},
|
| 10918 |
+
{
|
| 10919 |
+
"epoch": 12.96,
|
| 10920 |
+
"learning_rate": 7.53259701213528e-06,
|
| 10921 |
+
"loss": 1.7029,
|
| 10922 |
+
"step": 900500
|
| 10923 |
+
},
|
| 10924 |
+
{
|
| 10925 |
+
"epoch": 12.97,
|
| 10926 |
+
"learning_rate": 7.4807741238660495e-06,
|
| 10927 |
+
"loss": 1.7047,
|
| 10928 |
+
"step": 901000
|
| 10929 |
+
},
|
| 10930 |
+
{
|
| 10931 |
+
"epoch": 12.98,
|
| 10932 |
+
"learning_rate": 7.4288473821133144e-06,
|
| 10933 |
+
"loss": 1.7049,
|
| 10934 |
+
"step": 901500
|
| 10935 |
+
},
|
| 10936 |
+
{
|
| 10937 |
+
"epoch": 12.98,
|
| 10938 |
+
"learning_rate": 7.376920640360579e-06,
|
| 10939 |
+
"loss": 1.7073,
|
| 10940 |
+
"step": 902000
|
| 10941 |
+
},
|
| 10942 |
+
{
|
| 10943 |
+
"epoch": 12.99,
|
| 10944 |
+
"learning_rate": 7.324993898607844e-06,
|
| 10945 |
+
"loss": 1.7036,
|
| 10946 |
+
"step": 902500
|
| 10947 |
+
},
|
| 10948 |
+
{
|
| 10949 |
+
"epoch": 13.0,
|
| 10950 |
+
"learning_rate": 7.273171010338614e-06,
|
| 10951 |
+
"loss": 1.7011,
|
| 10952 |
+
"step": 903000
|
| 10953 |
+
},
|
| 10954 |
+
{
|
| 10955 |
+
"epoch": 13.0,
|
| 10956 |
+
"eval_accuracy": 0.6677293660710026,
|
| 10957 |
+
"eval_loss": 1.5736616849899292,
|
| 10958 |
+
"eval_runtime": 1292.8595,
|
| 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": 6.837275613018931e+18,
|
| 10967 |
"trial_name": null,
|
| 10968 |
"trial_params": null
|
| 10969 |
}
|
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 sha256:612de748233d159437078bb7f4ee519273e837c5c8f59ae0d256c06a48bde03c
|
| 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:e26da7adb61568c20149e3a3b8d6acebeecb52046b416908a25f52f3d232dca1
|
| 3 |
+
size 296897
|