Commit
•
45559ac
1
Parent(s):
f1dd9a3
ilsilfverskiold/iptc-newscodes-multilingual-text-classification
Browse files- README.md +28 -26
- all_results.json +22 -22
- eval_results.json +22 -22
- model.safetensors +1 -1
- runs/May16_10-03-31_f06314cdb888/events.out.tfevents.1715853811.f06314cdb888.1663.0 +3 -0
- runs/May16_10-03-31_f06314cdb888/events.out.tfevents.1715855400.f06314cdb888.1663.1 +3 -0
- trainer_state.json +1120 -861
- training_args.bin +1 -1
README.md
CHANGED
|
@@ -19,27 +19,27 @@ should probably proofread and complete it, then remove this comment. -->
|
|
| 19 |
|
| 20 |
This model is a fine-tuned version of [KB/bert-base-swedish-cased](https://huggingface.co/KB/bert-base-swedish-cased) on the None dataset.
|
| 21 |
It achieves the following results on the evaluation set:
|
| 22 |
-
- Loss: 0.
|
| 23 |
-
- Accuracy: 0.
|
| 24 |
-
- F1: 0.
|
| 25 |
-
- Precision: 0.
|
| 26 |
-
- Recall: 0.
|
| 27 |
-
- Accuracy Label Arts, culture, entertainment and media: 0.
|
| 28 |
-
- Accuracy Label Conflict, war and peace: 0.
|
| 29 |
-
- Accuracy Label Crime, law and justice: 0.
|
| 30 |
-
- Accuracy Label Disaster, accident, and emergency incident: 0.
|
| 31 |
-
- Accuracy Label Economy, business, and finance: 0.
|
| 32 |
-
- Accuracy Label Environment: 0.
|
| 33 |
-
- Accuracy Label Health: 0.
|
| 34 |
-
- Accuracy Label Human interest: 0.
|
| 35 |
- Accuracy Label Labour: 0.5
|
| 36 |
-
- Accuracy Label Lifestyle and leisure: 0.
|
| 37 |
-
- Accuracy Label Politics: 0.
|
| 38 |
- Accuracy Label Religion: 0.0
|
| 39 |
-
- Accuracy Label Science and technology: 0.
|
| 40 |
-
- Accuracy Label Society: 0.
|
| 41 |
- Accuracy Label Sport: 0.9615
|
| 42 |
-
- Accuracy Label Weather:
|
| 43 |
|
| 44 |
## Model description
|
| 45 |
|
|
@@ -73,14 +73,16 @@ The following hyperparameters were used during training:
|
|
| 73 |
|
| 74 |
| Training Loss | Epoch | Step | Validation Loss | Accuracy | F1 | Precision | Recall | Accuracy Label Arts, culture, entertainment and media | Accuracy Label Conflict, war and peace | Accuracy Label Crime, law and justice | Accuracy Label Disaster, accident, and emergency incident | Accuracy Label Economy, business, and finance | Accuracy Label Environment | Accuracy Label Health | Accuracy Label Human interest | Accuracy Label Labour | Accuracy Label Lifestyle and leisure | Accuracy Label Politics | Accuracy Label Religion | Accuracy Label Science and technology | Accuracy Label Society | Accuracy Label Sport | Accuracy Label Weather |
|
| 75 |
|:-------------:|:------:|:----:|:---------------:|:--------:|:------:|:---------:|:------:|:-----------------------------------------------------:|:--------------------------------------:|:-------------------------------------:|:---------------------------------------------------------:|:---------------------------------------------:|:--------------------------:|:---------------------:|:-----------------------------:|:---------------------:|:------------------------------------:|:-----------------------:|:-----------------------:|:-------------------------------------:|:----------------------:|:--------------------:|:----------------------:|
|
| 76 |
-
| 1.
|
| 77 |
-
| 1.
|
| 78 |
-
| 0.
|
| 79 |
-
| 0.
|
| 80 |
-
| 0.
|
| 81 |
-
| 0.
|
| 82 |
-
| 0.
|
| 83 |
-
| 0.
|
|
|
|
|
|
|
| 84 |
|
| 85 |
|
| 86 |
### Framework versions
|
|
|
|
| 19 |
|
| 20 |
This model is a fine-tuned version of [KB/bert-base-swedish-cased](https://huggingface.co/KB/bert-base-swedish-cased) on the None dataset.
|
| 21 |
It achieves the following results on the evaluation set:
|
| 22 |
+
- Loss: 0.8030
|
| 23 |
+
- Accuracy: 0.7431
|
| 24 |
+
- F1: 0.7474
|
| 25 |
+
- Precision: 0.7695
|
| 26 |
+
- Recall: 0.7431
|
| 27 |
+
- Accuracy Label Arts, culture, entertainment and media: 0.6842
|
| 28 |
+
- Accuracy Label Conflict, war and peace: 0.7351
|
| 29 |
+
- Accuracy Label Crime, law and justice: 0.8918
|
| 30 |
+
- Accuracy Label Disaster, accident, and emergency incident: 0.8699
|
| 31 |
+
- Accuracy Label Economy, business, and finance: 0.6893
|
| 32 |
+
- Accuracy Label Environment: 0.4483
|
| 33 |
+
- Accuracy Label Health: 0.7222
|
| 34 |
+
- Accuracy Label Human interest: 0.3182
|
| 35 |
- Accuracy Label Labour: 0.5
|
| 36 |
+
- Accuracy Label Lifestyle and leisure: 0.5556
|
| 37 |
+
- Accuracy Label Politics: 0.7909
|
| 38 |
- Accuracy Label Religion: 0.0
|
| 39 |
+
- Accuracy Label Science and technology: 0.4583
|
| 40 |
+
- Accuracy Label Society: 0.3538
|
| 41 |
- Accuracy Label Sport: 0.9615
|
| 42 |
+
- Accuracy Label Weather: 0.0
|
| 43 |
|
| 44 |
## Model description
|
| 45 |
|
|
|
|
| 73 |
|
| 74 |
| Training Loss | Epoch | Step | Validation Loss | Accuracy | F1 | Precision | Recall | Accuracy Label Arts, culture, entertainment and media | Accuracy Label Conflict, war and peace | Accuracy Label Crime, law and justice | Accuracy Label Disaster, accident, and emergency incident | Accuracy Label Economy, business, and finance | Accuracy Label Environment | Accuracy Label Health | Accuracy Label Human interest | Accuracy Label Labour | Accuracy Label Lifestyle and leisure | Accuracy Label Politics | Accuracy Label Religion | Accuracy Label Science and technology | Accuracy Label Society | Accuracy Label Sport | Accuracy Label Weather |
|
| 75 |
|:-------------:|:------:|:----:|:---------------:|:--------:|:------:|:---------:|:------:|:-----------------------------------------------------:|:--------------------------------------:|:-------------------------------------:|:---------------------------------------------------------:|:---------------------------------------------:|:--------------------------:|:---------------------:|:-----------------------------:|:---------------------:|:------------------------------------:|:-----------------------:|:-----------------------:|:-------------------------------------:|:----------------------:|:--------------------:|:----------------------:|
|
| 76 |
+
| 1.9761 | 0.2907 | 200 | 1.4046 | 0.6462 | 0.6164 | 0.6057 | 0.6462 | 0.3158 | 0.8315 | 0.7629 | 0.7055 | 0.5437 | 0.0 | 0.5 | 0.0 | 0.0 | 0.3333 | 0.4843 | 0.0 | 0.0833 | 0.0 | 0.9615 | 0.0 |
|
| 77 |
+
| 1.2153 | 0.5814 | 400 | 1.0225 | 0.6894 | 0.6868 | 0.7652 | 0.6894 | 0.7895 | 0.6554 | 0.8196 | 0.8562 | 0.6408 | 0.2414 | 0.8333 | 0.1364 | 0.0 | 0.6667 | 0.8467 | 0.0 | 0.375 | 0.0154 | 0.9615 | 1.0 |
|
| 78 |
+
| 0.954 | 0.8721 | 600 | 0.8858 | 0.7231 | 0.7138 | 0.7309 | 0.7231 | 0.7368 | 0.7795 | 0.8918 | 0.8699 | 0.6214 | 0.3448 | 0.8889 | 0.1818 | 1.0 | 0.5556 | 0.6899 | 0.0 | 0.25 | 0.0462 | 0.9615 | 1.0 |
|
| 79 |
+
| 0.6662 | 1.1628 | 800 | 0.9381 | 0.6881 | 0.7009 | 0.7618 | 0.6881 | 0.7895 | 0.6126 | 0.8454 | 0.8630 | 0.6505 | 0.4483 | 0.7222 | 0.2273 | 1.0 | 0.4444 | 0.8293 | 0.0 | 0.5417 | 0.2308 | 0.9615 | 1.0 |
|
| 80 |
+
| 0.5554 | 1.4535 | 1000 | 0.8791 | 0.7025 | 0.7124 | 0.7628 | 0.7025 | 0.7368 | 0.6478 | 0.9021 | 0.8562 | 0.6602 | 0.3103 | 0.7778 | 0.3636 | 0.5 | 0.5556 | 0.8084 | 0.0 | 0.5 | 0.1846 | 0.9615 | 1.0 |
|
| 81 |
+
| 0.4396 | 1.7442 | 1200 | 0.8275 | 0.7175 | 0.7280 | 0.7686 | 0.7175 | 0.7895 | 0.6631 | 0.8196 | 0.8836 | 0.6893 | 0.3793 | 0.8333 | 0.4091 | 0.5 | 0.5556 | 0.8362 | 0.0 | 0.4167 | 0.3692 | 0.9615 | 1.0 |
|
| 82 |
+
| 0.383 | 2.0349 | 1400 | 0.7929 | 0.745 | 0.7501 | 0.7653 | 0.745 | 0.6842 | 0.7841 | 0.8866 | 0.8767 | 0.7087 | 0.4483 | 0.7778 | 0.4091 | 0.5 | 0.5556 | 0.6899 | 0.0 | 0.4167 | 0.2923 | 0.9615 | 0.0 |
|
| 83 |
+
| 0.3418 | 2.3256 | 1600 | 0.8042 | 0.7438 | 0.7440 | 0.7686 | 0.7438 | 0.7895 | 0.7351 | 0.9072 | 0.8493 | 0.7864 | 0.4483 | 0.7778 | 0.3182 | 0.5 | 0.5556 | 0.7909 | 0.0 | 0.4167 | 0.1846 | 0.9615 | 0.0 |
|
| 84 |
+
| 0.248 | 2.6163 | 1800 | 0.8387 | 0.7275 | 0.7325 | 0.7610 | 0.7275 | 0.6842 | 0.6891 | 0.8814 | 0.8699 | 0.7573 | 0.4138 | 0.8333 | 0.4091 | 0.5 | 0.5556 | 0.8014 | 0.0 | 0.4167 | 0.2769 | 0.9615 | 0.0 |
|
| 85 |
+
| 0.2525 | 2.9070 | 2000 | 0.8137 | 0.735 | 0.7413 | 0.7697 | 0.735 | 0.6842 | 0.7106 | 0.8763 | 0.8699 | 0.6796 | 0.4483 | 0.7222 | 0.3636 | 0.5 | 0.5556 | 0.8153 | 0.0 | 0.4583 | 0.3385 | 0.9615 | 0.0 |
|
| 86 |
|
| 87 |
|
| 88 |
### Framework versions
|
all_results.json
CHANGED
|
@@ -1,28 +1,28 @@
|
|
| 1 |
{
|
| 2 |
"epoch": 3.0,
|
| 3 |
-
"eval_accuracy": 0.
|
| 4 |
-
"eval_accuracy_label_arts, culture, entertainment and media": 0.
|
| 5 |
-
"eval_accuracy_label_conflict, war and peace": 0.
|
| 6 |
-
"eval_accuracy_label_crime, law and justice": 0.
|
| 7 |
-
"eval_accuracy_label_disaster, accident, and emergency incident": 0.
|
| 8 |
-
"eval_accuracy_label_economy, business, and finance": 0.
|
| 9 |
-
"eval_accuracy_label_environment": 0.
|
| 10 |
-
"eval_accuracy_label_health": 0.
|
| 11 |
-
"eval_accuracy_label_human interest": 0.
|
| 12 |
"eval_accuracy_label_labour": 0.5,
|
| 13 |
-
"eval_accuracy_label_lifestyle and leisure": 0.
|
| 14 |
-
"eval_accuracy_label_politics": 0.
|
| 15 |
"eval_accuracy_label_religion": 0.0,
|
| 16 |
-
"eval_accuracy_label_science and technology": 0.
|
| 17 |
-
"eval_accuracy_label_society": 0.
|
| 18 |
"eval_accuracy_label_sport": 0.9615384615384616,
|
| 19 |
-
"eval_accuracy_label_weather":
|
| 20 |
-
"eval_f1": 0.
|
| 21 |
-
"eval_loss": 0.
|
| 22 |
-
"eval_precision": 0.
|
| 23 |
-
"eval_recall": 0.
|
| 24 |
-
"eval_runtime":
|
| 25 |
-
"eval_samples_per_second": 127.
|
| 26 |
-
"eval_steps_per_second":
|
| 27 |
-
"step":
|
| 28 |
}
|
|
|
|
| 1 |
{
|
| 2 |
"epoch": 3.0,
|
| 3 |
+
"eval_accuracy": 0.743125,
|
| 4 |
+
"eval_accuracy_label_arts, culture, entertainment and media": 0.6842105263157895,
|
| 5 |
+
"eval_accuracy_label_conflict, war and peace": 0.7350689127105666,
|
| 6 |
+
"eval_accuracy_label_crime, law and justice": 0.8917525773195877,
|
| 7 |
+
"eval_accuracy_label_disaster, accident, and emergency incident": 0.8698630136986302,
|
| 8 |
+
"eval_accuracy_label_economy, business, and finance": 0.6893203883495146,
|
| 9 |
+
"eval_accuracy_label_environment": 0.4482758620689655,
|
| 10 |
+
"eval_accuracy_label_health": 0.7222222222222222,
|
| 11 |
+
"eval_accuracy_label_human interest": 0.3181818181818182,
|
| 12 |
"eval_accuracy_label_labour": 0.5,
|
| 13 |
+
"eval_accuracy_label_lifestyle and leisure": 0.5555555555555556,
|
| 14 |
+
"eval_accuracy_label_politics": 0.7909407665505227,
|
| 15 |
"eval_accuracy_label_religion": 0.0,
|
| 16 |
+
"eval_accuracy_label_science and technology": 0.4583333333333333,
|
| 17 |
+
"eval_accuracy_label_society": 0.35384615384615387,
|
| 18 |
"eval_accuracy_label_sport": 0.9615384615384616,
|
| 19 |
+
"eval_accuracy_label_weather": 0.0,
|
| 20 |
+
"eval_f1": 0.747448709907019,
|
| 21 |
+
"eval_loss": 0.8030127882957458,
|
| 22 |
+
"eval_precision": 0.7695117491320402,
|
| 23 |
+
"eval_recall": 0.743125,
|
| 24 |
+
"eval_runtime": 12.5023,
|
| 25 |
+
"eval_samples_per_second": 127.976,
|
| 26 |
+
"eval_steps_per_second": 7.999,
|
| 27 |
+
"step": 2064
|
| 28 |
}
|
eval_results.json
CHANGED
|
@@ -1,28 +1,28 @@
|
|
| 1 |
{
|
| 2 |
"epoch": 3.0,
|
| 3 |
-
"eval_accuracy": 0.
|
| 4 |
-
"eval_accuracy_label_arts, culture, entertainment and media": 0.
|
| 5 |
-
"eval_accuracy_label_conflict, war and peace": 0.
|
| 6 |
-
"eval_accuracy_label_crime, law and justice": 0.
|
| 7 |
-
"eval_accuracy_label_disaster, accident, and emergency incident": 0.
|
| 8 |
-
"eval_accuracy_label_economy, business, and finance": 0.
|
| 9 |
-
"eval_accuracy_label_environment": 0.
|
| 10 |
-
"eval_accuracy_label_health": 0.
|
| 11 |
-
"eval_accuracy_label_human interest": 0.
|
| 12 |
"eval_accuracy_label_labour": 0.5,
|
| 13 |
-
"eval_accuracy_label_lifestyle and leisure": 0.
|
| 14 |
-
"eval_accuracy_label_politics": 0.
|
| 15 |
"eval_accuracy_label_religion": 0.0,
|
| 16 |
-
"eval_accuracy_label_science and technology": 0.
|
| 17 |
-
"eval_accuracy_label_society": 0.
|
| 18 |
"eval_accuracy_label_sport": 0.9615384615384616,
|
| 19 |
-
"eval_accuracy_label_weather":
|
| 20 |
-
"eval_f1": 0.
|
| 21 |
-
"eval_loss": 0.
|
| 22 |
-
"eval_precision": 0.
|
| 23 |
-
"eval_recall": 0.
|
| 24 |
-
"eval_runtime":
|
| 25 |
-
"eval_samples_per_second": 127.
|
| 26 |
-
"eval_steps_per_second":
|
| 27 |
-
"step":
|
| 28 |
}
|
|
|
|
| 1 |
{
|
| 2 |
"epoch": 3.0,
|
| 3 |
+
"eval_accuracy": 0.743125,
|
| 4 |
+
"eval_accuracy_label_arts, culture, entertainment and media": 0.6842105263157895,
|
| 5 |
+
"eval_accuracy_label_conflict, war and peace": 0.7350689127105666,
|
| 6 |
+
"eval_accuracy_label_crime, law and justice": 0.8917525773195877,
|
| 7 |
+
"eval_accuracy_label_disaster, accident, and emergency incident": 0.8698630136986302,
|
| 8 |
+
"eval_accuracy_label_economy, business, and finance": 0.6893203883495146,
|
| 9 |
+
"eval_accuracy_label_environment": 0.4482758620689655,
|
| 10 |
+
"eval_accuracy_label_health": 0.7222222222222222,
|
| 11 |
+
"eval_accuracy_label_human interest": 0.3181818181818182,
|
| 12 |
"eval_accuracy_label_labour": 0.5,
|
| 13 |
+
"eval_accuracy_label_lifestyle and leisure": 0.5555555555555556,
|
| 14 |
+
"eval_accuracy_label_politics": 0.7909407665505227,
|
| 15 |
"eval_accuracy_label_religion": 0.0,
|
| 16 |
+
"eval_accuracy_label_science and technology": 0.4583333333333333,
|
| 17 |
+
"eval_accuracy_label_society": 0.35384615384615387,
|
| 18 |
"eval_accuracy_label_sport": 0.9615384615384616,
|
| 19 |
+
"eval_accuracy_label_weather": 0.0,
|
| 20 |
+
"eval_f1": 0.747448709907019,
|
| 21 |
+
"eval_loss": 0.8030127882957458,
|
| 22 |
+
"eval_precision": 0.7695117491320402,
|
| 23 |
+
"eval_recall": 0.743125,
|
| 24 |
+
"eval_runtime": 12.5023,
|
| 25 |
+
"eval_samples_per_second": 127.976,
|
| 26 |
+
"eval_steps_per_second": 7.999,
|
| 27 |
+
"step": 2064
|
| 28 |
}
|
model.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 498836536
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:f966cfafb0cf3e8f20b1b1947021ad13c2f60723d52d1ebbc680443b433b5916
|
| 3 |
size 498836536
|
runs/May16_10-03-31_f06314cdb888/events.out.tfevents.1715853811.f06314cdb888.1663.0
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:72f9f6cacc514c81849156b4dab719ad9f3b54b8e337b9fe9a0ca3457d610af1
|
| 3 |
+
size 66341
|
runs/May16_10-03-31_f06314cdb888/events.out.tfevents.1715855400.f06314cdb888.1663.1
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:5007f802f60a833f27a74912c6f9044764e113584b5f4b7757feade04767c9b8
|
| 3 |
+
size 1774
|
trainer_state.json
CHANGED
|
@@ -3,1518 +3,1777 @@
|
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
"epoch": 3.0,
|
| 5 |
"eval_steps": 200,
|
| 6 |
-
"global_step":
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
| 10 |
"log_history": [
|
| 11 |
{
|
| 12 |
-
"epoch": 0.
|
| 13 |
-
"grad_norm": 5.
|
| 14 |
"learning_rate": 4.0000000000000003e-07,
|
| 15 |
-
"loss": 2.
|
| 16 |
"step": 10
|
| 17 |
},
|
| 18 |
{
|
| 19 |
-
"epoch": 0.
|
| 20 |
-
"grad_norm":
|
| 21 |
"learning_rate": 8.000000000000001e-07,
|
| 22 |
-
"loss": 2.
|
| 23 |
"step": 20
|
| 24 |
},
|
| 25 |
{
|
| 26 |
-
"epoch": 0.
|
| 27 |
-
"grad_norm":
|
| 28 |
"learning_rate": 1.2000000000000002e-06,
|
| 29 |
-
"loss": 2.
|
| 30 |
"step": 30
|
| 31 |
},
|
| 32 |
{
|
| 33 |
-
"epoch": 0.
|
| 34 |
-
"grad_norm":
|
| 35 |
"learning_rate": 1.6000000000000001e-06,
|
| 36 |
-
"loss": 2.
|
| 37 |
"step": 40
|
| 38 |
},
|
| 39 |
{
|
| 40 |
-
"epoch": 0.
|
| 41 |
-
"grad_norm":
|
| 42 |
"learning_rate": 2.0000000000000003e-06,
|
| 43 |
-
"loss": 2.
|
| 44 |
"step": 50
|
| 45 |
},
|
| 46 |
{
|
| 47 |
-
"epoch": 0.
|
| 48 |
-
"grad_norm":
|
| 49 |
"learning_rate": 2.4000000000000003e-06,
|
| 50 |
-
"loss": 2.
|
| 51 |
"step": 60
|
| 52 |
},
|
| 53 |
{
|
| 54 |
-
"epoch": 0.
|
| 55 |
-
"grad_norm": 6.
|
| 56 |
"learning_rate": 2.8000000000000003e-06,
|
| 57 |
-
"loss": 2.
|
| 58 |
"step": 70
|
| 59 |
},
|
| 60 |
{
|
| 61 |
-
"epoch": 0.
|
| 62 |
-
"grad_norm": 5.
|
| 63 |
"learning_rate": 3.2000000000000003e-06,
|
| 64 |
-
"loss": 2.
|
| 65 |
"step": 80
|
| 66 |
},
|
| 67 |
{
|
| 68 |
-
"epoch": 0.
|
| 69 |
-
"grad_norm": 5.
|
| 70 |
"learning_rate": 3.6000000000000003e-06,
|
| 71 |
-
"loss": 2.
|
| 72 |
"step": 90
|
| 73 |
},
|
| 74 |
{
|
| 75 |
-
"epoch": 0.
|
| 76 |
-
"grad_norm": 5.
|
| 77 |
"learning_rate": 4.000000000000001e-06,
|
| 78 |
-
"loss": 2.
|
| 79 |
"step": 100
|
| 80 |
},
|
| 81 |
{
|
| 82 |
-
"epoch": 0.
|
| 83 |
-
"grad_norm": 5.
|
| 84 |
"learning_rate": 4.4e-06,
|
| 85 |
-
"loss": 2.
|
| 86 |
"step": 110
|
| 87 |
},
|
| 88 |
{
|
| 89 |
-
"epoch": 0.
|
| 90 |
-
"grad_norm":
|
| 91 |
"learning_rate": 4.800000000000001e-06,
|
| 92 |
-
"loss": 2.
|
| 93 |
"step": 120
|
| 94 |
},
|
| 95 |
{
|
| 96 |
-
"epoch": 0.
|
| 97 |
-
"grad_norm":
|
| 98 |
"learning_rate": 5.2e-06,
|
| 99 |
-
"loss": 2.
|
| 100 |
"step": 130
|
| 101 |
},
|
| 102 |
{
|
| 103 |
-
"epoch": 0.
|
| 104 |
-
"grad_norm":
|
| 105 |
"learning_rate": 5.600000000000001e-06,
|
| 106 |
-
"loss": 2.
|
| 107 |
"step": 140
|
| 108 |
},
|
| 109 |
{
|
| 110 |
-
"epoch": 0.
|
| 111 |
-
"grad_norm":
|
| 112 |
"learning_rate": 6e-06,
|
| 113 |
-
"loss": 2.
|
| 114 |
"step": 150
|
| 115 |
},
|
| 116 |
{
|
| 117 |
-
"epoch": 0.
|
| 118 |
-
"grad_norm":
|
| 119 |
"learning_rate": 6.4000000000000006e-06,
|
| 120 |
-
"loss": 2.
|
| 121 |
"step": 160
|
| 122 |
},
|
| 123 |
{
|
| 124 |
-
"epoch": 0.
|
| 125 |
-
"grad_norm":
|
| 126 |
"learning_rate": 6.800000000000001e-06,
|
| 127 |
-
"loss":
|
| 128 |
"step": 170
|
| 129 |
},
|
| 130 |
{
|
| 131 |
-
"epoch": 0.
|
| 132 |
-
"grad_norm":
|
| 133 |
"learning_rate": 7.2000000000000005e-06,
|
| 134 |
-
"loss":
|
| 135 |
"step": 180
|
| 136 |
},
|
| 137 |
{
|
| 138 |
-
"epoch": 0.
|
| 139 |
-
"grad_norm": 7.
|
| 140 |
"learning_rate": 7.600000000000001e-06,
|
| 141 |
-
"loss": 1.
|
| 142 |
"step": 190
|
| 143 |
},
|
| 144 |
{
|
| 145 |
-
"epoch": 0.
|
| 146 |
-
"grad_norm":
|
| 147 |
"learning_rate": 8.000000000000001e-06,
|
| 148 |
-
"loss": 1.
|
| 149 |
"step": 200
|
| 150 |
},
|
| 151 |
{
|
| 152 |
-
"epoch": 0.
|
| 153 |
-
"eval_accuracy": 0.
|
| 154 |
-
"eval_accuracy_label_arts, culture, entertainment and media": 0.
|
| 155 |
-
"eval_accuracy_label_conflict, war and peace": 0.
|
| 156 |
-
"eval_accuracy_label_crime, law and justice": 0.
|
| 157 |
-
"eval_accuracy_label_disaster, accident, and emergency incident": 0.
|
| 158 |
-
"eval_accuracy_label_economy, business, and finance": 0.
|
| 159 |
-
"eval_accuracy_label_environment": 0.
|
| 160 |
-
"eval_accuracy_label_health": 0.
|
| 161 |
"eval_accuracy_label_human interest": 0.0,
|
| 162 |
-
"eval_accuracy_label_labour": 0.
|
| 163 |
-
"eval_accuracy_label_lifestyle and leisure": 0.
|
| 164 |
-
"eval_accuracy_label_politics": 0.
|
| 165 |
"eval_accuracy_label_religion": 0.0,
|
| 166 |
-
"eval_accuracy_label_science and technology": 0.
|
| 167 |
-
"eval_accuracy_label_society": 0.
|
| 168 |
"eval_accuracy_label_sport": 0.9615384615384616,
|
| 169 |
-
"eval_accuracy_label_weather":
|
| 170 |
-
"eval_f1": 0.
|
| 171 |
-
"eval_loss": 1.
|
| 172 |
-
"eval_precision": 0.
|
| 173 |
-
"eval_recall": 0.
|
| 174 |
-
"eval_runtime":
|
| 175 |
-
"eval_samples_per_second":
|
| 176 |
-
"eval_steps_per_second":
|
| 177 |
"step": 200
|
| 178 |
},
|
| 179 |
{
|
| 180 |
-
"epoch": 0.
|
| 181 |
-
"grad_norm": 7.
|
| 182 |
"learning_rate": 8.400000000000001e-06,
|
| 183 |
-
"loss": 1.
|
| 184 |
"step": 210
|
| 185 |
},
|
| 186 |
{
|
| 187 |
-
"epoch": 0.
|
| 188 |
-
"grad_norm": 8.
|
| 189 |
"learning_rate": 8.8e-06,
|
| 190 |
-
"loss": 1.
|
| 191 |
"step": 220
|
| 192 |
},
|
| 193 |
{
|
| 194 |
-
"epoch": 0.
|
| 195 |
-
"grad_norm":
|
| 196 |
"learning_rate": 9.200000000000002e-06,
|
| 197 |
-
"loss": 1.
|
| 198 |
"step": 230
|
| 199 |
},
|
| 200 |
{
|
| 201 |
-
"epoch": 0.
|
| 202 |
-
"grad_norm":
|
| 203 |
"learning_rate": 9.600000000000001e-06,
|
| 204 |
-
"loss": 1.
|
| 205 |
"step": 240
|
| 206 |
},
|
| 207 |
{
|
| 208 |
-
"epoch": 0.
|
| 209 |
-
"grad_norm": 13.
|
| 210 |
"learning_rate": 1e-05,
|
| 211 |
-
"loss": 1.
|
| 212 |
"step": 250
|
| 213 |
},
|
| 214 |
{
|
| 215 |
-
"epoch": 0.
|
| 216 |
-
"grad_norm":
|
| 217 |
"learning_rate": 1.04e-05,
|
| 218 |
-
"loss": 1.
|
| 219 |
"step": 260
|
| 220 |
},
|
| 221 |
{
|
| 222 |
-
"epoch": 0.
|
| 223 |
-
"grad_norm":
|
| 224 |
"learning_rate": 1.0800000000000002e-05,
|
| 225 |
-
"loss": 1.
|
| 226 |
"step": 270
|
| 227 |
},
|
| 228 |
{
|
| 229 |
-
"epoch": 0.
|
| 230 |
-
"grad_norm":
|
| 231 |
"learning_rate": 1.1200000000000001e-05,
|
| 232 |
-
"loss": 1.
|
| 233 |
"step": 280
|
| 234 |
},
|
| 235 |
{
|
| 236 |
-
"epoch": 0.
|
| 237 |
-
"grad_norm":
|
| 238 |
"learning_rate": 1.16e-05,
|
| 239 |
-
"loss": 1.
|
| 240 |
"step": 290
|
| 241 |
},
|
| 242 |
{
|
| 243 |
-
"epoch": 0.
|
| 244 |
-
"grad_norm":
|
| 245 |
"learning_rate": 1.2e-05,
|
| 246 |
-
"loss": 1.
|
| 247 |
"step": 300
|
| 248 |
},
|
| 249 |
{
|
| 250 |
-
"epoch": 0.
|
| 251 |
-
"grad_norm":
|
| 252 |
"learning_rate": 1.2400000000000002e-05,
|
| 253 |
-
"loss": 1.
|
| 254 |
"step": 310
|
| 255 |
},
|
| 256 |
{
|
| 257 |
-
"epoch": 0.
|
| 258 |
-
"grad_norm": 11.
|
| 259 |
"learning_rate": 1.2800000000000001e-05,
|
| 260 |
-
"loss": 1.
|
| 261 |
"step": 320
|
| 262 |
},
|
| 263 |
{
|
| 264 |
-
"epoch": 0.
|
| 265 |
-
"grad_norm":
|
| 266 |
"learning_rate": 1.3200000000000002e-05,
|
| 267 |
-
"loss": 1.
|
| 268 |
"step": 330
|
| 269 |
},
|
| 270 |
{
|
| 271 |
-
"epoch": 0.
|
| 272 |
-
"grad_norm":
|
| 273 |
"learning_rate": 1.3600000000000002e-05,
|
| 274 |
-
"loss": 1.
|
| 275 |
"step": 340
|
| 276 |
},
|
| 277 |
{
|
| 278 |
-
"epoch": 0.
|
| 279 |
-
"grad_norm":
|
| 280 |
"learning_rate": 1.4e-05,
|
| 281 |
-
"loss": 1.
|
| 282 |
"step": 350
|
| 283 |
},
|
| 284 |
{
|
| 285 |
-
"epoch": 0.
|
| 286 |
-
"grad_norm": 8.
|
| 287 |
"learning_rate": 1.4400000000000001e-05,
|
| 288 |
-
"loss": 1.
|
| 289 |
"step": 360
|
| 290 |
},
|
| 291 |
{
|
| 292 |
-
"epoch": 0.
|
| 293 |
-
"grad_norm":
|
| 294 |
"learning_rate": 1.48e-05,
|
| 295 |
-
"loss": 1.
|
| 296 |
"step": 370
|
| 297 |
},
|
| 298 |
{
|
| 299 |
-
"epoch": 0.
|
| 300 |
-
"grad_norm":
|
| 301 |
"learning_rate": 1.5200000000000002e-05,
|
| 302 |
-
"loss":
|
| 303 |
"step": 380
|
| 304 |
},
|
| 305 |
{
|
| 306 |
-
"epoch": 0.
|
| 307 |
-
"grad_norm":
|
| 308 |
"learning_rate": 1.5600000000000003e-05,
|
| 309 |
-
"loss":
|
| 310 |
"step": 390
|
| 311 |
},
|
| 312 |
{
|
| 313 |
-
"epoch": 0.
|
| 314 |
-
"grad_norm":
|
| 315 |
"learning_rate": 1.6000000000000003e-05,
|
| 316 |
-
"loss": 1.
|
| 317 |
"step": 400
|
| 318 |
},
|
| 319 |
{
|
| 320 |
-
"epoch": 0.
|
| 321 |
-
"eval_accuracy": 0.
|
| 322 |
-
"eval_accuracy_label_arts, culture, entertainment and media": 0.
|
| 323 |
-
"eval_accuracy_label_conflict, war and peace": 0.
|
| 324 |
-
"eval_accuracy_label_crime, law and justice": 0.
|
| 325 |
-
"eval_accuracy_label_disaster, accident, and emergency incident": 0.
|
| 326 |
-
"eval_accuracy_label_economy, business, and finance": 0.
|
| 327 |
-
"eval_accuracy_label_environment": 0.
|
| 328 |
-
"eval_accuracy_label_health": 0.
|
| 329 |
-
"eval_accuracy_label_human interest": 0.
|
| 330 |
-
"eval_accuracy_label_labour":
|
| 331 |
-
"eval_accuracy_label_lifestyle and leisure": 0.
|
| 332 |
-
"eval_accuracy_label_politics": 0.
|
| 333 |
"eval_accuracy_label_religion": 0.0,
|
| 334 |
-
"eval_accuracy_label_science and technology": 0.
|
| 335 |
-
"eval_accuracy_label_society": 0.
|
| 336 |
"eval_accuracy_label_sport": 0.9615384615384616,
|
| 337 |
"eval_accuracy_label_weather": 1.0,
|
| 338 |
-
"eval_f1": 0.
|
| 339 |
-
"eval_loss": 1.
|
| 340 |
-
"eval_precision": 0.
|
| 341 |
-
"eval_recall": 0.
|
| 342 |
-
"eval_runtime":
|
| 343 |
-
"eval_samples_per_second":
|
| 344 |
-
"eval_steps_per_second": 7.
|
| 345 |
"step": 400
|
| 346 |
},
|
| 347 |
{
|
| 348 |
-
"epoch": 0.
|
| 349 |
-
"grad_norm": 11.
|
| 350 |
"learning_rate": 1.64e-05,
|
| 351 |
-
"loss":
|
| 352 |
"step": 410
|
| 353 |
},
|
| 354 |
{
|
| 355 |
-
"epoch": 0.
|
| 356 |
-
"grad_norm":
|
| 357 |
"learning_rate": 1.6800000000000002e-05,
|
| 358 |
-
"loss":
|
| 359 |
"step": 420
|
| 360 |
},
|
| 361 |
{
|
| 362 |
-
"epoch": 0.
|
| 363 |
-
"grad_norm":
|
| 364 |
"learning_rate": 1.72e-05,
|
| 365 |
-
"loss":
|
| 366 |
"step": 430
|
| 367 |
},
|
| 368 |
{
|
| 369 |
-
"epoch": 0.
|
| 370 |
-
"grad_norm":
|
| 371 |
"learning_rate": 1.76e-05,
|
| 372 |
-
"loss": 1.
|
| 373 |
"step": 440
|
| 374 |
},
|
| 375 |
{
|
| 376 |
-
"epoch": 0.
|
| 377 |
-
"grad_norm":
|
| 378 |
"learning_rate": 1.8e-05,
|
| 379 |
-
"loss": 0.
|
| 380 |
"step": 450
|
| 381 |
},
|
| 382 |
{
|
| 383 |
-
"epoch": 0.
|
| 384 |
-
"grad_norm":
|
| 385 |
"learning_rate": 1.8400000000000003e-05,
|
| 386 |
-
"loss":
|
| 387 |
"step": 460
|
| 388 |
},
|
| 389 |
{
|
| 390 |
-
"epoch": 0.
|
| 391 |
-
"grad_norm":
|
| 392 |
"learning_rate": 1.88e-05,
|
| 393 |
-
"loss":
|
| 394 |
"step": 470
|
| 395 |
},
|
| 396 |
{
|
| 397 |
-
"epoch": 0.
|
| 398 |
-
"grad_norm":
|
| 399 |
"learning_rate": 1.9200000000000003e-05,
|
| 400 |
-
"loss":
|
| 401 |
"step": 480
|
| 402 |
},
|
| 403 |
{
|
| 404 |
-
"epoch": 0.
|
| 405 |
-
"grad_norm":
|
| 406 |
"learning_rate": 1.9600000000000002e-05,
|
| 407 |
-
"loss":
|
| 408 |
"step": 490
|
| 409 |
},
|
| 410 |
{
|
| 411 |
-
"epoch": 0.
|
| 412 |
-
"grad_norm":
|
| 413 |
"learning_rate": 2e-05,
|
| 414 |
-
"loss":
|
| 415 |
"step": 500
|
| 416 |
},
|
| 417 |
{
|
| 418 |
-
"epoch": 0.
|
| 419 |
-
"grad_norm":
|
| 420 |
-
"learning_rate": 1.
|
| 421 |
-
"loss": 0.
|
| 422 |
"step": 510
|
| 423 |
},
|
| 424 |
{
|
| 425 |
-
"epoch": 0.
|
| 426 |
-
"grad_norm":
|
| 427 |
-
"learning_rate": 1.
|
| 428 |
-
"loss": 0.
|
| 429 |
"step": 520
|
| 430 |
},
|
| 431 |
{
|
| 432 |
-
"epoch": 0.
|
| 433 |
-
"grad_norm":
|
| 434 |
-
"learning_rate": 1.
|
| 435 |
-
"loss": 0.
|
| 436 |
"step": 530
|
| 437 |
},
|
| 438 |
{
|
| 439 |
-
"epoch": 0.
|
| 440 |
-
"grad_norm":
|
| 441 |
-
"learning_rate": 1.
|
| 442 |
-
"loss": 0.
|
| 443 |
"step": 540
|
| 444 |
},
|
| 445 |
{
|
| 446 |
-
"epoch": 0.
|
| 447 |
-
"grad_norm":
|
| 448 |
-
"learning_rate": 1.
|
| 449 |
-
"loss":
|
| 450 |
"step": 550
|
| 451 |
},
|
| 452 |
{
|
| 453 |
-
"epoch": 0.
|
| 454 |
-
"grad_norm":
|
| 455 |
-
"learning_rate": 1.
|
| 456 |
-
"loss": 0.
|
| 457 |
"step": 560
|
| 458 |
},
|
| 459 |
{
|
| 460 |
-
"epoch": 0.
|
| 461 |
-
"grad_norm":
|
| 462 |
-
"learning_rate": 1.
|
| 463 |
-
"loss":
|
| 464 |
"step": 570
|
| 465 |
},
|
| 466 |
{
|
| 467 |
-
"epoch": 0.
|
| 468 |
-
"grad_norm":
|
| 469 |
-
"learning_rate": 1.
|
| 470 |
-
"loss": 0.
|
| 471 |
"step": 580
|
| 472 |
},
|
| 473 |
{
|
| 474 |
-
"epoch": 0.
|
| 475 |
-
"grad_norm":
|
| 476 |
-
"learning_rate": 1.
|
| 477 |
-
"loss": 0.
|
| 478 |
"step": 590
|
| 479 |
},
|
| 480 |
{
|
| 481 |
-
"epoch":
|
| 482 |
-
"grad_norm":
|
| 483 |
-
"learning_rate": 1.
|
| 484 |
-
"loss": 0.
|
| 485 |
"step": 600
|
| 486 |
},
|
| 487 |
{
|
| 488 |
-
"epoch":
|
| 489 |
-
"eval_accuracy": 0.
|
| 490 |
-
"eval_accuracy_label_arts, culture, entertainment and media": 0.
|
| 491 |
-
"eval_accuracy_label_conflict, war and peace": 0.
|
| 492 |
-
"eval_accuracy_label_crime, law and justice": 0.
|
| 493 |
-
"eval_accuracy_label_disaster, accident, and emergency incident": 0.
|
| 494 |
-
"eval_accuracy_label_economy, business, and finance": 0.
|
| 495 |
-
"eval_accuracy_label_environment": 0.
|
| 496 |
-
"eval_accuracy_label_health": 0.
|
| 497 |
-
"eval_accuracy_label_human interest": 0.
|
| 498 |
"eval_accuracy_label_labour": 1.0,
|
| 499 |
-
"eval_accuracy_label_lifestyle and leisure": 0.
|
| 500 |
-
"eval_accuracy_label_politics": 0.
|
| 501 |
"eval_accuracy_label_religion": 0.0,
|
| 502 |
-
"eval_accuracy_label_science and technology": 0.
|
| 503 |
-
"eval_accuracy_label_society": 0.
|
| 504 |
-
"eval_accuracy_label_sport": 0.
|
| 505 |
"eval_accuracy_label_weather": 1.0,
|
| 506 |
-
"eval_f1": 0.
|
| 507 |
-
"eval_loss": 0.
|
| 508 |
-
"eval_precision": 0.
|
| 509 |
-
"eval_recall": 0.
|
| 510 |
-
"eval_runtime":
|
| 511 |
-
"eval_samples_per_second":
|
| 512 |
-
"eval_steps_per_second": 7.
|
| 513 |
"step": 600
|
| 514 |
},
|
| 515 |
{
|
| 516 |
-
"epoch":
|
| 517 |
-
"grad_norm":
|
| 518 |
-
"learning_rate": 1.
|
| 519 |
-
"loss": 0.
|
| 520 |
"step": 610
|
| 521 |
},
|
| 522 |
{
|
| 523 |
-
"epoch":
|
| 524 |
-
"grad_norm":
|
| 525 |
-
"learning_rate": 1.
|
| 526 |
-
"loss": 0.
|
| 527 |
"step": 620
|
| 528 |
},
|
| 529 |
{
|
| 530 |
-
"epoch":
|
| 531 |
-
"grad_norm":
|
| 532 |
-
"learning_rate": 1.
|
| 533 |
-
"loss": 0.
|
| 534 |
"step": 630
|
| 535 |
},
|
| 536 |
{
|
| 537 |
-
"epoch":
|
| 538 |
-
"grad_norm":
|
| 539 |
-
"learning_rate": 1.
|
| 540 |
-
"loss": 0.
|
| 541 |
"step": 640
|
| 542 |
},
|
| 543 |
{
|
| 544 |
-
"epoch":
|
| 545 |
-
"grad_norm":
|
| 546 |
-
"learning_rate": 1.
|
| 547 |
-
"loss": 0.
|
| 548 |
"step": 650
|
| 549 |
},
|
| 550 |
{
|
| 551 |
-
"epoch":
|
| 552 |
-
"grad_norm":
|
| 553 |
-
"learning_rate": 1.
|
| 554 |
-
"loss": 0.
|
| 555 |
"step": 660
|
| 556 |
},
|
| 557 |
{
|
| 558 |
-
"epoch":
|
| 559 |
-
"grad_norm":
|
| 560 |
-
"learning_rate": 1.
|
| 561 |
-
"loss": 0.
|
| 562 |
"step": 670
|
| 563 |
},
|
| 564 |
{
|
| 565 |
-
"epoch":
|
| 566 |
-
"grad_norm":
|
| 567 |
-
"learning_rate": 1.
|
| 568 |
-
"loss": 0.
|
| 569 |
"step": 680
|
| 570 |
},
|
| 571 |
{
|
| 572 |
-
"epoch": 1.
|
| 573 |
-
"grad_norm":
|
| 574 |
-
"learning_rate": 1.
|
| 575 |
-
"loss": 0.
|
| 576 |
"step": 690
|
| 577 |
},
|
| 578 |
{
|
| 579 |
-
"epoch": 1.
|
| 580 |
-
"grad_norm":
|
| 581 |
-
"learning_rate": 1.
|
| 582 |
-
"loss": 0.
|
| 583 |
"step": 700
|
| 584 |
},
|
| 585 |
{
|
| 586 |
-
"epoch": 1.
|
| 587 |
-
"grad_norm":
|
| 588 |
-
"learning_rate": 1.
|
| 589 |
-
"loss": 0.
|
| 590 |
"step": 710
|
| 591 |
},
|
| 592 |
{
|
| 593 |
-
"epoch": 1.
|
| 594 |
-
"grad_norm":
|
| 595 |
-
"learning_rate": 1.
|
| 596 |
-
"loss": 0.
|
| 597 |
"step": 720
|
| 598 |
},
|
| 599 |
{
|
| 600 |
-
"epoch": 1.
|
| 601 |
-
"grad_norm":
|
| 602 |
-
"learning_rate": 1.
|
| 603 |
-
"loss": 0.
|
| 604 |
"step": 730
|
| 605 |
},
|
| 606 |
{
|
| 607 |
-
"epoch": 1.
|
| 608 |
-
"grad_norm":
|
| 609 |
-
"learning_rate": 1.
|
| 610 |
-
"loss": 0.
|
| 611 |
"step": 740
|
| 612 |
},
|
| 613 |
{
|
| 614 |
-
"epoch": 1.
|
| 615 |
-
"grad_norm":
|
| 616 |
-
"learning_rate": 1.
|
| 617 |
-
"loss": 0.
|
| 618 |
"step": 750
|
| 619 |
},
|
| 620 |
{
|
| 621 |
-
"epoch": 1.
|
| 622 |
-
"grad_norm":
|
| 623 |
-
"learning_rate": 1.
|
| 624 |
-
"loss": 0.
|
| 625 |
"step": 760
|
| 626 |
},
|
| 627 |
{
|
| 628 |
-
"epoch": 1.
|
| 629 |
-
"grad_norm":
|
| 630 |
-
"learning_rate": 1.
|
| 631 |
-
"loss": 0.
|
| 632 |
"step": 770
|
| 633 |
},
|
| 634 |
{
|
| 635 |
-
"epoch": 1.
|
| 636 |
-
"grad_norm":
|
| 637 |
-
"learning_rate": 1.
|
| 638 |
-
"loss": 0.
|
| 639 |
"step": 780
|
| 640 |
},
|
| 641 |
{
|
| 642 |
-
"epoch": 1.
|
| 643 |
-
"grad_norm":
|
| 644 |
-
"learning_rate": 1.
|
| 645 |
-
"loss": 0.
|
| 646 |
"step": 790
|
| 647 |
},
|
| 648 |
{
|
| 649 |
-
"epoch": 1.
|
| 650 |
-
"grad_norm":
|
| 651 |
-
"learning_rate": 1.
|
| 652 |
-
"loss": 0.
|
| 653 |
"step": 800
|
| 654 |
},
|
| 655 |
{
|
| 656 |
-
"epoch": 1.
|
| 657 |
-
"eval_accuracy": 0.
|
| 658 |
-
"eval_accuracy_label_arts, culture, entertainment and media": 0.
|
| 659 |
-
"eval_accuracy_label_conflict, war and peace": 0.
|
| 660 |
-
"eval_accuracy_label_crime, law and justice": 0.
|
| 661 |
-
"eval_accuracy_label_disaster, accident, and emergency incident": 0.
|
| 662 |
-
"eval_accuracy_label_economy, business, and finance": 0.
|
| 663 |
-
"eval_accuracy_label_environment": 0.
|
| 664 |
-
"eval_accuracy_label_health": 0.
|
| 665 |
-
"eval_accuracy_label_human interest": 0.
|
| 666 |
"eval_accuracy_label_labour": 1.0,
|
| 667 |
-
"eval_accuracy_label_lifestyle and leisure": 0.
|
| 668 |
-
"eval_accuracy_label_politics": 0.
|
| 669 |
"eval_accuracy_label_religion": 0.0,
|
| 670 |
-
"eval_accuracy_label_science and technology": 0.
|
| 671 |
-
"eval_accuracy_label_society": 0.
|
| 672 |
"eval_accuracy_label_sport": 0.9615384615384616,
|
| 673 |
"eval_accuracy_label_weather": 1.0,
|
| 674 |
-
"eval_f1": 0.
|
| 675 |
-
"eval_loss": 0.
|
| 676 |
-
"eval_precision": 0.
|
| 677 |
-
"eval_recall": 0.
|
| 678 |
-
"eval_runtime":
|
| 679 |
-
"eval_samples_per_second":
|
| 680 |
-
"eval_steps_per_second": 7.
|
| 681 |
"step": 800
|
| 682 |
},
|
| 683 |
{
|
| 684 |
-
"epoch": 1.
|
| 685 |
-
"grad_norm": 12.
|
| 686 |
-
"learning_rate": 1.
|
| 687 |
-
"loss": 0.
|
| 688 |
"step": 810
|
| 689 |
},
|
| 690 |
{
|
| 691 |
-
"epoch": 1.
|
| 692 |
-
"grad_norm":
|
| 693 |
-
"learning_rate": 1.
|
| 694 |
-
"loss": 0.
|
| 695 |
"step": 820
|
| 696 |
},
|
| 697 |
{
|
| 698 |
-
"epoch": 1.
|
| 699 |
-
"grad_norm":
|
| 700 |
-
"learning_rate": 1.
|
| 701 |
-
"loss": 0.
|
| 702 |
"step": 830
|
| 703 |
},
|
| 704 |
{
|
| 705 |
-
"epoch": 1.
|
| 706 |
-
"grad_norm":
|
| 707 |
-
"learning_rate": 1.
|
| 708 |
-
"loss": 0.
|
| 709 |
"step": 840
|
| 710 |
},
|
| 711 |
{
|
| 712 |
-
"epoch": 1.
|
| 713 |
-
"grad_norm":
|
| 714 |
-
"learning_rate": 1.
|
| 715 |
-
"loss": 0.
|
| 716 |
"step": 850
|
| 717 |
},
|
| 718 |
{
|
| 719 |
-
"epoch": 1.
|
| 720 |
-
"grad_norm": 7.
|
| 721 |
-
"learning_rate": 1.
|
| 722 |
-
"loss": 0.
|
| 723 |
"step": 860
|
| 724 |
},
|
| 725 |
{
|
| 726 |
-
"epoch": 1.
|
| 727 |
-
"grad_norm":
|
| 728 |
-
"learning_rate": 1.
|
| 729 |
-
"loss": 0.
|
| 730 |
"step": 870
|
| 731 |
},
|
| 732 |
{
|
| 733 |
-
"epoch": 1.
|
| 734 |
-
"grad_norm":
|
| 735 |
-
"learning_rate": 1.
|
| 736 |
-
"loss": 0.
|
| 737 |
"step": 880
|
| 738 |
},
|
| 739 |
{
|
| 740 |
-
"epoch": 1.
|
| 741 |
-
"grad_norm":
|
| 742 |
-
"learning_rate": 1.
|
| 743 |
-
"loss": 0.
|
| 744 |
"step": 890
|
| 745 |
},
|
| 746 |
{
|
| 747 |
-
"epoch": 1.
|
| 748 |
-
"grad_norm":
|
| 749 |
-
"learning_rate": 1.
|
| 750 |
-
"loss": 0.
|
| 751 |
"step": 900
|
| 752 |
},
|
| 753 |
{
|
| 754 |
-
"epoch": 1.
|
| 755 |
-
"grad_norm":
|
| 756 |
-
"learning_rate": 1.
|
| 757 |
-
"loss": 0.
|
| 758 |
"step": 910
|
| 759 |
},
|
| 760 |
{
|
| 761 |
-
"epoch": 1.
|
| 762 |
-
"grad_norm":
|
| 763 |
-
"learning_rate": 1.
|
| 764 |
-
"loss": 0.
|
| 765 |
"step": 920
|
| 766 |
},
|
| 767 |
{
|
| 768 |
-
"epoch": 1.
|
| 769 |
-
"grad_norm":
|
| 770 |
-
"learning_rate": 1.
|
| 771 |
-
"loss": 0.
|
| 772 |
"step": 930
|
| 773 |
},
|
| 774 |
{
|
| 775 |
-
"epoch": 1.
|
| 776 |
-
"grad_norm": 9.
|
| 777 |
-
"learning_rate": 1.
|
| 778 |
-
"loss": 0.
|
| 779 |
"step": 940
|
| 780 |
},
|
| 781 |
{
|
| 782 |
-
"epoch": 1.
|
| 783 |
-
"grad_norm":
|
| 784 |
-
"learning_rate": 1.
|
| 785 |
-
"loss": 0.
|
| 786 |
"step": 950
|
| 787 |
},
|
| 788 |
{
|
| 789 |
-
"epoch": 1.
|
| 790 |
-
"grad_norm":
|
| 791 |
-
"learning_rate": 1.
|
| 792 |
-
"loss": 0.
|
| 793 |
"step": 960
|
| 794 |
},
|
| 795 |
{
|
| 796 |
-
"epoch": 1.
|
| 797 |
-
"grad_norm": 8.
|
| 798 |
-
"learning_rate": 1.
|
| 799 |
-
"loss": 0.
|
| 800 |
"step": 970
|
| 801 |
},
|
| 802 |
{
|
| 803 |
-
"epoch": 1.
|
| 804 |
-
"grad_norm":
|
| 805 |
-
"learning_rate": 1.
|
| 806 |
-
"loss": 0.
|
| 807 |
"step": 980
|
| 808 |
},
|
| 809 |
{
|
| 810 |
-
"epoch": 1.
|
| 811 |
-
"grad_norm":
|
| 812 |
-
"learning_rate": 1.
|
| 813 |
-
"loss": 0.
|
| 814 |
"step": 990
|
| 815 |
},
|
| 816 |
{
|
| 817 |
-
"epoch": 1.
|
| 818 |
-
"grad_norm":
|
| 819 |
-
"learning_rate": 1.
|
| 820 |
-
"loss": 0.
|
| 821 |
"step": 1000
|
| 822 |
},
|
| 823 |
{
|
| 824 |
-
"epoch": 1.
|
| 825 |
-
"eval_accuracy": 0.
|
| 826 |
-
"eval_accuracy_label_arts, culture, entertainment and media": 0.
|
| 827 |
-
"eval_accuracy_label_conflict, war and peace": 0.
|
| 828 |
-
"eval_accuracy_label_crime, law and justice": 0.
|
| 829 |
-
"eval_accuracy_label_disaster, accident, and emergency incident": 0.
|
| 830 |
-
"eval_accuracy_label_economy, business, and finance": 0.
|
| 831 |
-
"eval_accuracy_label_environment": 0.
|
| 832 |
-
"eval_accuracy_label_health": 0.
|
| 833 |
-
"eval_accuracy_label_human interest": 0.
|
| 834 |
-
"eval_accuracy_label_labour":
|
| 835 |
-
"eval_accuracy_label_lifestyle and leisure": 0.
|
| 836 |
-
"eval_accuracy_label_politics": 0.
|
| 837 |
"eval_accuracy_label_religion": 0.0,
|
| 838 |
-
"eval_accuracy_label_science and technology": 0.
|
| 839 |
-
"eval_accuracy_label_society": 0.
|
| 840 |
"eval_accuracy_label_sport": 0.9615384615384616,
|
| 841 |
-
"eval_accuracy_label_weather":
|
| 842 |
-
"eval_f1": 0.
|
| 843 |
-
"eval_loss": 0.
|
| 844 |
-
"eval_precision": 0.
|
| 845 |
-
"eval_recall": 0.
|
| 846 |
-
"eval_runtime":
|
| 847 |
-
"eval_samples_per_second":
|
| 848 |
-
"eval_steps_per_second": 7.
|
| 849 |
"step": 1000
|
| 850 |
},
|
| 851 |
{
|
| 852 |
-
"epoch": 1.
|
| 853 |
-
"grad_norm":
|
| 854 |
-
"learning_rate": 1.
|
| 855 |
-
"loss": 0.
|
| 856 |
"step": 1010
|
| 857 |
},
|
| 858 |
{
|
| 859 |
-
"epoch": 1.
|
| 860 |
-
"grad_norm":
|
| 861 |
-
"learning_rate": 1.
|
| 862 |
-
"loss": 0.
|
| 863 |
"step": 1020
|
| 864 |
},
|
| 865 |
{
|
| 866 |
-
"epoch": 1.
|
| 867 |
-
"grad_norm":
|
| 868 |
-
"learning_rate": 1.
|
| 869 |
-
"loss": 0.
|
| 870 |
"step": 1030
|
| 871 |
},
|
| 872 |
{
|
| 873 |
-
"epoch": 1.
|
| 874 |
-
"grad_norm":
|
| 875 |
-
"learning_rate": 1.
|
| 876 |
-
"loss": 0.
|
| 877 |
"step": 1040
|
| 878 |
},
|
| 879 |
{
|
| 880 |
-
"epoch": 1.
|
| 881 |
-
"grad_norm":
|
| 882 |
-
"learning_rate": 1.
|
| 883 |
-
"loss": 0.
|
| 884 |
"step": 1050
|
| 885 |
},
|
| 886 |
{
|
| 887 |
-
"epoch": 1.
|
| 888 |
-
"grad_norm":
|
| 889 |
-
"learning_rate": 1.
|
| 890 |
-
"loss": 0.
|
| 891 |
"step": 1060
|
| 892 |
},
|
| 893 |
{
|
| 894 |
-
"epoch": 1.
|
| 895 |
-
"grad_norm":
|
| 896 |
-
"learning_rate": 1.
|
| 897 |
-
"loss": 0.
|
| 898 |
"step": 1070
|
| 899 |
},
|
| 900 |
{
|
| 901 |
-
"epoch": 1.
|
| 902 |
-
"grad_norm":
|
| 903 |
-
"learning_rate": 1.
|
| 904 |
-
"loss": 0.
|
| 905 |
"step": 1080
|
| 906 |
},
|
| 907 |
{
|
| 908 |
-
"epoch": 1.
|
| 909 |
-
"grad_norm":
|
| 910 |
-
"learning_rate": 1.
|
| 911 |
-
"loss": 0.
|
| 912 |
"step": 1090
|
| 913 |
},
|
| 914 |
{
|
| 915 |
-
"epoch": 1.
|
| 916 |
-
"grad_norm": 10.
|
| 917 |
-
"learning_rate": 1.
|
| 918 |
-
"loss": 0.
|
| 919 |
"step": 1100
|
| 920 |
},
|
| 921 |
{
|
| 922 |
-
"epoch": 1.
|
| 923 |
-
"grad_norm":
|
| 924 |
-
"learning_rate": 1.
|
| 925 |
-
"loss": 0.
|
| 926 |
"step": 1110
|
| 927 |
},
|
| 928 |
{
|
| 929 |
-
"epoch": 1.
|
| 930 |
-
"grad_norm":
|
| 931 |
-
"learning_rate": 1.
|
| 932 |
-
"loss": 0.
|
| 933 |
"step": 1120
|
| 934 |
},
|
| 935 |
{
|
| 936 |
-
"epoch": 1.
|
| 937 |
-
"grad_norm":
|
| 938 |
-
"learning_rate": 1.
|
| 939 |
-
"loss": 0.
|
| 940 |
"step": 1130
|
| 941 |
},
|
| 942 |
{
|
| 943 |
-
"epoch": 1.
|
| 944 |
-
"grad_norm":
|
| 945 |
-
"learning_rate":
|
| 946 |
-
"loss": 0.
|
| 947 |
"step": 1140
|
| 948 |
},
|
| 949 |
{
|
| 950 |
-
"epoch": 1.
|
| 951 |
-
"grad_norm":
|
| 952 |
-
"learning_rate":
|
| 953 |
-
"loss": 0.
|
| 954 |
"step": 1150
|
| 955 |
},
|
| 956 |
{
|
| 957 |
-
"epoch": 1.
|
| 958 |
-
"grad_norm":
|
| 959 |
-
"learning_rate":
|
| 960 |
-
"loss": 0.
|
| 961 |
"step": 1160
|
| 962 |
},
|
| 963 |
{
|
| 964 |
-
"epoch": 1.
|
| 965 |
-
"grad_norm":
|
| 966 |
-
"learning_rate":
|
| 967 |
-
"loss": 0.
|
| 968 |
"step": 1170
|
| 969 |
},
|
| 970 |
{
|
| 971 |
-
"epoch": 1.
|
| 972 |
-
"grad_norm":
|
| 973 |
-
"learning_rate":
|
| 974 |
-
"loss": 0.
|
| 975 |
"step": 1180
|
| 976 |
},
|
| 977 |
{
|
| 978 |
-
"epoch":
|
| 979 |
-
"grad_norm":
|
| 980 |
-
"learning_rate":
|
| 981 |
-
"loss": 0.
|
| 982 |
"step": 1190
|
| 983 |
},
|
| 984 |
{
|
| 985 |
-
"epoch":
|
| 986 |
-
"grad_norm":
|
| 987 |
-
"learning_rate":
|
| 988 |
-
"loss": 0.
|
| 989 |
"step": 1200
|
| 990 |
},
|
| 991 |
{
|
| 992 |
-
"epoch":
|
| 993 |
-
"eval_accuracy": 0.
|
| 994 |
-
"eval_accuracy_label_arts, culture, entertainment and media": 0.
|
| 995 |
-
"eval_accuracy_label_conflict, war and peace": 0.
|
| 996 |
-
"eval_accuracy_label_crime, law and justice": 0.
|
| 997 |
-
"eval_accuracy_label_disaster, accident, and emergency incident": 0.
|
| 998 |
-
"eval_accuracy_label_economy, business, and finance": 0.
|
| 999 |
-
"eval_accuracy_label_environment": 0.
|
| 1000 |
-
"eval_accuracy_label_health": 0.
|
| 1001 |
-
"eval_accuracy_label_human interest": 0.
|
| 1002 |
"eval_accuracy_label_labour": 0.5,
|
| 1003 |
-
"eval_accuracy_label_lifestyle and leisure": 0.
|
| 1004 |
-
"eval_accuracy_label_politics": 0.
|
| 1005 |
"eval_accuracy_label_religion": 0.0,
|
| 1006 |
"eval_accuracy_label_science and technology": 0.4166666666666667,
|
| 1007 |
-
"eval_accuracy_label_society": 0.
|
| 1008 |
"eval_accuracy_label_sport": 0.9615384615384616,
|
| 1009 |
"eval_accuracy_label_weather": 1.0,
|
| 1010 |
-
"eval_f1": 0.
|
| 1011 |
-
"eval_loss": 0.
|
| 1012 |
-
"eval_precision": 0.
|
| 1013 |
-
"eval_recall": 0.
|
| 1014 |
-
"eval_runtime":
|
| 1015 |
-
"eval_samples_per_second":
|
| 1016 |
-
"eval_steps_per_second": 7.
|
| 1017 |
"step": 1200
|
| 1018 |
},
|
| 1019 |
{
|
| 1020 |
-
"epoch":
|
| 1021 |
-
"grad_norm":
|
| 1022 |
-
"learning_rate":
|
| 1023 |
-
"loss": 0.
|
| 1024 |
"step": 1210
|
| 1025 |
},
|
| 1026 |
{
|
| 1027 |
-
"epoch":
|
| 1028 |
-
"grad_norm":
|
| 1029 |
-
"learning_rate":
|
| 1030 |
-
"loss": 0.
|
| 1031 |
"step": 1220
|
| 1032 |
},
|
| 1033 |
{
|
| 1034 |
-
"epoch":
|
| 1035 |
-
"grad_norm":
|
| 1036 |
-
"learning_rate":
|
| 1037 |
-
"loss": 0.
|
| 1038 |
"step": 1230
|
| 1039 |
},
|
| 1040 |
{
|
| 1041 |
-
"epoch":
|
| 1042 |
-
"grad_norm":
|
| 1043 |
-
"learning_rate":
|
| 1044 |
-
"loss": 0.
|
| 1045 |
"step": 1240
|
| 1046 |
},
|
| 1047 |
{
|
| 1048 |
-
"epoch":
|
| 1049 |
-
"grad_norm":
|
| 1050 |
-
"learning_rate":
|
| 1051 |
-
"loss": 0.
|
| 1052 |
"step": 1250
|
| 1053 |
},
|
| 1054 |
{
|
| 1055 |
-
"epoch":
|
| 1056 |
-
"grad_norm":
|
| 1057 |
-
"learning_rate":
|
| 1058 |
-
"loss": 0.
|
| 1059 |
"step": 1260
|
| 1060 |
},
|
| 1061 |
{
|
| 1062 |
-
"epoch":
|
| 1063 |
-
"grad_norm":
|
| 1064 |
-
"learning_rate":
|
| 1065 |
-
"loss": 0.
|
| 1066 |
"step": 1270
|
| 1067 |
},
|
| 1068 |
{
|
| 1069 |
-
"epoch":
|
| 1070 |
-
"grad_norm":
|
| 1071 |
-
"learning_rate":
|
| 1072 |
-
"loss": 0.
|
| 1073 |
"step": 1280
|
| 1074 |
},
|
| 1075 |
{
|
| 1076 |
-
"epoch":
|
| 1077 |
-
"grad_norm":
|
| 1078 |
-
"learning_rate":
|
| 1079 |
-
"loss": 0.
|
| 1080 |
"step": 1290
|
| 1081 |
},
|
| 1082 |
{
|
| 1083 |
-
"epoch":
|
| 1084 |
-
"grad_norm":
|
| 1085 |
-
"learning_rate":
|
| 1086 |
-
"loss": 0.
|
| 1087 |
"step": 1300
|
| 1088 |
},
|
| 1089 |
{
|
| 1090 |
-
"epoch":
|
| 1091 |
-
"grad_norm":
|
| 1092 |
-
"learning_rate":
|
| 1093 |
-
"loss": 0.
|
| 1094 |
"step": 1310
|
| 1095 |
},
|
| 1096 |
{
|
| 1097 |
-
"epoch":
|
| 1098 |
-
"grad_norm":
|
| 1099 |
-
"learning_rate":
|
| 1100 |
-
"loss": 0.
|
| 1101 |
"step": 1320
|
| 1102 |
},
|
| 1103 |
{
|
| 1104 |
-
"epoch":
|
| 1105 |
-
"grad_norm":
|
| 1106 |
-
"learning_rate":
|
| 1107 |
-
"loss": 0.
|
| 1108 |
"step": 1330
|
| 1109 |
},
|
| 1110 |
{
|
| 1111 |
-
"epoch":
|
| 1112 |
-
"grad_norm":
|
| 1113 |
-
"learning_rate":
|
| 1114 |
-
"loss": 0.
|
| 1115 |
"step": 1340
|
| 1116 |
},
|
| 1117 |
{
|
| 1118 |
-
"epoch":
|
| 1119 |
-
"grad_norm":
|
| 1120 |
-
"learning_rate":
|
| 1121 |
-
"loss": 0.
|
| 1122 |
"step": 1350
|
| 1123 |
},
|
| 1124 |
{
|
| 1125 |
-
"epoch":
|
| 1126 |
-
"grad_norm":
|
| 1127 |
-
"learning_rate":
|
| 1128 |
-
"loss": 0.
|
| 1129 |
"step": 1360
|
| 1130 |
},
|
| 1131 |
{
|
| 1132 |
-
"epoch":
|
| 1133 |
-
"grad_norm":
|
| 1134 |
-
"learning_rate":
|
| 1135 |
-
"loss": 0.
|
| 1136 |
"step": 1370
|
| 1137 |
},
|
| 1138 |
{
|
| 1139 |
-
"epoch": 2.
|
| 1140 |
-
"grad_norm":
|
| 1141 |
-
"learning_rate":
|
| 1142 |
-
"loss": 0.
|
| 1143 |
"step": 1380
|
| 1144 |
},
|
| 1145 |
{
|
| 1146 |
-
"epoch": 2.
|
| 1147 |
-
"grad_norm":
|
| 1148 |
-
"learning_rate":
|
| 1149 |
-
"loss": 0.
|
| 1150 |
"step": 1390
|
| 1151 |
},
|
| 1152 |
{
|
| 1153 |
-
"epoch": 2.
|
| 1154 |
-
"grad_norm":
|
| 1155 |
-
"learning_rate":
|
| 1156 |
-
"loss": 0.
|
| 1157 |
"step": 1400
|
| 1158 |
},
|
| 1159 |
{
|
| 1160 |
-
"epoch": 2.
|
| 1161 |
-
"eval_accuracy": 0.
|
| 1162 |
-
"eval_accuracy_label_arts, culture, entertainment and media": 0.
|
| 1163 |
-
"eval_accuracy_label_conflict, war and peace": 0.
|
| 1164 |
-
"eval_accuracy_label_crime, law and justice": 0.
|
| 1165 |
-
"eval_accuracy_label_disaster, accident, and emergency incident": 0.
|
| 1166 |
-
"eval_accuracy_label_economy, business, and finance": 0.
|
| 1167 |
-
"eval_accuracy_label_environment": 0.
|
| 1168 |
-
"eval_accuracy_label_health": 0.
|
| 1169 |
-
"eval_accuracy_label_human interest": 0.
|
| 1170 |
"eval_accuracy_label_labour": 0.5,
|
| 1171 |
-
"eval_accuracy_label_lifestyle and leisure": 0.
|
| 1172 |
-
"eval_accuracy_label_politics": 0.
|
| 1173 |
"eval_accuracy_label_religion": 0.0,
|
| 1174 |
"eval_accuracy_label_science and technology": 0.4166666666666667,
|
| 1175 |
-
"eval_accuracy_label_society": 0.
|
| 1176 |
"eval_accuracy_label_sport": 0.9615384615384616,
|
| 1177 |
-
"eval_accuracy_label_weather":
|
| 1178 |
-
"eval_f1": 0.
|
| 1179 |
-
"eval_loss": 0.
|
| 1180 |
-
"eval_precision": 0.
|
| 1181 |
-
"eval_recall": 0.
|
| 1182 |
-
"eval_runtime":
|
| 1183 |
-
"eval_samples_per_second":
|
| 1184 |
-
"eval_steps_per_second": 7.
|
| 1185 |
"step": 1400
|
| 1186 |
},
|
| 1187 |
{
|
| 1188 |
-
"epoch": 2.
|
| 1189 |
-
"grad_norm":
|
| 1190 |
-
"learning_rate":
|
| 1191 |
-
"loss": 0.
|
| 1192 |
"step": 1410
|
| 1193 |
},
|
| 1194 |
{
|
| 1195 |
-
"epoch": 2.
|
| 1196 |
-
"grad_norm":
|
| 1197 |
-
"learning_rate":
|
| 1198 |
-
"loss": 0.
|
| 1199 |
"step": 1420
|
| 1200 |
},
|
| 1201 |
{
|
| 1202 |
-
"epoch": 2.
|
| 1203 |
-
"grad_norm":
|
| 1204 |
-
"learning_rate":
|
| 1205 |
-
"loss": 0.
|
| 1206 |
"step": 1430
|
| 1207 |
},
|
| 1208 |
{
|
| 1209 |
-
"epoch": 2.
|
| 1210 |
-
"grad_norm":
|
| 1211 |
-
"learning_rate":
|
| 1212 |
-
"loss": 0.
|
| 1213 |
"step": 1440
|
| 1214 |
},
|
| 1215 |
{
|
| 1216 |
-
"epoch": 2.
|
| 1217 |
-
"grad_norm":
|
| 1218 |
-
"learning_rate":
|
| 1219 |
-
"loss": 0.
|
| 1220 |
"step": 1450
|
| 1221 |
},
|
| 1222 |
{
|
| 1223 |
-
"epoch": 2.
|
| 1224 |
-
"grad_norm":
|
| 1225 |
-
"learning_rate":
|
| 1226 |
-
"loss": 0.
|
| 1227 |
"step": 1460
|
| 1228 |
},
|
| 1229 |
{
|
| 1230 |
-
"epoch": 2.
|
| 1231 |
-
"grad_norm":
|
| 1232 |
-
"learning_rate":
|
| 1233 |
-
"loss": 0.
|
| 1234 |
"step": 1470
|
| 1235 |
},
|
| 1236 |
{
|
| 1237 |
-
"epoch": 2.
|
| 1238 |
-
"grad_norm":
|
| 1239 |
-
"learning_rate":
|
| 1240 |
-
"loss": 0.
|
| 1241 |
"step": 1480
|
| 1242 |
},
|
| 1243 |
{
|
| 1244 |
-
"epoch": 2.
|
| 1245 |
-
"grad_norm":
|
| 1246 |
-
"learning_rate":
|
| 1247 |
-
"loss": 0.
|
| 1248 |
"step": 1490
|
| 1249 |
},
|
| 1250 |
{
|
| 1251 |
-
"epoch": 2.
|
| 1252 |
-
"grad_norm":
|
| 1253 |
-
"learning_rate":
|
| 1254 |
-
"loss": 0.
|
| 1255 |
"step": 1500
|
| 1256 |
},
|
| 1257 |
{
|
| 1258 |
-
"epoch": 2.
|
| 1259 |
-
"grad_norm":
|
| 1260 |
-
"learning_rate":
|
| 1261 |
-
"loss": 0.
|
| 1262 |
"step": 1510
|
| 1263 |
},
|
| 1264 |
{
|
| 1265 |
-
"epoch": 2.
|
| 1266 |
-
"grad_norm":
|
| 1267 |
-
"learning_rate":
|
| 1268 |
-
"loss": 0.
|
| 1269 |
"step": 1520
|
| 1270 |
},
|
| 1271 |
{
|
| 1272 |
-
"epoch": 2.
|
| 1273 |
-
"grad_norm":
|
| 1274 |
-
"learning_rate":
|
| 1275 |
-
"loss": 0.
|
| 1276 |
"step": 1530
|
| 1277 |
},
|
| 1278 |
{
|
| 1279 |
-
"epoch": 2.
|
| 1280 |
-
"grad_norm":
|
| 1281 |
-
"learning_rate":
|
| 1282 |
-
"loss": 0.
|
| 1283 |
"step": 1540
|
| 1284 |
},
|
| 1285 |
{
|
| 1286 |
-
"epoch": 2.
|
| 1287 |
-
"grad_norm":
|
| 1288 |
-
"learning_rate":
|
| 1289 |
-
"loss": 0.
|
| 1290 |
"step": 1550
|
| 1291 |
},
|
| 1292 |
{
|
| 1293 |
-
"epoch": 2.
|
| 1294 |
-
"grad_norm":
|
| 1295 |
-
"learning_rate":
|
| 1296 |
-
"loss": 0.
|
| 1297 |
"step": 1560
|
| 1298 |
},
|
| 1299 |
{
|
| 1300 |
-
"epoch": 2.
|
| 1301 |
-
"grad_norm":
|
| 1302 |
-
"learning_rate":
|
| 1303 |
-
"loss": 0.
|
| 1304 |
"step": 1570
|
| 1305 |
},
|
| 1306 |
{
|
| 1307 |
-
"epoch": 2.
|
| 1308 |
-
"grad_norm":
|
| 1309 |
-
"learning_rate":
|
| 1310 |
-
"loss": 0.
|
| 1311 |
"step": 1580
|
| 1312 |
},
|
| 1313 |
{
|
| 1314 |
-
"epoch": 2.
|
| 1315 |
-
"grad_norm":
|
| 1316 |
-
"learning_rate":
|
| 1317 |
-
"loss": 0.
|
| 1318 |
"step": 1590
|
| 1319 |
},
|
| 1320 |
{
|
| 1321 |
-
"epoch": 2.
|
| 1322 |
-
"grad_norm": 6.
|
| 1323 |
-
"learning_rate":
|
| 1324 |
-
"loss": 0.
|
| 1325 |
"step": 1600
|
| 1326 |
},
|
| 1327 |
{
|
| 1328 |
-
"epoch": 2.
|
| 1329 |
-
"eval_accuracy": 0.
|
| 1330 |
-
"eval_accuracy_label_arts, culture, entertainment and media": 0.
|
| 1331 |
-
"eval_accuracy_label_conflict, war and peace": 0.
|
| 1332 |
-
"eval_accuracy_label_crime, law and justice": 0.
|
| 1333 |
-
"eval_accuracy_label_disaster, accident, and emergency incident": 0.
|
| 1334 |
-
"eval_accuracy_label_economy, business, and finance": 0.
|
| 1335 |
-
"eval_accuracy_label_environment": 0.
|
| 1336 |
-
"eval_accuracy_label_health": 0.
|
| 1337 |
-
"eval_accuracy_label_human interest": 0.
|
| 1338 |
"eval_accuracy_label_labour": 0.5,
|
| 1339 |
-
"eval_accuracy_label_lifestyle and leisure": 0.
|
| 1340 |
-
"eval_accuracy_label_politics": 0.
|
| 1341 |
"eval_accuracy_label_religion": 0.0,
|
| 1342 |
"eval_accuracy_label_science and technology": 0.4166666666666667,
|
| 1343 |
-
"eval_accuracy_label_society": 0.
|
| 1344 |
"eval_accuracy_label_sport": 0.9615384615384616,
|
| 1345 |
-
"eval_accuracy_label_weather":
|
| 1346 |
-
"eval_f1": 0.
|
| 1347 |
-
"eval_loss": 0.
|
| 1348 |
-
"eval_precision": 0.
|
| 1349 |
-
"eval_recall": 0.
|
| 1350 |
-
"eval_runtime":
|
| 1351 |
-
"eval_samples_per_second":
|
| 1352 |
-
"eval_steps_per_second": 7.
|
| 1353 |
"step": 1600
|
| 1354 |
},
|
| 1355 |
{
|
| 1356 |
-
"epoch": 2.
|
| 1357 |
-
"grad_norm":
|
| 1358 |
-
"learning_rate":
|
| 1359 |
-
"loss": 0.
|
| 1360 |
"step": 1610
|
| 1361 |
},
|
| 1362 |
{
|
| 1363 |
-
"epoch": 2.
|
| 1364 |
-
"grad_norm":
|
| 1365 |
-
"learning_rate":
|
| 1366 |
-
"loss": 0.
|
| 1367 |
"step": 1620
|
| 1368 |
},
|
| 1369 |
{
|
| 1370 |
-
"epoch": 2.
|
| 1371 |
-
"grad_norm":
|
| 1372 |
-
"learning_rate":
|
| 1373 |
-
"loss": 0.
|
| 1374 |
"step": 1630
|
| 1375 |
},
|
| 1376 |
{
|
| 1377 |
-
"epoch": 2.
|
| 1378 |
-
"grad_norm":
|
| 1379 |
-
"learning_rate":
|
| 1380 |
-
"loss": 0.
|
| 1381 |
"step": 1640
|
| 1382 |
},
|
| 1383 |
{
|
| 1384 |
-
"epoch": 2.
|
| 1385 |
-
"grad_norm":
|
| 1386 |
-
"learning_rate":
|
| 1387 |
-
"loss": 0.
|
| 1388 |
"step": 1650
|
| 1389 |
},
|
| 1390 |
{
|
| 1391 |
-
"epoch": 2.
|
| 1392 |
-
"grad_norm":
|
| 1393 |
-
"learning_rate":
|
| 1394 |
-
"loss": 0.
|
| 1395 |
"step": 1660
|
| 1396 |
},
|
| 1397 |
{
|
| 1398 |
-
"epoch": 2.
|
| 1399 |
-
"grad_norm":
|
| 1400 |
-
"learning_rate":
|
| 1401 |
-
"loss": 0.
|
| 1402 |
"step": 1670
|
| 1403 |
},
|
| 1404 |
{
|
| 1405 |
-
"epoch": 2.
|
| 1406 |
-
"grad_norm":
|
| 1407 |
-
"learning_rate":
|
| 1408 |
-
"loss": 0.
|
| 1409 |
"step": 1680
|
| 1410 |
},
|
| 1411 |
{
|
| 1412 |
-
"epoch": 2.
|
| 1413 |
-
"grad_norm":
|
| 1414 |
-
"learning_rate":
|
| 1415 |
-
"loss": 0.
|
| 1416 |
"step": 1690
|
| 1417 |
},
|
| 1418 |
{
|
| 1419 |
-
"epoch": 2.
|
| 1420 |
-
"grad_norm":
|
| 1421 |
-
"learning_rate":
|
| 1422 |
-
"loss": 0.
|
| 1423 |
"step": 1700
|
| 1424 |
},
|
| 1425 |
{
|
| 1426 |
-
"epoch": 2.
|
| 1427 |
-
"grad_norm": 8.
|
| 1428 |
-
"learning_rate":
|
| 1429 |
-
"loss": 0.
|
| 1430 |
"step": 1710
|
| 1431 |
},
|
| 1432 |
{
|
| 1433 |
-
"epoch": 2.
|
| 1434 |
-
"grad_norm":
|
| 1435 |
-
"learning_rate":
|
| 1436 |
-
"loss": 0.
|
| 1437 |
"step": 1720
|
| 1438 |
},
|
| 1439 |
{
|
| 1440 |
-
"epoch": 2.
|
| 1441 |
-
"grad_norm":
|
| 1442 |
-
"learning_rate":
|
| 1443 |
-
"loss": 0.
|
| 1444 |
"step": 1730
|
| 1445 |
},
|
| 1446 |
{
|
| 1447 |
-
"epoch": 2.
|
| 1448 |
-
"grad_norm":
|
| 1449 |
-
"learning_rate":
|
| 1450 |
-
"loss": 0.
|
| 1451 |
"step": 1740
|
| 1452 |
},
|
| 1453 |
{
|
| 1454 |
-
"epoch": 2.
|
| 1455 |
-
"grad_norm":
|
| 1456 |
-
"learning_rate": 4.
|
| 1457 |
-
"loss": 0.
|
| 1458 |
"step": 1750
|
| 1459 |
},
|
| 1460 |
{
|
| 1461 |
-
"epoch": 2.
|
| 1462 |
-
"grad_norm":
|
| 1463 |
-
"learning_rate":
|
| 1464 |
-
"loss": 0.
|
| 1465 |
"step": 1760
|
| 1466 |
},
|
| 1467 |
{
|
| 1468 |
-
"epoch": 2.
|
| 1469 |
-
"grad_norm":
|
| 1470 |
-
"learning_rate":
|
| 1471 |
-
"loss": 0.
|
| 1472 |
"step": 1770
|
| 1473 |
},
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1474 |
{
|
| 1475 |
"epoch": 3.0,
|
| 1476 |
-
"step":
|
| 1477 |
-
"total_flos":
|
| 1478 |
-
"train_loss": 0.
|
| 1479 |
-
"train_runtime":
|
| 1480 |
-
"train_samples_per_second":
|
| 1481 |
-
"train_steps_per_second": 1.
|
| 1482 |
},
|
| 1483 |
{
|
| 1484 |
"epoch": 3.0,
|
| 1485 |
-
"eval_accuracy": 0.
|
| 1486 |
-
"eval_accuracy_label_arts, culture, entertainment and media": 0.
|
| 1487 |
-
"eval_accuracy_label_conflict, war and peace": 0.
|
| 1488 |
-
"eval_accuracy_label_crime, law and justice": 0.
|
| 1489 |
-
"eval_accuracy_label_disaster, accident, and emergency incident": 0.
|
| 1490 |
-
"eval_accuracy_label_economy, business, and finance": 0.
|
| 1491 |
-
"eval_accuracy_label_environment": 0.
|
| 1492 |
-
"eval_accuracy_label_health": 0.
|
| 1493 |
-
"eval_accuracy_label_human interest": 0.
|
| 1494 |
"eval_accuracy_label_labour": 0.5,
|
| 1495 |
-
"eval_accuracy_label_lifestyle and leisure": 0.
|
| 1496 |
-
"eval_accuracy_label_politics": 0.
|
| 1497 |
"eval_accuracy_label_religion": 0.0,
|
| 1498 |
-
"eval_accuracy_label_science and technology": 0.
|
| 1499 |
-
"eval_accuracy_label_society": 0.
|
| 1500 |
"eval_accuracy_label_sport": 0.9615384615384616,
|
| 1501 |
-
"eval_accuracy_label_weather":
|
| 1502 |
-
"eval_f1": 0.
|
| 1503 |
-
"eval_loss": 0.
|
| 1504 |
-
"eval_precision": 0.
|
| 1505 |
-
"eval_recall": 0.
|
| 1506 |
-
"eval_runtime":
|
| 1507 |
-
"eval_samples_per_second": 127.
|
| 1508 |
-
"eval_steps_per_second":
|
| 1509 |
-
"step":
|
| 1510 |
}
|
| 1511 |
],
|
| 1512 |
"logging_steps": 10,
|
| 1513 |
-
"max_steps":
|
| 1514 |
"num_input_tokens_seen": 0,
|
| 1515 |
"num_train_epochs": 3,
|
| 1516 |
"save_steps": 1000,
|
| 1517 |
-
"total_flos":
|
| 1518 |
"train_batch_size": 16,
|
| 1519 |
"trial_name": null,
|
| 1520 |
"trial_params": null
|
|
|
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
"epoch": 3.0,
|
| 5 |
"eval_steps": 200,
|
| 6 |
+
"global_step": 2064,
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
| 10 |
"log_history": [
|
| 11 |
{
|
| 12 |
+
"epoch": 0.014534883720930232,
|
| 13 |
+
"grad_norm": 5.562132358551025,
|
| 14 |
"learning_rate": 4.0000000000000003e-07,
|
| 15 |
+
"loss": 2.7844,
|
| 16 |
"step": 10
|
| 17 |
},
|
| 18 |
{
|
| 19 |
+
"epoch": 0.029069767441860465,
|
| 20 |
+
"grad_norm": 6.067043781280518,
|
| 21 |
"learning_rate": 8.000000000000001e-07,
|
| 22 |
+
"loss": 2.7725,
|
| 23 |
"step": 20
|
| 24 |
},
|
| 25 |
{
|
| 26 |
+
"epoch": 0.0436046511627907,
|
| 27 |
+
"grad_norm": 6.281836986541748,
|
| 28 |
"learning_rate": 1.2000000000000002e-06,
|
| 29 |
+
"loss": 2.7484,
|
| 30 |
"step": 30
|
| 31 |
},
|
| 32 |
{
|
| 33 |
+
"epoch": 0.05813953488372093,
|
| 34 |
+
"grad_norm": 5.292967319488525,
|
| 35 |
"learning_rate": 1.6000000000000001e-06,
|
| 36 |
+
"loss": 2.6898,
|
| 37 |
"step": 40
|
| 38 |
},
|
| 39 |
{
|
| 40 |
+
"epoch": 0.07267441860465117,
|
| 41 |
+
"grad_norm": 6.3012261390686035,
|
| 42 |
"learning_rate": 2.0000000000000003e-06,
|
| 43 |
+
"loss": 2.6246,
|
| 44 |
"step": 50
|
| 45 |
},
|
| 46 |
{
|
| 47 |
+
"epoch": 0.0872093023255814,
|
| 48 |
+
"grad_norm": 5.181797981262207,
|
| 49 |
"learning_rate": 2.4000000000000003e-06,
|
| 50 |
+
"loss": 2.593,
|
| 51 |
"step": 60
|
| 52 |
},
|
| 53 |
{
|
| 54 |
+
"epoch": 0.10174418604651163,
|
| 55 |
+
"grad_norm": 6.388678073883057,
|
| 56 |
"learning_rate": 2.8000000000000003e-06,
|
| 57 |
+
"loss": 2.5312,
|
| 58 |
"step": 70
|
| 59 |
},
|
| 60 |
{
|
| 61 |
+
"epoch": 0.11627906976744186,
|
| 62 |
+
"grad_norm": 5.643568992614746,
|
| 63 |
"learning_rate": 3.2000000000000003e-06,
|
| 64 |
+
"loss": 2.4826,
|
| 65 |
"step": 80
|
| 66 |
},
|
| 67 |
{
|
| 68 |
+
"epoch": 0.1308139534883721,
|
| 69 |
+
"grad_norm": 5.568506240844727,
|
| 70 |
"learning_rate": 3.6000000000000003e-06,
|
| 71 |
+
"loss": 2.4131,
|
| 72 |
"step": 90
|
| 73 |
},
|
| 74 |
{
|
| 75 |
+
"epoch": 0.14534883720930233,
|
| 76 |
+
"grad_norm": 5.19083833694458,
|
| 77 |
"learning_rate": 4.000000000000001e-06,
|
| 78 |
+
"loss": 2.415,
|
| 79 |
"step": 100
|
| 80 |
},
|
| 81 |
{
|
| 82 |
+
"epoch": 0.15988372093023256,
|
| 83 |
+
"grad_norm": 5.537081241607666,
|
| 84 |
"learning_rate": 4.4e-06,
|
| 85 |
+
"loss": 2.3401,
|
| 86 |
"step": 110
|
| 87 |
},
|
| 88 |
{
|
| 89 |
+
"epoch": 0.1744186046511628,
|
| 90 |
+
"grad_norm": 5.086550235748291,
|
| 91 |
"learning_rate": 4.800000000000001e-06,
|
| 92 |
+
"loss": 2.3943,
|
| 93 |
"step": 120
|
| 94 |
},
|
| 95 |
{
|
| 96 |
+
"epoch": 0.18895348837209303,
|
| 97 |
+
"grad_norm": 7.545917987823486,
|
| 98 |
"learning_rate": 5.2e-06,
|
| 99 |
+
"loss": 2.2582,
|
| 100 |
"step": 130
|
| 101 |
},
|
| 102 |
{
|
| 103 |
+
"epoch": 0.20348837209302326,
|
| 104 |
+
"grad_norm": 7.227110862731934,
|
| 105 |
"learning_rate": 5.600000000000001e-06,
|
| 106 |
+
"loss": 2.1587,
|
| 107 |
"step": 140
|
| 108 |
},
|
| 109 |
{
|
| 110 |
+
"epoch": 0.2180232558139535,
|
| 111 |
+
"grad_norm": 6.003934860229492,
|
| 112 |
"learning_rate": 6e-06,
|
| 113 |
+
"loss": 2.0893,
|
| 114 |
"step": 150
|
| 115 |
},
|
| 116 |
{
|
| 117 |
+
"epoch": 0.23255813953488372,
|
| 118 |
+
"grad_norm": 6.239010810852051,
|
| 119 |
"learning_rate": 6.4000000000000006e-06,
|
| 120 |
+
"loss": 2.0743,
|
| 121 |
"step": 160
|
| 122 |
},
|
| 123 |
{
|
| 124 |
+
"epoch": 0.24709302325581395,
|
| 125 |
+
"grad_norm": 7.215450763702393,
|
| 126 |
"learning_rate": 6.800000000000001e-06,
|
| 127 |
+
"loss": 2.0095,
|
| 128 |
"step": 170
|
| 129 |
},
|
| 130 |
{
|
| 131 |
+
"epoch": 0.2616279069767442,
|
| 132 |
+
"grad_norm": 5.769580364227295,
|
| 133 |
"learning_rate": 7.2000000000000005e-06,
|
| 134 |
+
"loss": 2.0063,
|
| 135 |
"step": 180
|
| 136 |
},
|
| 137 |
{
|
| 138 |
+
"epoch": 0.2761627906976744,
|
| 139 |
+
"grad_norm": 7.701093673706055,
|
| 140 |
"learning_rate": 7.600000000000001e-06,
|
| 141 |
+
"loss": 1.9538,
|
| 142 |
"step": 190
|
| 143 |
},
|
| 144 |
{
|
| 145 |
+
"epoch": 0.29069767441860467,
|
| 146 |
+
"grad_norm": 8.804961204528809,
|
| 147 |
"learning_rate": 8.000000000000001e-06,
|
| 148 |
+
"loss": 1.9761,
|
| 149 |
"step": 200
|
| 150 |
},
|
| 151 |
{
|
| 152 |
+
"epoch": 0.29069767441860467,
|
| 153 |
+
"eval_accuracy": 0.64625,
|
| 154 |
+
"eval_accuracy_label_arts, culture, entertainment and media": 0.3157894736842105,
|
| 155 |
+
"eval_accuracy_label_conflict, war and peace": 0.8315467075038285,
|
| 156 |
+
"eval_accuracy_label_crime, law and justice": 0.7628865979381443,
|
| 157 |
+
"eval_accuracy_label_disaster, accident, and emergency incident": 0.7054794520547946,
|
| 158 |
+
"eval_accuracy_label_economy, business, and finance": 0.5436893203883495,
|
| 159 |
+
"eval_accuracy_label_environment": 0.0,
|
| 160 |
+
"eval_accuracy_label_health": 0.5,
|
| 161 |
"eval_accuracy_label_human interest": 0.0,
|
| 162 |
+
"eval_accuracy_label_labour": 0.0,
|
| 163 |
+
"eval_accuracy_label_lifestyle and leisure": 0.3333333333333333,
|
| 164 |
+
"eval_accuracy_label_politics": 0.4843205574912892,
|
| 165 |
"eval_accuracy_label_religion": 0.0,
|
| 166 |
+
"eval_accuracy_label_science and technology": 0.08333333333333333,
|
| 167 |
+
"eval_accuracy_label_society": 0.0,
|
| 168 |
"eval_accuracy_label_sport": 0.9615384615384616,
|
| 169 |
+
"eval_accuracy_label_weather": 0.0,
|
| 170 |
+
"eval_f1": 0.6164394441800294,
|
| 171 |
+
"eval_loss": 1.4046330451965332,
|
| 172 |
+
"eval_precision": 0.6057417123934041,
|
| 173 |
+
"eval_recall": 0.64625,
|
| 174 |
+
"eval_runtime": 12.4531,
|
| 175 |
+
"eval_samples_per_second": 128.482,
|
| 176 |
+
"eval_steps_per_second": 8.03,
|
| 177 |
"step": 200
|
| 178 |
},
|
| 179 |
{
|
| 180 |
+
"epoch": 0.30523255813953487,
|
| 181 |
+
"grad_norm": 7.173561096191406,
|
| 182 |
"learning_rate": 8.400000000000001e-06,
|
| 183 |
+
"loss": 1.7684,
|
| 184 |
"step": 210
|
| 185 |
},
|
| 186 |
{
|
| 187 |
+
"epoch": 0.31976744186046513,
|
| 188 |
+
"grad_norm": 8.086793899536133,
|
| 189 |
"learning_rate": 8.8e-06,
|
| 190 |
+
"loss": 1.7253,
|
| 191 |
"step": 220
|
| 192 |
},
|
| 193 |
{
|
| 194 |
+
"epoch": 0.33430232558139533,
|
| 195 |
+
"grad_norm": 8.559186935424805,
|
| 196 |
"learning_rate": 9.200000000000002e-06,
|
| 197 |
+
"loss": 1.795,
|
| 198 |
"step": 230
|
| 199 |
},
|
| 200 |
{
|
| 201 |
+
"epoch": 0.3488372093023256,
|
| 202 |
+
"grad_norm": 6.066705226898193,
|
| 203 |
"learning_rate": 9.600000000000001e-06,
|
| 204 |
+
"loss": 1.6181,
|
| 205 |
"step": 240
|
| 206 |
},
|
| 207 |
{
|
| 208 |
+
"epoch": 0.3633720930232558,
|
| 209 |
+
"grad_norm": 13.028596878051758,
|
| 210 |
"learning_rate": 1e-05,
|
| 211 |
+
"loss": 1.6328,
|
| 212 |
"step": 250
|
| 213 |
},
|
| 214 |
{
|
| 215 |
+
"epoch": 0.37790697674418605,
|
| 216 |
+
"grad_norm": 9.958759307861328,
|
| 217 |
"learning_rate": 1.04e-05,
|
| 218 |
+
"loss": 1.6997,
|
| 219 |
"step": 260
|
| 220 |
},
|
| 221 |
{
|
| 222 |
+
"epoch": 0.39244186046511625,
|
| 223 |
+
"grad_norm": 11.519359588623047,
|
| 224 |
"learning_rate": 1.0800000000000002e-05,
|
| 225 |
+
"loss": 1.6046,
|
| 226 |
"step": 270
|
| 227 |
},
|
| 228 |
{
|
| 229 |
+
"epoch": 0.4069767441860465,
|
| 230 |
+
"grad_norm": 8.298735618591309,
|
| 231 |
"learning_rate": 1.1200000000000001e-05,
|
| 232 |
+
"loss": 1.4126,
|
| 233 |
"step": 280
|
| 234 |
},
|
| 235 |
{
|
| 236 |
+
"epoch": 0.42151162790697677,
|
| 237 |
+
"grad_norm": 10.555624961853027,
|
| 238 |
"learning_rate": 1.16e-05,
|
| 239 |
+
"loss": 1.445,
|
| 240 |
"step": 290
|
| 241 |
},
|
| 242 |
{
|
| 243 |
+
"epoch": 0.436046511627907,
|
| 244 |
+
"grad_norm": 8.33277702331543,
|
| 245 |
"learning_rate": 1.2e-05,
|
| 246 |
+
"loss": 1.4088,
|
| 247 |
"step": 300
|
| 248 |
},
|
| 249 |
{
|
| 250 |
+
"epoch": 0.45058139534883723,
|
| 251 |
+
"grad_norm": 10.028512001037598,
|
| 252 |
"learning_rate": 1.2400000000000002e-05,
|
| 253 |
+
"loss": 1.3765,
|
| 254 |
"step": 310
|
| 255 |
},
|
| 256 |
{
|
| 257 |
+
"epoch": 0.46511627906976744,
|
| 258 |
+
"grad_norm": 11.585610389709473,
|
| 259 |
"learning_rate": 1.2800000000000001e-05,
|
| 260 |
+
"loss": 1.409,
|
| 261 |
"step": 320
|
| 262 |
},
|
| 263 |
{
|
| 264 |
+
"epoch": 0.4796511627906977,
|
| 265 |
+
"grad_norm": 13.202520370483398,
|
| 266 |
"learning_rate": 1.3200000000000002e-05,
|
| 267 |
+
"loss": 1.3285,
|
| 268 |
"step": 330
|
| 269 |
},
|
| 270 |
{
|
| 271 |
+
"epoch": 0.4941860465116279,
|
| 272 |
+
"grad_norm": 8.829900741577148,
|
| 273 |
"learning_rate": 1.3600000000000002e-05,
|
| 274 |
+
"loss": 1.3261,
|
| 275 |
"step": 340
|
| 276 |
},
|
| 277 |
{
|
| 278 |
+
"epoch": 0.5087209302325582,
|
| 279 |
+
"grad_norm": 11.259300231933594,
|
| 280 |
"learning_rate": 1.4e-05,
|
| 281 |
+
"loss": 1.4546,
|
| 282 |
"step": 350
|
| 283 |
},
|
| 284 |
{
|
| 285 |
+
"epoch": 0.5232558139534884,
|
| 286 |
+
"grad_norm": 8.918652534484863,
|
| 287 |
"learning_rate": 1.4400000000000001e-05,
|
| 288 |
+
"loss": 1.2455,
|
| 289 |
"step": 360
|
| 290 |
},
|
| 291 |
{
|
| 292 |
+
"epoch": 0.5377906976744186,
|
| 293 |
+
"grad_norm": 10.523372650146484,
|
| 294 |
"learning_rate": 1.48e-05,
|
| 295 |
+
"loss": 1.3496,
|
| 296 |
"step": 370
|
| 297 |
},
|
| 298 |
{
|
| 299 |
+
"epoch": 0.5523255813953488,
|
| 300 |
+
"grad_norm": 13.649139404296875,
|
| 301 |
"learning_rate": 1.5200000000000002e-05,
|
| 302 |
+
"loss": 1.2688,
|
| 303 |
"step": 380
|
| 304 |
},
|
| 305 |
{
|
| 306 |
+
"epoch": 0.5668604651162791,
|
| 307 |
+
"grad_norm": 15.79927921295166,
|
| 308 |
"learning_rate": 1.5600000000000003e-05,
|
| 309 |
+
"loss": 1.22,
|
| 310 |
"step": 390
|
| 311 |
},
|
| 312 |
{
|
| 313 |
+
"epoch": 0.5813953488372093,
|
| 314 |
+
"grad_norm": 12.712923049926758,
|
| 315 |
"learning_rate": 1.6000000000000003e-05,
|
| 316 |
+
"loss": 1.2153,
|
| 317 |
"step": 400
|
| 318 |
},
|
| 319 |
{
|
| 320 |
+
"epoch": 0.5813953488372093,
|
| 321 |
+
"eval_accuracy": 0.689375,
|
| 322 |
+
"eval_accuracy_label_arts, culture, entertainment and media": 0.7894736842105263,
|
| 323 |
+
"eval_accuracy_label_conflict, war and peace": 0.655436447166922,
|
| 324 |
+
"eval_accuracy_label_crime, law and justice": 0.8195876288659794,
|
| 325 |
+
"eval_accuracy_label_disaster, accident, and emergency incident": 0.8561643835616438,
|
| 326 |
+
"eval_accuracy_label_economy, business, and finance": 0.6407766990291263,
|
| 327 |
+
"eval_accuracy_label_environment": 0.2413793103448276,
|
| 328 |
+
"eval_accuracy_label_health": 0.8333333333333334,
|
| 329 |
+
"eval_accuracy_label_human interest": 0.13636363636363635,
|
| 330 |
+
"eval_accuracy_label_labour": 0.0,
|
| 331 |
+
"eval_accuracy_label_lifestyle and leisure": 0.6666666666666666,
|
| 332 |
+
"eval_accuracy_label_politics": 0.8466898954703833,
|
| 333 |
"eval_accuracy_label_religion": 0.0,
|
| 334 |
+
"eval_accuracy_label_science and technology": 0.375,
|
| 335 |
+
"eval_accuracy_label_society": 0.015384615384615385,
|
| 336 |
"eval_accuracy_label_sport": 0.9615384615384616,
|
| 337 |
"eval_accuracy_label_weather": 1.0,
|
| 338 |
+
"eval_f1": 0.6867853551801604,
|
| 339 |
+
"eval_loss": 1.0224822759628296,
|
| 340 |
+
"eval_precision": 0.765204058143133,
|
| 341 |
+
"eval_recall": 0.689375,
|
| 342 |
+
"eval_runtime": 12.5726,
|
| 343 |
+
"eval_samples_per_second": 127.261,
|
| 344 |
+
"eval_steps_per_second": 7.954,
|
| 345 |
"step": 400
|
| 346 |
},
|
| 347 |
{
|
| 348 |
+
"epoch": 0.5959302325581395,
|
| 349 |
+
"grad_norm": 11.235016822814941,
|
| 350 |
"learning_rate": 1.64e-05,
|
| 351 |
+
"loss": 1.2041,
|
| 352 |
"step": 410
|
| 353 |
},
|
| 354 |
{
|
| 355 |
+
"epoch": 0.6104651162790697,
|
| 356 |
+
"grad_norm": 10.808572769165039,
|
| 357 |
"learning_rate": 1.6800000000000002e-05,
|
| 358 |
+
"loss": 1.1168,
|
| 359 |
"step": 420
|
| 360 |
},
|
| 361 |
{
|
| 362 |
+
"epoch": 0.625,
|
| 363 |
+
"grad_norm": 16.74306869506836,
|
| 364 |
"learning_rate": 1.72e-05,
|
| 365 |
+
"loss": 1.1227,
|
| 366 |
"step": 430
|
| 367 |
},
|
| 368 |
{
|
| 369 |
+
"epoch": 0.6395348837209303,
|
| 370 |
+
"grad_norm": 9.389023780822754,
|
| 371 |
"learning_rate": 1.76e-05,
|
| 372 |
+
"loss": 1.1575,
|
| 373 |
"step": 440
|
| 374 |
},
|
| 375 |
{
|
| 376 |
+
"epoch": 0.6540697674418605,
|
| 377 |
+
"grad_norm": 15.347270965576172,
|
| 378 |
"learning_rate": 1.8e-05,
|
| 379 |
+
"loss": 0.9979,
|
| 380 |
"step": 450
|
| 381 |
},
|
| 382 |
{
|
| 383 |
+
"epoch": 0.6686046511627907,
|
| 384 |
+
"grad_norm": 11.799999237060547,
|
| 385 |
"learning_rate": 1.8400000000000003e-05,
|
| 386 |
+
"loss": 1.0845,
|
| 387 |
"step": 460
|
| 388 |
},
|
| 389 |
{
|
| 390 |
+
"epoch": 0.6831395348837209,
|
| 391 |
+
"grad_norm": 12.181020736694336,
|
| 392 |
"learning_rate": 1.88e-05,
|
| 393 |
+
"loss": 1.0833,
|
| 394 |
"step": 470
|
| 395 |
},
|
| 396 |
{
|
| 397 |
+
"epoch": 0.6976744186046512,
|
| 398 |
+
"grad_norm": 12.819634437561035,
|
| 399 |
"learning_rate": 1.9200000000000003e-05,
|
| 400 |
+
"loss": 1.137,
|
| 401 |
"step": 480
|
| 402 |
},
|
| 403 |
{
|
| 404 |
+
"epoch": 0.7122093023255814,
|
| 405 |
+
"grad_norm": 13.189191818237305,
|
| 406 |
"learning_rate": 1.9600000000000002e-05,
|
| 407 |
+
"loss": 1.0458,
|
| 408 |
"step": 490
|
| 409 |
},
|
| 410 |
{
|
| 411 |
+
"epoch": 0.7267441860465116,
|
| 412 |
+
"grad_norm": 9.888482093811035,
|
| 413 |
"learning_rate": 2e-05,
|
| 414 |
+
"loss": 1.0184,
|
| 415 |
"step": 500
|
| 416 |
},
|
| 417 |
{
|
| 418 |
+
"epoch": 0.7412790697674418,
|
| 419 |
+
"grad_norm": 9.71278190612793,
|
| 420 |
+
"learning_rate": 1.987212276214834e-05,
|
| 421 |
+
"loss": 0.98,
|
| 422 |
"step": 510
|
| 423 |
},
|
| 424 |
{
|
| 425 |
+
"epoch": 0.7558139534883721,
|
| 426 |
+
"grad_norm": 14.008501052856445,
|
| 427 |
+
"learning_rate": 1.9744245524296677e-05,
|
| 428 |
+
"loss": 0.9985,
|
| 429 |
"step": 520
|
| 430 |
},
|
| 431 |
{
|
| 432 |
+
"epoch": 0.7703488372093024,
|
| 433 |
+
"grad_norm": 16.69378662109375,
|
| 434 |
+
"learning_rate": 1.9616368286445014e-05,
|
| 435 |
+
"loss": 0.9146,
|
| 436 |
"step": 530
|
| 437 |
},
|
| 438 |
{
|
| 439 |
+
"epoch": 0.7848837209302325,
|
| 440 |
+
"grad_norm": 15.569829940795898,
|
| 441 |
+
"learning_rate": 1.9488491048593352e-05,
|
| 442 |
+
"loss": 0.9717,
|
| 443 |
"step": 540
|
| 444 |
},
|
| 445 |
{
|
| 446 |
+
"epoch": 0.7994186046511628,
|
| 447 |
+
"grad_norm": 10.90029239654541,
|
| 448 |
+
"learning_rate": 1.936061381074169e-05,
|
| 449 |
+
"loss": 1.0832,
|
| 450 |
"step": 550
|
| 451 |
},
|
| 452 |
{
|
| 453 |
+
"epoch": 0.813953488372093,
|
| 454 |
+
"grad_norm": 16.751508712768555,
|
| 455 |
+
"learning_rate": 1.9232736572890027e-05,
|
| 456 |
+
"loss": 0.9345,
|
| 457 |
"step": 560
|
| 458 |
},
|
| 459 |
{
|
| 460 |
+
"epoch": 0.8284883720930233,
|
| 461 |
+
"grad_norm": 14.394732475280762,
|
| 462 |
+
"learning_rate": 1.9104859335038365e-05,
|
| 463 |
+
"loss": 1.0405,
|
| 464 |
"step": 570
|
| 465 |
},
|
| 466 |
{
|
| 467 |
+
"epoch": 0.8430232558139535,
|
| 468 |
+
"grad_norm": 11.01040267944336,
|
| 469 |
+
"learning_rate": 1.8976982097186702e-05,
|
| 470 |
+
"loss": 0.9115,
|
| 471 |
"step": 580
|
| 472 |
},
|
| 473 |
{
|
| 474 |
+
"epoch": 0.8575581395348837,
|
| 475 |
+
"grad_norm": 15.872978210449219,
|
| 476 |
+
"learning_rate": 1.884910485933504e-05,
|
| 477 |
+
"loss": 0.8443,
|
| 478 |
"step": 590
|
| 479 |
},
|
| 480 |
{
|
| 481 |
+
"epoch": 0.872093023255814,
|
| 482 |
+
"grad_norm": 19.019338607788086,
|
| 483 |
+
"learning_rate": 1.8721227621483378e-05,
|
| 484 |
+
"loss": 0.954,
|
| 485 |
"step": 600
|
| 486 |
},
|
| 487 |
{
|
| 488 |
+
"epoch": 0.872093023255814,
|
| 489 |
+
"eval_accuracy": 0.723125,
|
| 490 |
+
"eval_accuracy_label_arts, culture, entertainment and media": 0.7368421052631579,
|
| 491 |
+
"eval_accuracy_label_conflict, war and peace": 0.77947932618683,
|
| 492 |
+
"eval_accuracy_label_crime, law and justice": 0.8917525773195877,
|
| 493 |
+
"eval_accuracy_label_disaster, accident, and emergency incident": 0.8698630136986302,
|
| 494 |
+
"eval_accuracy_label_economy, business, and finance": 0.6213592233009708,
|
| 495 |
+
"eval_accuracy_label_environment": 0.3448275862068966,
|
| 496 |
+
"eval_accuracy_label_health": 0.8888888888888888,
|
| 497 |
+
"eval_accuracy_label_human interest": 0.18181818181818182,
|
| 498 |
"eval_accuracy_label_labour": 1.0,
|
| 499 |
+
"eval_accuracy_label_lifestyle and leisure": 0.5555555555555556,
|
| 500 |
+
"eval_accuracy_label_politics": 0.6898954703832753,
|
| 501 |
"eval_accuracy_label_religion": 0.0,
|
| 502 |
+
"eval_accuracy_label_science and technology": 0.25,
|
| 503 |
+
"eval_accuracy_label_society": 0.046153846153846156,
|
| 504 |
+
"eval_accuracy_label_sport": 0.9615384615384616,
|
| 505 |
"eval_accuracy_label_weather": 1.0,
|
| 506 |
+
"eval_f1": 0.7137964265878213,
|
| 507 |
+
"eval_loss": 0.8858376145362854,
|
| 508 |
+
"eval_precision": 0.7308873219026191,
|
| 509 |
+
"eval_recall": 0.723125,
|
| 510 |
+
"eval_runtime": 12.6079,
|
| 511 |
+
"eval_samples_per_second": 126.905,
|
| 512 |
+
"eval_steps_per_second": 7.932,
|
| 513 |
"step": 600
|
| 514 |
},
|
| 515 |
{
|
| 516 |
+
"epoch": 0.8866279069767442,
|
| 517 |
+
"grad_norm": 8.310386657714844,
|
| 518 |
+
"learning_rate": 1.8593350383631715e-05,
|
| 519 |
+
"loss": 0.8133,
|
| 520 |
"step": 610
|
| 521 |
},
|
| 522 |
{
|
| 523 |
+
"epoch": 0.9011627906976745,
|
| 524 |
+
"grad_norm": 12.042872428894043,
|
| 525 |
+
"learning_rate": 1.8465473145780053e-05,
|
| 526 |
+
"loss": 0.9059,
|
| 527 |
"step": 620
|
| 528 |
},
|
| 529 |
{
|
| 530 |
+
"epoch": 0.9156976744186046,
|
| 531 |
+
"grad_norm": 10.83311653137207,
|
| 532 |
+
"learning_rate": 1.833759590792839e-05,
|
| 533 |
+
"loss": 0.9357,
|
| 534 |
"step": 630
|
| 535 |
},
|
| 536 |
{
|
| 537 |
+
"epoch": 0.9302325581395349,
|
| 538 |
+
"grad_norm": 11.619269371032715,
|
| 539 |
+
"learning_rate": 1.8209718670076728e-05,
|
| 540 |
+
"loss": 0.967,
|
| 541 |
"step": 640
|
| 542 |
},
|
| 543 |
{
|
| 544 |
+
"epoch": 0.9447674418604651,
|
| 545 |
+
"grad_norm": 11.699959754943848,
|
| 546 |
+
"learning_rate": 1.8081841432225066e-05,
|
| 547 |
+
"loss": 0.748,
|
| 548 |
"step": 650
|
| 549 |
},
|
| 550 |
{
|
| 551 |
+
"epoch": 0.9593023255813954,
|
| 552 |
+
"grad_norm": 15.047694206237793,
|
| 553 |
+
"learning_rate": 1.7953964194373403e-05,
|
| 554 |
+
"loss": 0.8928,
|
| 555 |
"step": 660
|
| 556 |
},
|
| 557 |
{
|
| 558 |
+
"epoch": 0.9738372093023255,
|
| 559 |
+
"grad_norm": 13.320290565490723,
|
| 560 |
+
"learning_rate": 1.782608695652174e-05,
|
| 561 |
+
"loss": 0.8786,
|
| 562 |
"step": 670
|
| 563 |
},
|
| 564 |
{
|
| 565 |
+
"epoch": 0.9883720930232558,
|
| 566 |
+
"grad_norm": 13.041028022766113,
|
| 567 |
+
"learning_rate": 1.769820971867008e-05,
|
| 568 |
+
"loss": 0.9478,
|
| 569 |
"step": 680
|
| 570 |
},
|
| 571 |
{
|
| 572 |
+
"epoch": 1.002906976744186,
|
| 573 |
+
"grad_norm": 14.132765769958496,
|
| 574 |
+
"learning_rate": 1.7570332480818416e-05,
|
| 575 |
+
"loss": 0.8231,
|
| 576 |
"step": 690
|
| 577 |
},
|
| 578 |
{
|
| 579 |
+
"epoch": 1.0174418604651163,
|
| 580 |
+
"grad_norm": 7.3977837562561035,
|
| 581 |
+
"learning_rate": 1.7442455242966754e-05,
|
| 582 |
+
"loss": 0.7806,
|
| 583 |
"step": 700
|
| 584 |
},
|
| 585 |
{
|
| 586 |
+
"epoch": 1.0319767441860466,
|
| 587 |
+
"grad_norm": 16.572118759155273,
|
| 588 |
+
"learning_rate": 1.731457800511509e-05,
|
| 589 |
+
"loss": 0.7654,
|
| 590 |
"step": 710
|
| 591 |
},
|
| 592 |
{
|
| 593 |
+
"epoch": 1.0465116279069768,
|
| 594 |
+
"grad_norm": 21.040008544921875,
|
| 595 |
+
"learning_rate": 1.718670076726343e-05,
|
| 596 |
+
"loss": 0.7856,
|
| 597 |
"step": 720
|
| 598 |
},
|
| 599 |
{
|
| 600 |
+
"epoch": 1.0610465116279069,
|
| 601 |
+
"grad_norm": 21.33478355407715,
|
| 602 |
+
"learning_rate": 1.7058823529411767e-05,
|
| 603 |
+
"loss": 0.7323,
|
| 604 |
"step": 730
|
| 605 |
},
|
| 606 |
{
|
| 607 |
+
"epoch": 1.0755813953488371,
|
| 608 |
+
"grad_norm": 13.03322696685791,
|
| 609 |
+
"learning_rate": 1.6930946291560104e-05,
|
| 610 |
+
"loss": 0.7396,
|
| 611 |
"step": 740
|
| 612 |
},
|
| 613 |
{
|
| 614 |
+
"epoch": 1.0901162790697674,
|
| 615 |
+
"grad_norm": 19.618709564208984,
|
| 616 |
+
"learning_rate": 1.6803069053708442e-05,
|
| 617 |
+
"loss": 0.8284,
|
| 618 |
"step": 750
|
| 619 |
},
|
| 620 |
{
|
| 621 |
+
"epoch": 1.1046511627906976,
|
| 622 |
+
"grad_norm": 21.11569595336914,
|
| 623 |
+
"learning_rate": 1.667519181585678e-05,
|
| 624 |
+
"loss": 0.7032,
|
| 625 |
"step": 760
|
| 626 |
},
|
| 627 |
{
|
| 628 |
+
"epoch": 1.119186046511628,
|
| 629 |
+
"grad_norm": 13.00526237487793,
|
| 630 |
+
"learning_rate": 1.6547314578005117e-05,
|
| 631 |
+
"loss": 0.7199,
|
| 632 |
"step": 770
|
| 633 |
},
|
| 634 |
{
|
| 635 |
+
"epoch": 1.1337209302325582,
|
| 636 |
+
"grad_norm": 12.57153034210205,
|
| 637 |
+
"learning_rate": 1.6419437340153455e-05,
|
| 638 |
+
"loss": 0.6688,
|
| 639 |
"step": 780
|
| 640 |
},
|
| 641 |
{
|
| 642 |
+
"epoch": 1.1482558139534884,
|
| 643 |
+
"grad_norm": 11.098487854003906,
|
| 644 |
+
"learning_rate": 1.6291560102301792e-05,
|
| 645 |
+
"loss": 0.6463,
|
| 646 |
"step": 790
|
| 647 |
},
|
| 648 |
{
|
| 649 |
+
"epoch": 1.1627906976744187,
|
| 650 |
+
"grad_norm": 12.68520736694336,
|
| 651 |
+
"learning_rate": 1.616368286445013e-05,
|
| 652 |
+
"loss": 0.6662,
|
| 653 |
"step": 800
|
| 654 |
},
|
| 655 |
{
|
| 656 |
+
"epoch": 1.1627906976744187,
|
| 657 |
+
"eval_accuracy": 0.688125,
|
| 658 |
+
"eval_accuracy_label_arts, culture, entertainment and media": 0.7894736842105263,
|
| 659 |
+
"eval_accuracy_label_conflict, war and peace": 0.6125574272588055,
|
| 660 |
+
"eval_accuracy_label_crime, law and justice": 0.845360824742268,
|
| 661 |
+
"eval_accuracy_label_disaster, accident, and emergency incident": 0.863013698630137,
|
| 662 |
+
"eval_accuracy_label_economy, business, and finance": 0.6504854368932039,
|
| 663 |
+
"eval_accuracy_label_environment": 0.4482758620689655,
|
| 664 |
+
"eval_accuracy_label_health": 0.7222222222222222,
|
| 665 |
+
"eval_accuracy_label_human interest": 0.22727272727272727,
|
| 666 |
"eval_accuracy_label_labour": 1.0,
|
| 667 |
+
"eval_accuracy_label_lifestyle and leisure": 0.4444444444444444,
|
| 668 |
+
"eval_accuracy_label_politics": 0.8292682926829268,
|
| 669 |
"eval_accuracy_label_religion": 0.0,
|
| 670 |
+
"eval_accuracy_label_science and technology": 0.5416666666666666,
|
| 671 |
+
"eval_accuracy_label_society": 0.23076923076923078,
|
| 672 |
"eval_accuracy_label_sport": 0.9615384615384616,
|
| 673 |
"eval_accuracy_label_weather": 1.0,
|
| 674 |
+
"eval_f1": 0.7008818694331442,
|
| 675 |
+
"eval_loss": 0.9380582571029663,
|
| 676 |
+
"eval_precision": 0.7617561903814418,
|
| 677 |
+
"eval_recall": 0.688125,
|
| 678 |
+
"eval_runtime": 12.5844,
|
| 679 |
+
"eval_samples_per_second": 127.142,
|
| 680 |
+
"eval_steps_per_second": 7.946,
|
| 681 |
"step": 800
|
| 682 |
},
|
| 683 |
{
|
| 684 |
+
"epoch": 1.177325581395349,
|
| 685 |
+
"grad_norm": 12.958664894104004,
|
| 686 |
+
"learning_rate": 1.6035805626598467e-05,
|
| 687 |
+
"loss": 0.6513,
|
| 688 |
"step": 810
|
| 689 |
},
|
| 690 |
{
|
| 691 |
+
"epoch": 1.191860465116279,
|
| 692 |
+
"grad_norm": 13.338906288146973,
|
| 693 |
+
"learning_rate": 1.5907928388746805e-05,
|
| 694 |
+
"loss": 0.7241,
|
| 695 |
"step": 820
|
| 696 |
},
|
| 697 |
{
|
| 698 |
+
"epoch": 1.2063953488372092,
|
| 699 |
+
"grad_norm": 12.350173950195312,
|
| 700 |
+
"learning_rate": 1.5780051150895143e-05,
|
| 701 |
+
"loss": 0.6857,
|
| 702 |
"step": 830
|
| 703 |
},
|
| 704 |
{
|
| 705 |
+
"epoch": 1.2209302325581395,
|
| 706 |
+
"grad_norm": 9.547300338745117,
|
| 707 |
+
"learning_rate": 1.565217391304348e-05,
|
| 708 |
+
"loss": 0.5884,
|
| 709 |
"step": 840
|
| 710 |
},
|
| 711 |
{
|
| 712 |
+
"epoch": 1.2354651162790697,
|
| 713 |
+
"grad_norm": 9.108930587768555,
|
| 714 |
+
"learning_rate": 1.5524296675191818e-05,
|
| 715 |
+
"loss": 0.5416,
|
| 716 |
"step": 850
|
| 717 |
},
|
| 718 |
{
|
| 719 |
+
"epoch": 1.25,
|
| 720 |
+
"grad_norm": 7.098880767822266,
|
| 721 |
+
"learning_rate": 1.5396419437340155e-05,
|
| 722 |
+
"loss": 0.5443,
|
| 723 |
"step": 860
|
| 724 |
},
|
| 725 |
{
|
| 726 |
+
"epoch": 1.2645348837209303,
|
| 727 |
+
"grad_norm": 15.5491304397583,
|
| 728 |
+
"learning_rate": 1.5268542199488493e-05,
|
| 729 |
+
"loss": 0.638,
|
| 730 |
"step": 870
|
| 731 |
},
|
| 732 |
{
|
| 733 |
+
"epoch": 1.2790697674418605,
|
| 734 |
+
"grad_norm": 14.731600761413574,
|
| 735 |
+
"learning_rate": 1.514066496163683e-05,
|
| 736 |
+
"loss": 0.6516,
|
| 737 |
"step": 880
|
| 738 |
},
|
| 739 |
{
|
| 740 |
+
"epoch": 1.2936046511627908,
|
| 741 |
+
"grad_norm": 6.682343006134033,
|
| 742 |
+
"learning_rate": 1.5012787723785167e-05,
|
| 743 |
+
"loss": 0.6165,
|
| 744 |
"step": 890
|
| 745 |
},
|
| 746 |
{
|
| 747 |
+
"epoch": 1.308139534883721,
|
| 748 |
+
"grad_norm": 14.463234901428223,
|
| 749 |
+
"learning_rate": 1.4884910485933506e-05,
|
| 750 |
+
"loss": 0.6455,
|
| 751 |
"step": 900
|
| 752 |
},
|
| 753 |
{
|
| 754 |
+
"epoch": 1.322674418604651,
|
| 755 |
+
"grad_norm": 14.159065246582031,
|
| 756 |
+
"learning_rate": 1.4757033248081842e-05,
|
| 757 |
+
"loss": 0.6129,
|
| 758 |
"step": 910
|
| 759 |
},
|
| 760 |
{
|
| 761 |
+
"epoch": 1.3372093023255813,
|
| 762 |
+
"grad_norm": 14.65009880065918,
|
| 763 |
+
"learning_rate": 1.4629156010230181e-05,
|
| 764 |
+
"loss": 0.5702,
|
| 765 |
"step": 920
|
| 766 |
},
|
| 767 |
{
|
| 768 |
+
"epoch": 1.3517441860465116,
|
| 769 |
+
"grad_norm": 16.38220977783203,
|
| 770 |
+
"learning_rate": 1.4501278772378517e-05,
|
| 771 |
+
"loss": 0.6183,
|
| 772 |
"step": 930
|
| 773 |
},
|
| 774 |
{
|
| 775 |
+
"epoch": 1.3662790697674418,
|
| 776 |
+
"grad_norm": 9.162236213684082,
|
| 777 |
+
"learning_rate": 1.4373401534526856e-05,
|
| 778 |
+
"loss": 0.547,
|
| 779 |
"step": 940
|
| 780 |
},
|
| 781 |
{
|
| 782 |
+
"epoch": 1.380813953488372,
|
| 783 |
+
"grad_norm": 9.222379684448242,
|
| 784 |
+
"learning_rate": 1.4245524296675192e-05,
|
| 785 |
+
"loss": 0.5629,
|
| 786 |
"step": 950
|
| 787 |
},
|
| 788 |
{
|
| 789 |
+
"epoch": 1.3953488372093024,
|
| 790 |
+
"grad_norm": 8.016491889953613,
|
| 791 |
+
"learning_rate": 1.4117647058823532e-05,
|
| 792 |
+
"loss": 0.6529,
|
| 793 |
"step": 960
|
| 794 |
},
|
| 795 |
{
|
| 796 |
+
"epoch": 1.4098837209302326,
|
| 797 |
+
"grad_norm": 8.260580062866211,
|
| 798 |
+
"learning_rate": 1.3989769820971867e-05,
|
| 799 |
+
"loss": 0.5309,
|
| 800 |
"step": 970
|
| 801 |
},
|
| 802 |
{
|
| 803 |
+
"epoch": 1.4244186046511627,
|
| 804 |
+
"grad_norm": 16.81390953063965,
|
| 805 |
+
"learning_rate": 1.3861892583120207e-05,
|
| 806 |
+
"loss": 0.6785,
|
| 807 |
"step": 980
|
| 808 |
},
|
| 809 |
{
|
| 810 |
+
"epoch": 1.4389534883720931,
|
| 811 |
+
"grad_norm": 11.895804405212402,
|
| 812 |
+
"learning_rate": 1.3734015345268543e-05,
|
| 813 |
+
"loss": 0.5976,
|
| 814 |
"step": 990
|
| 815 |
},
|
| 816 |
{
|
| 817 |
+
"epoch": 1.4534883720930232,
|
| 818 |
+
"grad_norm": 8.548684120178223,
|
| 819 |
+
"learning_rate": 1.3606138107416882e-05,
|
| 820 |
+
"loss": 0.5554,
|
| 821 |
"step": 1000
|
| 822 |
},
|
| 823 |
{
|
| 824 |
+
"epoch": 1.4534883720930232,
|
| 825 |
+
"eval_accuracy": 0.7025,
|
| 826 |
+
"eval_accuracy_label_arts, culture, entertainment and media": 0.7368421052631579,
|
| 827 |
+
"eval_accuracy_label_conflict, war and peace": 0.6477794793261868,
|
| 828 |
+
"eval_accuracy_label_crime, law and justice": 0.9020618556701031,
|
| 829 |
+
"eval_accuracy_label_disaster, accident, and emergency incident": 0.8561643835616438,
|
| 830 |
+
"eval_accuracy_label_economy, business, and finance": 0.6601941747572816,
|
| 831 |
+
"eval_accuracy_label_environment": 0.3103448275862069,
|
| 832 |
+
"eval_accuracy_label_health": 0.7777777777777778,
|
| 833 |
+
"eval_accuracy_label_human interest": 0.36363636363636365,
|
| 834 |
+
"eval_accuracy_label_labour": 0.5,
|
| 835 |
+
"eval_accuracy_label_lifestyle and leisure": 0.5555555555555556,
|
| 836 |
+
"eval_accuracy_label_politics": 0.8083623693379791,
|
| 837 |
"eval_accuracy_label_religion": 0.0,
|
| 838 |
+
"eval_accuracy_label_science and technology": 0.5,
|
| 839 |
+
"eval_accuracy_label_society": 0.18461538461538463,
|
| 840 |
"eval_accuracy_label_sport": 0.9615384615384616,
|
| 841 |
+
"eval_accuracy_label_weather": 1.0,
|
| 842 |
+
"eval_f1": 0.7123723699365401,
|
| 843 |
+
"eval_loss": 0.8791087865829468,
|
| 844 |
+
"eval_precision": 0.7628439748212975,
|
| 845 |
+
"eval_recall": 0.7025,
|
| 846 |
+
"eval_runtime": 12.5844,
|
| 847 |
+
"eval_samples_per_second": 127.142,
|
| 848 |
+
"eval_steps_per_second": 7.946,
|
| 849 |
"step": 1000
|
| 850 |
},
|
| 851 |
{
|
| 852 |
+
"epoch": 1.4680232558139534,
|
| 853 |
+
"grad_norm": 13.218795776367188,
|
| 854 |
+
"learning_rate": 1.3478260869565218e-05,
|
| 855 |
+
"loss": 0.5587,
|
| 856 |
"step": 1010
|
| 857 |
},
|
| 858 |
{
|
| 859 |
+
"epoch": 1.4825581395348837,
|
| 860 |
+
"grad_norm": 21.222410202026367,
|
| 861 |
+
"learning_rate": 1.3350383631713557e-05,
|
| 862 |
+
"loss": 0.6128,
|
| 863 |
"step": 1020
|
| 864 |
},
|
| 865 |
{
|
| 866 |
+
"epoch": 1.497093023255814,
|
| 867 |
+
"grad_norm": 10.608304977416992,
|
| 868 |
+
"learning_rate": 1.3222506393861893e-05,
|
| 869 |
+
"loss": 0.5334,
|
| 870 |
"step": 1030
|
| 871 |
},
|
| 872 |
{
|
| 873 |
+
"epoch": 1.5116279069767442,
|
| 874 |
+
"grad_norm": 9.839329719543457,
|
| 875 |
+
"learning_rate": 1.3094629156010232e-05,
|
| 876 |
+
"loss": 0.4974,
|
| 877 |
"step": 1040
|
| 878 |
},
|
| 879 |
{
|
| 880 |
+
"epoch": 1.5261627906976745,
|
| 881 |
+
"grad_norm": 16.351816177368164,
|
| 882 |
+
"learning_rate": 1.2966751918158568e-05,
|
| 883 |
+
"loss": 0.5387,
|
| 884 |
"step": 1050
|
| 885 |
},
|
| 886 |
{
|
| 887 |
+
"epoch": 1.5406976744186047,
|
| 888 |
+
"grad_norm": 13.274540901184082,
|
| 889 |
+
"learning_rate": 1.2838874680306906e-05,
|
| 890 |
+
"loss": 0.5928,
|
| 891 |
"step": 1060
|
| 892 |
},
|
| 893 |
{
|
| 894 |
+
"epoch": 1.5552325581395348,
|
| 895 |
+
"grad_norm": 11.493636131286621,
|
| 896 |
+
"learning_rate": 1.2710997442455244e-05,
|
| 897 |
+
"loss": 0.4391,
|
| 898 |
"step": 1070
|
| 899 |
},
|
| 900 |
{
|
| 901 |
+
"epoch": 1.5697674418604652,
|
| 902 |
+
"grad_norm": 7.543900012969971,
|
| 903 |
+
"learning_rate": 1.2583120204603581e-05,
|
| 904 |
+
"loss": 0.5845,
|
| 905 |
"step": 1080
|
| 906 |
},
|
| 907 |
{
|
| 908 |
+
"epoch": 1.5843023255813953,
|
| 909 |
+
"grad_norm": 18.52677345275879,
|
| 910 |
+
"learning_rate": 1.2455242966751919e-05,
|
| 911 |
+
"loss": 0.4857,
|
| 912 |
"step": 1090
|
| 913 |
},
|
| 914 |
{
|
| 915 |
+
"epoch": 1.5988372093023255,
|
| 916 |
+
"grad_norm": 10.437915802001953,
|
| 917 |
+
"learning_rate": 1.2327365728900256e-05,
|
| 918 |
+
"loss": 0.4946,
|
| 919 |
"step": 1100
|
| 920 |
},
|
| 921 |
{
|
| 922 |
+
"epoch": 1.6133720930232558,
|
| 923 |
+
"grad_norm": 9.237434387207031,
|
| 924 |
+
"learning_rate": 1.2199488491048594e-05,
|
| 925 |
+
"loss": 0.6215,
|
| 926 |
"step": 1110
|
| 927 |
},
|
| 928 |
{
|
| 929 |
+
"epoch": 1.627906976744186,
|
| 930 |
+
"grad_norm": 15.632594108581543,
|
| 931 |
+
"learning_rate": 1.2071611253196932e-05,
|
| 932 |
+
"loss": 0.5642,
|
| 933 |
"step": 1120
|
| 934 |
},
|
| 935 |
{
|
| 936 |
+
"epoch": 1.6424418604651163,
|
| 937 |
+
"grad_norm": 16.443044662475586,
|
| 938 |
+
"learning_rate": 1.194373401534527e-05,
|
| 939 |
+
"loss": 0.5714,
|
| 940 |
"step": 1130
|
| 941 |
},
|
| 942 |
{
|
| 943 |
+
"epoch": 1.6569767441860463,
|
| 944 |
+
"grad_norm": 15.662179946899414,
|
| 945 |
+
"learning_rate": 1.1815856777493607e-05,
|
| 946 |
+
"loss": 0.4438,
|
| 947 |
"step": 1140
|
| 948 |
},
|
| 949 |
{
|
| 950 |
+
"epoch": 1.6715116279069768,
|
| 951 |
+
"grad_norm": 14.527420043945312,
|
| 952 |
+
"learning_rate": 1.1687979539641944e-05,
|
| 953 |
+
"loss": 0.5733,
|
| 954 |
"step": 1150
|
| 955 |
},
|
| 956 |
{
|
| 957 |
+
"epoch": 1.6860465116279069,
|
| 958 |
+
"grad_norm": 20.69162368774414,
|
| 959 |
+
"learning_rate": 1.1560102301790282e-05,
|
| 960 |
+
"loss": 0.5324,
|
| 961 |
"step": 1160
|
| 962 |
},
|
| 963 |
{
|
| 964 |
+
"epoch": 1.7005813953488373,
|
| 965 |
+
"grad_norm": 11.502646446228027,
|
| 966 |
+
"learning_rate": 1.143222506393862e-05,
|
| 967 |
+
"loss": 0.4495,
|
| 968 |
"step": 1170
|
| 969 |
},
|
| 970 |
{
|
| 971 |
+
"epoch": 1.7151162790697674,
|
| 972 |
+
"grad_norm": 8.692178726196289,
|
| 973 |
+
"learning_rate": 1.1304347826086957e-05,
|
| 974 |
+
"loss": 0.4218,
|
| 975 |
"step": 1180
|
| 976 |
},
|
| 977 |
{
|
| 978 |
+
"epoch": 1.7296511627906976,
|
| 979 |
+
"grad_norm": 11.3637113571167,
|
| 980 |
+
"learning_rate": 1.1176470588235295e-05,
|
| 981 |
+
"loss": 0.5778,
|
| 982 |
"step": 1190
|
| 983 |
},
|
| 984 |
{
|
| 985 |
+
"epoch": 1.744186046511628,
|
| 986 |
+
"grad_norm": 7.9896721839904785,
|
| 987 |
+
"learning_rate": 1.1048593350383632e-05,
|
| 988 |
+
"loss": 0.4396,
|
| 989 |
"step": 1200
|
| 990 |
},
|
| 991 |
{
|
| 992 |
+
"epoch": 1.744186046511628,
|
| 993 |
+
"eval_accuracy": 0.7175,
|
| 994 |
+
"eval_accuracy_label_arts, culture, entertainment and media": 0.7894736842105263,
|
| 995 |
+
"eval_accuracy_label_conflict, war and peace": 0.663093415007657,
|
| 996 |
+
"eval_accuracy_label_crime, law and justice": 0.8195876288659794,
|
| 997 |
+
"eval_accuracy_label_disaster, accident, and emergency incident": 0.8835616438356164,
|
| 998 |
+
"eval_accuracy_label_economy, business, and finance": 0.6893203883495146,
|
| 999 |
+
"eval_accuracy_label_environment": 0.3793103448275862,
|
| 1000 |
+
"eval_accuracy_label_health": 0.8333333333333334,
|
| 1001 |
+
"eval_accuracy_label_human interest": 0.4090909090909091,
|
| 1002 |
"eval_accuracy_label_labour": 0.5,
|
| 1003 |
+
"eval_accuracy_label_lifestyle and leisure": 0.5555555555555556,
|
| 1004 |
+
"eval_accuracy_label_politics": 0.8362369337979094,
|
| 1005 |
"eval_accuracy_label_religion": 0.0,
|
| 1006 |
"eval_accuracy_label_science and technology": 0.4166666666666667,
|
| 1007 |
+
"eval_accuracy_label_society": 0.36923076923076925,
|
| 1008 |
"eval_accuracy_label_sport": 0.9615384615384616,
|
| 1009 |
"eval_accuracy_label_weather": 1.0,
|
| 1010 |
+
"eval_f1": 0.7279616125865503,
|
| 1011 |
+
"eval_loss": 0.8275023698806763,
|
| 1012 |
+
"eval_precision": 0.7685711157026855,
|
| 1013 |
+
"eval_recall": 0.7175,
|
| 1014 |
+
"eval_runtime": 12.6195,
|
| 1015 |
+
"eval_samples_per_second": 126.788,
|
| 1016 |
+
"eval_steps_per_second": 7.924,
|
| 1017 |
"step": 1200
|
| 1018 |
},
|
| 1019 |
{
|
| 1020 |
+
"epoch": 1.7587209302325582,
|
| 1021 |
+
"grad_norm": 8.378173828125,
|
| 1022 |
+
"learning_rate": 1.092071611253197e-05,
|
| 1023 |
+
"loss": 0.4187,
|
| 1024 |
"step": 1210
|
| 1025 |
},
|
| 1026 |
{
|
| 1027 |
+
"epoch": 1.7732558139534884,
|
| 1028 |
+
"grad_norm": 22.63587760925293,
|
| 1029 |
+
"learning_rate": 1.0792838874680308e-05,
|
| 1030 |
+
"loss": 0.4558,
|
| 1031 |
"step": 1220
|
| 1032 |
},
|
| 1033 |
{
|
| 1034 |
+
"epoch": 1.7877906976744184,
|
| 1035 |
+
"grad_norm": 16.91942596435547,
|
| 1036 |
+
"learning_rate": 1.0664961636828645e-05,
|
| 1037 |
+
"loss": 0.6778,
|
| 1038 |
"step": 1230
|
| 1039 |
},
|
| 1040 |
{
|
| 1041 |
+
"epoch": 1.802325581395349,
|
| 1042 |
+
"grad_norm": 11.042551040649414,
|
| 1043 |
+
"learning_rate": 1.0537084398976983e-05,
|
| 1044 |
+
"loss": 0.5229,
|
| 1045 |
"step": 1240
|
| 1046 |
},
|
| 1047 |
{
|
| 1048 |
+
"epoch": 1.816860465116279,
|
| 1049 |
+
"grad_norm": 20.5400390625,
|
| 1050 |
+
"learning_rate": 1.040920716112532e-05,
|
| 1051 |
+
"loss": 0.5121,
|
| 1052 |
"step": 1250
|
| 1053 |
},
|
| 1054 |
{
|
| 1055 |
+
"epoch": 1.8313953488372094,
|
| 1056 |
+
"grad_norm": 12.495570182800293,
|
| 1057 |
+
"learning_rate": 1.0281329923273658e-05,
|
| 1058 |
+
"loss": 0.4737,
|
| 1059 |
"step": 1260
|
| 1060 |
},
|
| 1061 |
{
|
| 1062 |
+
"epoch": 1.8459302325581395,
|
| 1063 |
+
"grad_norm": 11.538856506347656,
|
| 1064 |
+
"learning_rate": 1.0153452685421996e-05,
|
| 1065 |
+
"loss": 0.3766,
|
| 1066 |
"step": 1270
|
| 1067 |
},
|
| 1068 |
{
|
| 1069 |
+
"epoch": 1.8604651162790697,
|
| 1070 |
+
"grad_norm": 8.165515899658203,
|
| 1071 |
+
"learning_rate": 1.0025575447570333e-05,
|
| 1072 |
+
"loss": 0.5152,
|
| 1073 |
"step": 1280
|
| 1074 |
},
|
| 1075 |
{
|
| 1076 |
+
"epoch": 1.875,
|
| 1077 |
+
"grad_norm": 8.704108238220215,
|
| 1078 |
+
"learning_rate": 9.897698209718671e-06,
|
| 1079 |
+
"loss": 0.5027,
|
| 1080 |
"step": 1290
|
| 1081 |
},
|
| 1082 |
{
|
| 1083 |
+
"epoch": 1.8895348837209303,
|
| 1084 |
+
"grad_norm": 14.96538257598877,
|
| 1085 |
+
"learning_rate": 9.769820971867009e-06,
|
| 1086 |
+
"loss": 0.5427,
|
| 1087 |
"step": 1300
|
| 1088 |
},
|
| 1089 |
{
|
| 1090 |
+
"epoch": 1.9040697674418605,
|
| 1091 |
+
"grad_norm": 11.60605525970459,
|
| 1092 |
+
"learning_rate": 9.641943734015346e-06,
|
| 1093 |
+
"loss": 0.4236,
|
| 1094 |
"step": 1310
|
| 1095 |
},
|
| 1096 |
{
|
| 1097 |
+
"epoch": 1.9186046511627906,
|
| 1098 |
+
"grad_norm": 10.070377349853516,
|
| 1099 |
+
"learning_rate": 9.514066496163684e-06,
|
| 1100 |
+
"loss": 0.4683,
|
| 1101 |
"step": 1320
|
| 1102 |
},
|
| 1103 |
{
|
| 1104 |
+
"epoch": 1.933139534883721,
|
| 1105 |
+
"grad_norm": 8.57789134979248,
|
| 1106 |
+
"learning_rate": 9.386189258312021e-06,
|
| 1107 |
+
"loss": 0.505,
|
| 1108 |
"step": 1330
|
| 1109 |
},
|
| 1110 |
{
|
| 1111 |
+
"epoch": 1.947674418604651,
|
| 1112 |
+
"grad_norm": 14.64469051361084,
|
| 1113 |
+
"learning_rate": 9.258312020460359e-06,
|
| 1114 |
+
"loss": 0.4311,
|
| 1115 |
"step": 1340
|
| 1116 |
},
|
| 1117 |
{
|
| 1118 |
+
"epoch": 1.9622093023255816,
|
| 1119 |
+
"grad_norm": 16.99574851989746,
|
| 1120 |
+
"learning_rate": 9.130434782608697e-06,
|
| 1121 |
+
"loss": 0.4414,
|
| 1122 |
"step": 1350
|
| 1123 |
},
|
| 1124 |
{
|
| 1125 |
+
"epoch": 1.9767441860465116,
|
| 1126 |
+
"grad_norm": 14.059187889099121,
|
| 1127 |
+
"learning_rate": 9.002557544757034e-06,
|
| 1128 |
+
"loss": 0.5001,
|
| 1129 |
"step": 1360
|
| 1130 |
},
|
| 1131 |
{
|
| 1132 |
+
"epoch": 1.9912790697674418,
|
| 1133 |
+
"grad_norm": 7.389652729034424,
|
| 1134 |
+
"learning_rate": 8.874680306905372e-06,
|
| 1135 |
+
"loss": 0.4546,
|
| 1136 |
"step": 1370
|
| 1137 |
},
|
| 1138 |
{
|
| 1139 |
+
"epoch": 2.005813953488372,
|
| 1140 |
+
"grad_norm": 11.378294944763184,
|
| 1141 |
+
"learning_rate": 8.74680306905371e-06,
|
| 1142 |
+
"loss": 0.4447,
|
| 1143 |
"step": 1380
|
| 1144 |
},
|
| 1145 |
{
|
| 1146 |
+
"epoch": 2.020348837209302,
|
| 1147 |
+
"grad_norm": 9.877668380737305,
|
| 1148 |
+
"learning_rate": 8.618925831202047e-06,
|
| 1149 |
+
"loss": 0.4254,
|
| 1150 |
"step": 1390
|
| 1151 |
},
|
| 1152 |
{
|
| 1153 |
+
"epoch": 2.0348837209302326,
|
| 1154 |
+
"grad_norm": 5.385466575622559,
|
| 1155 |
+
"learning_rate": 8.491048593350385e-06,
|
| 1156 |
+
"loss": 0.383,
|
| 1157 |
"step": 1400
|
| 1158 |
},
|
| 1159 |
{
|
| 1160 |
+
"epoch": 2.0348837209302326,
|
| 1161 |
+
"eval_accuracy": 0.745,
|
| 1162 |
+
"eval_accuracy_label_arts, culture, entertainment and media": 0.6842105263157895,
|
| 1163 |
+
"eval_accuracy_label_conflict, war and peace": 0.7840735068912711,
|
| 1164 |
+
"eval_accuracy_label_crime, law and justice": 0.8865979381443299,
|
| 1165 |
+
"eval_accuracy_label_disaster, accident, and emergency incident": 0.8767123287671232,
|
| 1166 |
+
"eval_accuracy_label_economy, business, and finance": 0.7087378640776699,
|
| 1167 |
+
"eval_accuracy_label_environment": 0.4482758620689655,
|
| 1168 |
+
"eval_accuracy_label_health": 0.7777777777777778,
|
| 1169 |
+
"eval_accuracy_label_human interest": 0.4090909090909091,
|
| 1170 |
"eval_accuracy_label_labour": 0.5,
|
| 1171 |
+
"eval_accuracy_label_lifestyle and leisure": 0.5555555555555556,
|
| 1172 |
+
"eval_accuracy_label_politics": 0.6898954703832753,
|
| 1173 |
"eval_accuracy_label_religion": 0.0,
|
| 1174 |
"eval_accuracy_label_science and technology": 0.4166666666666667,
|
| 1175 |
+
"eval_accuracy_label_society": 0.2923076923076923,
|
| 1176 |
"eval_accuracy_label_sport": 0.9615384615384616,
|
| 1177 |
+
"eval_accuracy_label_weather": 0.0,
|
| 1178 |
+
"eval_f1": 0.7500828579968577,
|
| 1179 |
+
"eval_loss": 0.7928646206855774,
|
| 1180 |
+
"eval_precision": 0.7652956903435105,
|
| 1181 |
+
"eval_recall": 0.745,
|
| 1182 |
+
"eval_runtime": 12.617,
|
| 1183 |
+
"eval_samples_per_second": 126.813,
|
| 1184 |
+
"eval_steps_per_second": 7.926,
|
| 1185 |
"step": 1400
|
| 1186 |
},
|
| 1187 |
{
|
| 1188 |
+
"epoch": 2.0494186046511627,
|
| 1189 |
+
"grad_norm": 10.072980880737305,
|
| 1190 |
+
"learning_rate": 8.363171355498722e-06,
|
| 1191 |
+
"loss": 0.3935,
|
| 1192 |
"step": 1410
|
| 1193 |
},
|
| 1194 |
{
|
| 1195 |
+
"epoch": 2.063953488372093,
|
| 1196 |
+
"grad_norm": 8.466812133789062,
|
| 1197 |
+
"learning_rate": 8.23529411764706e-06,
|
| 1198 |
+
"loss": 0.3343,
|
| 1199 |
"step": 1420
|
| 1200 |
},
|
| 1201 |
{
|
| 1202 |
+
"epoch": 2.078488372093023,
|
| 1203 |
+
"grad_norm": 8.366528511047363,
|
| 1204 |
+
"learning_rate": 8.107416879795397e-06,
|
| 1205 |
+
"loss": 0.3865,
|
| 1206 |
"step": 1430
|
| 1207 |
},
|
| 1208 |
{
|
| 1209 |
+
"epoch": 2.0930232558139537,
|
| 1210 |
+
"grad_norm": 9.42149543762207,
|
| 1211 |
+
"learning_rate": 7.979539641943735e-06,
|
| 1212 |
+
"loss": 0.3071,
|
| 1213 |
"step": 1440
|
| 1214 |
},
|
| 1215 |
{
|
| 1216 |
+
"epoch": 2.1075581395348837,
|
| 1217 |
+
"grad_norm": 7.807155609130859,
|
| 1218 |
+
"learning_rate": 7.851662404092073e-06,
|
| 1219 |
+
"loss": 0.3173,
|
| 1220 |
"step": 1450
|
| 1221 |
},
|
| 1222 |
{
|
| 1223 |
+
"epoch": 2.1220930232558137,
|
| 1224 |
+
"grad_norm": 9.655524253845215,
|
| 1225 |
+
"learning_rate": 7.72378516624041e-06,
|
| 1226 |
+
"loss": 0.4508,
|
| 1227 |
"step": 1460
|
| 1228 |
},
|
| 1229 |
{
|
| 1230 |
+
"epoch": 2.136627906976744,
|
| 1231 |
+
"grad_norm": 8.439408302307129,
|
| 1232 |
+
"learning_rate": 7.595907928388747e-06,
|
| 1233 |
+
"loss": 0.3745,
|
| 1234 |
"step": 1470
|
| 1235 |
},
|
| 1236 |
{
|
| 1237 |
+
"epoch": 2.1511627906976742,
|
| 1238 |
+
"grad_norm": 8.737176895141602,
|
| 1239 |
+
"learning_rate": 7.468030690537085e-06,
|
| 1240 |
+
"loss": 0.3132,
|
| 1241 |
"step": 1480
|
| 1242 |
},
|
| 1243 |
{
|
| 1244 |
+
"epoch": 2.1656976744186047,
|
| 1245 |
+
"grad_norm": 14.63242244720459,
|
| 1246 |
+
"learning_rate": 7.340153452685422e-06,
|
| 1247 |
+
"loss": 0.3235,
|
| 1248 |
"step": 1490
|
| 1249 |
},
|
| 1250 |
{
|
| 1251 |
+
"epoch": 2.1802325581395348,
|
| 1252 |
+
"grad_norm": 10.674861907958984,
|
| 1253 |
+
"learning_rate": 7.21227621483376e-06,
|
| 1254 |
+
"loss": 0.39,
|
| 1255 |
"step": 1500
|
| 1256 |
},
|
| 1257 |
{
|
| 1258 |
+
"epoch": 2.1947674418604652,
|
| 1259 |
+
"grad_norm": 21.4650936126709,
|
| 1260 |
+
"learning_rate": 7.084398976982097e-06,
|
| 1261 |
+
"loss": 0.3526,
|
| 1262 |
"step": 1510
|
| 1263 |
},
|
| 1264 |
{
|
| 1265 |
+
"epoch": 2.2093023255813953,
|
| 1266 |
+
"grad_norm": 9.0557222366333,
|
| 1267 |
+
"learning_rate": 6.956521739130435e-06,
|
| 1268 |
+
"loss": 0.3315,
|
| 1269 |
"step": 1520
|
| 1270 |
},
|
| 1271 |
{
|
| 1272 |
+
"epoch": 2.2238372093023258,
|
| 1273 |
+
"grad_norm": 9.88924789428711,
|
| 1274 |
+
"learning_rate": 6.828644501278773e-06,
|
| 1275 |
+
"loss": 0.342,
|
| 1276 |
"step": 1530
|
| 1277 |
},
|
| 1278 |
{
|
| 1279 |
+
"epoch": 2.238372093023256,
|
| 1280 |
+
"grad_norm": 10.470682144165039,
|
| 1281 |
+
"learning_rate": 6.70076726342711e-06,
|
| 1282 |
+
"loss": 0.3666,
|
| 1283 |
"step": 1540
|
| 1284 |
},
|
| 1285 |
{
|
| 1286 |
+
"epoch": 2.2529069767441863,
|
| 1287 |
+
"grad_norm": 10.087296485900879,
|
| 1288 |
+
"learning_rate": 6.572890025575448e-06,
|
| 1289 |
+
"loss": 0.3509,
|
| 1290 |
"step": 1550
|
| 1291 |
},
|
| 1292 |
{
|
| 1293 |
+
"epoch": 2.2674418604651163,
|
| 1294 |
+
"grad_norm": 9.986245155334473,
|
| 1295 |
+
"learning_rate": 6.4450127877237854e-06,
|
| 1296 |
+
"loss": 0.2961,
|
| 1297 |
"step": 1560
|
| 1298 |
},
|
| 1299 |
{
|
| 1300 |
+
"epoch": 2.2819767441860463,
|
| 1301 |
+
"grad_norm": 4.27475643157959,
|
| 1302 |
+
"learning_rate": 6.317135549872123e-06,
|
| 1303 |
+
"loss": 0.3514,
|
| 1304 |
"step": 1570
|
| 1305 |
},
|
| 1306 |
{
|
| 1307 |
+
"epoch": 2.296511627906977,
|
| 1308 |
+
"grad_norm": 18.000774383544922,
|
| 1309 |
+
"learning_rate": 6.189258312020461e-06,
|
| 1310 |
+
"loss": 0.2998,
|
| 1311 |
"step": 1580
|
| 1312 |
},
|
| 1313 |
{
|
| 1314 |
+
"epoch": 2.311046511627907,
|
| 1315 |
+
"grad_norm": 11.935790061950684,
|
| 1316 |
+
"learning_rate": 6.061381074168798e-06,
|
| 1317 |
+
"loss": 0.3849,
|
| 1318 |
"step": 1590
|
| 1319 |
},
|
| 1320 |
{
|
| 1321 |
+
"epoch": 2.3255813953488373,
|
| 1322 |
+
"grad_norm": 6.844956874847412,
|
| 1323 |
+
"learning_rate": 5.933503836317136e-06,
|
| 1324 |
+
"loss": 0.3418,
|
| 1325 |
"step": 1600
|
| 1326 |
},
|
| 1327 |
{
|
| 1328 |
+
"epoch": 2.3255813953488373,
|
| 1329 |
+
"eval_accuracy": 0.74375,
|
| 1330 |
+
"eval_accuracy_label_arts, culture, entertainment and media": 0.7894736842105263,
|
| 1331 |
+
"eval_accuracy_label_conflict, war and peace": 0.7350689127105666,
|
| 1332 |
+
"eval_accuracy_label_crime, law and justice": 0.9072164948453608,
|
| 1333 |
+
"eval_accuracy_label_disaster, accident, and emergency incident": 0.8493150684931506,
|
| 1334 |
+
"eval_accuracy_label_economy, business, and finance": 0.7864077669902912,
|
| 1335 |
+
"eval_accuracy_label_environment": 0.4482758620689655,
|
| 1336 |
+
"eval_accuracy_label_health": 0.7777777777777778,
|
| 1337 |
+
"eval_accuracy_label_human interest": 0.3181818181818182,
|
| 1338 |
"eval_accuracy_label_labour": 0.5,
|
| 1339 |
+
"eval_accuracy_label_lifestyle and leisure": 0.5555555555555556,
|
| 1340 |
+
"eval_accuracy_label_politics": 0.7909407665505227,
|
| 1341 |
"eval_accuracy_label_religion": 0.0,
|
| 1342 |
"eval_accuracy_label_science and technology": 0.4166666666666667,
|
| 1343 |
+
"eval_accuracy_label_society": 0.18461538461538463,
|
| 1344 |
"eval_accuracy_label_sport": 0.9615384615384616,
|
| 1345 |
+
"eval_accuracy_label_weather": 0.0,
|
| 1346 |
+
"eval_f1": 0.7440049917955611,
|
| 1347 |
+
"eval_loss": 0.804217517375946,
|
| 1348 |
+
"eval_precision": 0.768551401471047,
|
| 1349 |
+
"eval_recall": 0.74375,
|
| 1350 |
+
"eval_runtime": 12.6056,
|
| 1351 |
+
"eval_samples_per_second": 126.928,
|
| 1352 |
+
"eval_steps_per_second": 7.933,
|
| 1353 |
"step": 1600
|
| 1354 |
},
|
| 1355 |
{
|
| 1356 |
+
"epoch": 2.3401162790697674,
|
| 1357 |
+
"grad_norm": 14.338434219360352,
|
| 1358 |
+
"learning_rate": 5.8056265984654735e-06,
|
| 1359 |
+
"loss": 0.2883,
|
| 1360 |
"step": 1610
|
| 1361 |
},
|
| 1362 |
{
|
| 1363 |
+
"epoch": 2.354651162790698,
|
| 1364 |
+
"grad_norm": 5.656998634338379,
|
| 1365 |
+
"learning_rate": 5.677749360613811e-06,
|
| 1366 |
+
"loss": 0.2737,
|
| 1367 |
"step": 1620
|
| 1368 |
},
|
| 1369 |
{
|
| 1370 |
+
"epoch": 2.369186046511628,
|
| 1371 |
+
"grad_norm": 6.466466903686523,
|
| 1372 |
+
"learning_rate": 5.549872122762149e-06,
|
| 1373 |
+
"loss": 0.331,
|
| 1374 |
"step": 1630
|
| 1375 |
},
|
| 1376 |
{
|
| 1377 |
+
"epoch": 2.383720930232558,
|
| 1378 |
+
"grad_norm": 11.816296577453613,
|
| 1379 |
+
"learning_rate": 5.421994884910486e-06,
|
| 1380 |
+
"loss": 0.3733,
|
| 1381 |
"step": 1640
|
| 1382 |
},
|
| 1383 |
{
|
| 1384 |
+
"epoch": 2.3982558139534884,
|
| 1385 |
+
"grad_norm": 8.958051681518555,
|
| 1386 |
+
"learning_rate": 5.294117647058824e-06,
|
| 1387 |
+
"loss": 0.2802,
|
| 1388 |
"step": 1650
|
| 1389 |
},
|
| 1390 |
{
|
| 1391 |
+
"epoch": 2.4127906976744184,
|
| 1392 |
+
"grad_norm": 13.985147476196289,
|
| 1393 |
+
"learning_rate": 5.1662404092071615e-06,
|
| 1394 |
+
"loss": 0.353,
|
| 1395 |
"step": 1660
|
| 1396 |
},
|
| 1397 |
{
|
| 1398 |
+
"epoch": 2.427325581395349,
|
| 1399 |
+
"grad_norm": 17.26644515991211,
|
| 1400 |
+
"learning_rate": 5.038363171355499e-06,
|
| 1401 |
+
"loss": 0.3399,
|
| 1402 |
"step": 1670
|
| 1403 |
},
|
| 1404 |
{
|
| 1405 |
+
"epoch": 2.441860465116279,
|
| 1406 |
+
"grad_norm": 13.139063835144043,
|
| 1407 |
+
"learning_rate": 4.910485933503837e-06,
|
| 1408 |
+
"loss": 0.2949,
|
| 1409 |
"step": 1680
|
| 1410 |
},
|
| 1411 |
{
|
| 1412 |
+
"epoch": 2.4563953488372094,
|
| 1413 |
+
"grad_norm": 15.950389862060547,
|
| 1414 |
+
"learning_rate": 4.782608695652174e-06,
|
| 1415 |
+
"loss": 0.3742,
|
| 1416 |
"step": 1690
|
| 1417 |
},
|
| 1418 |
{
|
| 1419 |
+
"epoch": 2.4709302325581395,
|
| 1420 |
+
"grad_norm": 8.10091495513916,
|
| 1421 |
+
"learning_rate": 4.654731457800512e-06,
|
| 1422 |
+
"loss": 0.2605,
|
| 1423 |
"step": 1700
|
| 1424 |
},
|
| 1425 |
{
|
| 1426 |
+
"epoch": 2.4854651162790695,
|
| 1427 |
+
"grad_norm": 8.73356819152832,
|
| 1428 |
+
"learning_rate": 4.5268542199488496e-06,
|
| 1429 |
+
"loss": 0.2077,
|
| 1430 |
"step": 1710
|
| 1431 |
},
|
| 1432 |
{
|
| 1433 |
+
"epoch": 2.5,
|
| 1434 |
+
"grad_norm": 7.236685276031494,
|
| 1435 |
+
"learning_rate": 4.398976982097187e-06,
|
| 1436 |
+
"loss": 0.3526,
|
| 1437 |
"step": 1720
|
| 1438 |
},
|
| 1439 |
{
|
| 1440 |
+
"epoch": 2.5145348837209305,
|
| 1441 |
+
"grad_norm": 6.902716159820557,
|
| 1442 |
+
"learning_rate": 4.271099744245525e-06,
|
| 1443 |
+
"loss": 0.2633,
|
| 1444 |
"step": 1730
|
| 1445 |
},
|
| 1446 |
{
|
| 1447 |
+
"epoch": 2.5290697674418605,
|
| 1448 |
+
"grad_norm": 4.061647891998291,
|
| 1449 |
+
"learning_rate": 4.143222506393862e-06,
|
| 1450 |
+
"loss": 0.2682,
|
| 1451 |
"step": 1740
|
| 1452 |
},
|
| 1453 |
{
|
| 1454 |
+
"epoch": 2.5436046511627906,
|
| 1455 |
+
"grad_norm": 3.789299488067627,
|
| 1456 |
+
"learning_rate": 4.0153452685422e-06,
|
| 1457 |
+
"loss": 0.2996,
|
| 1458 |
"step": 1750
|
| 1459 |
},
|
| 1460 |
{
|
| 1461 |
+
"epoch": 2.558139534883721,
|
| 1462 |
+
"grad_norm": 7.185976505279541,
|
| 1463 |
+
"learning_rate": 3.887468030690537e-06,
|
| 1464 |
+
"loss": 0.323,
|
| 1465 |
"step": 1760
|
| 1466 |
},
|
| 1467 |
{
|
| 1468 |
+
"epoch": 2.572674418604651,
|
| 1469 |
+
"grad_norm": 9.113693237304688,
|
| 1470 |
+
"learning_rate": 3.7595907928388748e-06,
|
| 1471 |
+
"loss": 0.2854,
|
| 1472 |
"step": 1770
|
| 1473 |
},
|
| 1474 |
+
{
|
| 1475 |
+
"epoch": 2.5872093023255816,
|
| 1476 |
+
"grad_norm": 10.132606506347656,
|
| 1477 |
+
"learning_rate": 3.6317135549872124e-06,
|
| 1478 |
+
"loss": 0.351,
|
| 1479 |
+
"step": 1780
|
| 1480 |
+
},
|
| 1481 |
+
{
|
| 1482 |
+
"epoch": 2.6017441860465116,
|
| 1483 |
+
"grad_norm": 12.435748100280762,
|
| 1484 |
+
"learning_rate": 3.50383631713555e-06,
|
| 1485 |
+
"loss": 0.3271,
|
| 1486 |
+
"step": 1790
|
| 1487 |
+
},
|
| 1488 |
+
{
|
| 1489 |
+
"epoch": 2.616279069767442,
|
| 1490 |
+
"grad_norm": 15.90312671661377,
|
| 1491 |
+
"learning_rate": 3.3759590792838876e-06,
|
| 1492 |
+
"loss": 0.248,
|
| 1493 |
+
"step": 1800
|
| 1494 |
+
},
|
| 1495 |
+
{
|
| 1496 |
+
"epoch": 2.616279069767442,
|
| 1497 |
+
"eval_accuracy": 0.7275,
|
| 1498 |
+
"eval_accuracy_label_arts, culture, entertainment and media": 0.6842105263157895,
|
| 1499 |
+
"eval_accuracy_label_conflict, war and peace": 0.6891271056661562,
|
| 1500 |
+
"eval_accuracy_label_crime, law and justice": 0.8814432989690721,
|
| 1501 |
+
"eval_accuracy_label_disaster, accident, and emergency incident": 0.8698630136986302,
|
| 1502 |
+
"eval_accuracy_label_economy, business, and finance": 0.7572815533980582,
|
| 1503 |
+
"eval_accuracy_label_environment": 0.41379310344827586,
|
| 1504 |
+
"eval_accuracy_label_health": 0.8333333333333334,
|
| 1505 |
+
"eval_accuracy_label_human interest": 0.4090909090909091,
|
| 1506 |
+
"eval_accuracy_label_labour": 0.5,
|
| 1507 |
+
"eval_accuracy_label_lifestyle and leisure": 0.5555555555555556,
|
| 1508 |
+
"eval_accuracy_label_politics": 0.8013937282229965,
|
| 1509 |
+
"eval_accuracy_label_religion": 0.0,
|
| 1510 |
+
"eval_accuracy_label_science and technology": 0.4166666666666667,
|
| 1511 |
+
"eval_accuracy_label_society": 0.27692307692307694,
|
| 1512 |
+
"eval_accuracy_label_sport": 0.9615384615384616,
|
| 1513 |
+
"eval_accuracy_label_weather": 0.0,
|
| 1514 |
+
"eval_f1": 0.7324830328014649,
|
| 1515 |
+
"eval_loss": 0.8387188911437988,
|
| 1516 |
+
"eval_precision": 0.7610492043473545,
|
| 1517 |
+
"eval_recall": 0.7275,
|
| 1518 |
+
"eval_runtime": 12.5904,
|
| 1519 |
+
"eval_samples_per_second": 127.081,
|
| 1520 |
+
"eval_steps_per_second": 7.943,
|
| 1521 |
+
"step": 1800
|
| 1522 |
+
},
|
| 1523 |
+
{
|
| 1524 |
+
"epoch": 2.630813953488372,
|
| 1525 |
+
"grad_norm": 14.707446098327637,
|
| 1526 |
+
"learning_rate": 3.248081841432225e-06,
|
| 1527 |
+
"loss": 0.3358,
|
| 1528 |
+
"step": 1810
|
| 1529 |
+
},
|
| 1530 |
+
{
|
| 1531 |
+
"epoch": 2.645348837209302,
|
| 1532 |
+
"grad_norm": 8.684798240661621,
|
| 1533 |
+
"learning_rate": 3.120204603580563e-06,
|
| 1534 |
+
"loss": 0.3082,
|
| 1535 |
+
"step": 1820
|
| 1536 |
+
},
|
| 1537 |
+
{
|
| 1538 |
+
"epoch": 2.6598837209302326,
|
| 1539 |
+
"grad_norm": 8.125092506408691,
|
| 1540 |
+
"learning_rate": 2.9923273657289004e-06,
|
| 1541 |
+
"loss": 0.2818,
|
| 1542 |
+
"step": 1830
|
| 1543 |
+
},
|
| 1544 |
+
{
|
| 1545 |
+
"epoch": 2.6744186046511627,
|
| 1546 |
+
"grad_norm": 10.351339340209961,
|
| 1547 |
+
"learning_rate": 2.864450127877238e-06,
|
| 1548 |
+
"loss": 0.3465,
|
| 1549 |
+
"step": 1840
|
| 1550 |
+
},
|
| 1551 |
+
{
|
| 1552 |
+
"epoch": 2.688953488372093,
|
| 1553 |
+
"grad_norm": 9.536513328552246,
|
| 1554 |
+
"learning_rate": 2.7365728900255756e-06,
|
| 1555 |
+
"loss": 0.3022,
|
| 1556 |
+
"step": 1850
|
| 1557 |
+
},
|
| 1558 |
+
{
|
| 1559 |
+
"epoch": 2.703488372093023,
|
| 1560 |
+
"grad_norm": 7.122174263000488,
|
| 1561 |
+
"learning_rate": 2.6086956521739132e-06,
|
| 1562 |
+
"loss": 0.3742,
|
| 1563 |
+
"step": 1860
|
| 1564 |
+
},
|
| 1565 |
+
{
|
| 1566 |
+
"epoch": 2.7180232558139537,
|
| 1567 |
+
"grad_norm": 11.325377464294434,
|
| 1568 |
+
"learning_rate": 2.480818414322251e-06,
|
| 1569 |
+
"loss": 0.2912,
|
| 1570 |
+
"step": 1870
|
| 1571 |
+
},
|
| 1572 |
+
{
|
| 1573 |
+
"epoch": 2.7325581395348837,
|
| 1574 |
+
"grad_norm": 9.651542663574219,
|
| 1575 |
+
"learning_rate": 2.3529411764705885e-06,
|
| 1576 |
+
"loss": 0.3156,
|
| 1577 |
+
"step": 1880
|
| 1578 |
+
},
|
| 1579 |
+
{
|
| 1580 |
+
"epoch": 2.7470930232558137,
|
| 1581 |
+
"grad_norm": 16.223478317260742,
|
| 1582 |
+
"learning_rate": 2.225063938618926e-06,
|
| 1583 |
+
"loss": 0.3405,
|
| 1584 |
+
"step": 1890
|
| 1585 |
+
},
|
| 1586 |
+
{
|
| 1587 |
+
"epoch": 2.761627906976744,
|
| 1588 |
+
"grad_norm": 6.137476921081543,
|
| 1589 |
+
"learning_rate": 2.0971867007672637e-06,
|
| 1590 |
+
"loss": 0.1922,
|
| 1591 |
+
"step": 1900
|
| 1592 |
+
},
|
| 1593 |
+
{
|
| 1594 |
+
"epoch": 2.7761627906976747,
|
| 1595 |
+
"grad_norm": 9.50839614868164,
|
| 1596 |
+
"learning_rate": 1.9693094629156013e-06,
|
| 1597 |
+
"loss": 0.2896,
|
| 1598 |
+
"step": 1910
|
| 1599 |
+
},
|
| 1600 |
+
{
|
| 1601 |
+
"epoch": 2.7906976744186047,
|
| 1602 |
+
"grad_norm": 8.988372802734375,
|
| 1603 |
+
"learning_rate": 1.8414322250639389e-06,
|
| 1604 |
+
"loss": 0.3286,
|
| 1605 |
+
"step": 1920
|
| 1606 |
+
},
|
| 1607 |
+
{
|
| 1608 |
+
"epoch": 2.8052325581395348,
|
| 1609 |
+
"grad_norm": 9.569466590881348,
|
| 1610 |
+
"learning_rate": 1.7135549872122765e-06,
|
| 1611 |
+
"loss": 0.3463,
|
| 1612 |
+
"step": 1930
|
| 1613 |
+
},
|
| 1614 |
+
{
|
| 1615 |
+
"epoch": 2.8197674418604652,
|
| 1616 |
+
"grad_norm": 7.808605194091797,
|
| 1617 |
+
"learning_rate": 1.585677749360614e-06,
|
| 1618 |
+
"loss": 0.3117,
|
| 1619 |
+
"step": 1940
|
| 1620 |
+
},
|
| 1621 |
+
{
|
| 1622 |
+
"epoch": 2.8343023255813953,
|
| 1623 |
+
"grad_norm": 9.752486228942871,
|
| 1624 |
+
"learning_rate": 1.4578005115089517e-06,
|
| 1625 |
+
"loss": 0.3708,
|
| 1626 |
+
"step": 1950
|
| 1627 |
+
},
|
| 1628 |
+
{
|
| 1629 |
+
"epoch": 2.8488372093023253,
|
| 1630 |
+
"grad_norm": 6.103060722351074,
|
| 1631 |
+
"learning_rate": 1.3299232736572893e-06,
|
| 1632 |
+
"loss": 0.2766,
|
| 1633 |
+
"step": 1960
|
| 1634 |
+
},
|
| 1635 |
+
{
|
| 1636 |
+
"epoch": 2.863372093023256,
|
| 1637 |
+
"grad_norm": 8.869214057922363,
|
| 1638 |
+
"learning_rate": 1.2020460358056267e-06,
|
| 1639 |
+
"loss": 0.2762,
|
| 1640 |
+
"step": 1970
|
| 1641 |
+
},
|
| 1642 |
+
{
|
| 1643 |
+
"epoch": 2.8779069767441863,
|
| 1644 |
+
"grad_norm": 8.96507740020752,
|
| 1645 |
+
"learning_rate": 1.0741687979539643e-06,
|
| 1646 |
+
"loss": 0.2874,
|
| 1647 |
+
"step": 1980
|
| 1648 |
+
},
|
| 1649 |
+
{
|
| 1650 |
+
"epoch": 2.8924418604651163,
|
| 1651 |
+
"grad_norm": 8.362478256225586,
|
| 1652 |
+
"learning_rate": 9.462915601023019e-07,
|
| 1653 |
+
"loss": 0.2944,
|
| 1654 |
+
"step": 1990
|
| 1655 |
+
},
|
| 1656 |
+
{
|
| 1657 |
+
"epoch": 2.9069767441860463,
|
| 1658 |
+
"grad_norm": 7.523708820343018,
|
| 1659 |
+
"learning_rate": 8.184143222506395e-07,
|
| 1660 |
+
"loss": 0.2525,
|
| 1661 |
+
"step": 2000
|
| 1662 |
+
},
|
| 1663 |
+
{
|
| 1664 |
+
"epoch": 2.9069767441860463,
|
| 1665 |
+
"eval_accuracy": 0.735,
|
| 1666 |
+
"eval_accuracy_label_arts, culture, entertainment and media": 0.6842105263157895,
|
| 1667 |
+
"eval_accuracy_label_conflict, war and peace": 0.7105666156202144,
|
| 1668 |
+
"eval_accuracy_label_crime, law and justice": 0.8762886597938144,
|
| 1669 |
+
"eval_accuracy_label_disaster, accident, and emergency incident": 0.8698630136986302,
|
| 1670 |
+
"eval_accuracy_label_economy, business, and finance": 0.6796116504854369,
|
| 1671 |
+
"eval_accuracy_label_environment": 0.4482758620689655,
|
| 1672 |
+
"eval_accuracy_label_health": 0.7222222222222222,
|
| 1673 |
+
"eval_accuracy_label_human interest": 0.36363636363636365,
|
| 1674 |
+
"eval_accuracy_label_labour": 0.5,
|
| 1675 |
+
"eval_accuracy_label_lifestyle and leisure": 0.5555555555555556,
|
| 1676 |
+
"eval_accuracy_label_politics": 0.8153310104529616,
|
| 1677 |
+
"eval_accuracy_label_religion": 0.0,
|
| 1678 |
+
"eval_accuracy_label_science and technology": 0.4583333333333333,
|
| 1679 |
+
"eval_accuracy_label_society": 0.3384615384615385,
|
| 1680 |
+
"eval_accuracy_label_sport": 0.9615384615384616,
|
| 1681 |
+
"eval_accuracy_label_weather": 0.0,
|
| 1682 |
+
"eval_f1": 0.7412823760546817,
|
| 1683 |
+
"eval_loss": 0.8137220740318298,
|
| 1684 |
+
"eval_precision": 0.7696596009813345,
|
| 1685 |
+
"eval_recall": 0.735,
|
| 1686 |
+
"eval_runtime": 12.5933,
|
| 1687 |
+
"eval_samples_per_second": 127.051,
|
| 1688 |
+
"eval_steps_per_second": 7.941,
|
| 1689 |
+
"step": 2000
|
| 1690 |
+
},
|
| 1691 |
+
{
|
| 1692 |
+
"epoch": 2.921511627906977,
|
| 1693 |
+
"grad_norm": 9.854249954223633,
|
| 1694 |
+
"learning_rate": 6.90537084398977e-07,
|
| 1695 |
+
"loss": 0.2625,
|
| 1696 |
+
"step": 2010
|
| 1697 |
+
},
|
| 1698 |
+
{
|
| 1699 |
+
"epoch": 2.936046511627907,
|
| 1700 |
+
"grad_norm": 7.306544780731201,
|
| 1701 |
+
"learning_rate": 5.626598465473146e-07,
|
| 1702 |
+
"loss": 0.3285,
|
| 1703 |
+
"step": 2020
|
| 1704 |
+
},
|
| 1705 |
+
{
|
| 1706 |
+
"epoch": 2.9505813953488373,
|
| 1707 |
+
"grad_norm": 4.267274856567383,
|
| 1708 |
+
"learning_rate": 4.347826086956522e-07,
|
| 1709 |
+
"loss": 0.3445,
|
| 1710 |
+
"step": 2030
|
| 1711 |
+
},
|
| 1712 |
+
{
|
| 1713 |
+
"epoch": 2.9651162790697674,
|
| 1714 |
+
"grad_norm": 15.480578422546387,
|
| 1715 |
+
"learning_rate": 3.069053708439898e-07,
|
| 1716 |
+
"loss": 0.3066,
|
| 1717 |
+
"step": 2040
|
| 1718 |
+
},
|
| 1719 |
+
{
|
| 1720 |
+
"epoch": 2.979651162790698,
|
| 1721 |
+
"grad_norm": 11.125273704528809,
|
| 1722 |
+
"learning_rate": 1.7902813299232738e-07,
|
| 1723 |
+
"loss": 0.3026,
|
| 1724 |
+
"step": 2050
|
| 1725 |
+
},
|
| 1726 |
+
{
|
| 1727 |
+
"epoch": 2.994186046511628,
|
| 1728 |
+
"grad_norm": 12.110825538635254,
|
| 1729 |
+
"learning_rate": 5.115089514066497e-08,
|
| 1730 |
+
"loss": 0.2815,
|
| 1731 |
+
"step": 2060
|
| 1732 |
+
},
|
| 1733 |
{
|
| 1734 |
"epoch": 3.0,
|
| 1735 |
+
"step": 2064,
|
| 1736 |
+
"total_flos": 8687308686630912.0,
|
| 1737 |
+
"train_loss": 0.8068887479603291,
|
| 1738 |
+
"train_runtime": 1561.1428,
|
| 1739 |
+
"train_samples_per_second": 42.294,
|
| 1740 |
+
"train_steps_per_second": 1.322
|
| 1741 |
},
|
| 1742 |
{
|
| 1743 |
"epoch": 3.0,
|
| 1744 |
+
"eval_accuracy": 0.743125,
|
| 1745 |
+
"eval_accuracy_label_arts, culture, entertainment and media": 0.6842105263157895,
|
| 1746 |
+
"eval_accuracy_label_conflict, war and peace": 0.7350689127105666,
|
| 1747 |
+
"eval_accuracy_label_crime, law and justice": 0.8917525773195877,
|
| 1748 |
+
"eval_accuracy_label_disaster, accident, and emergency incident": 0.8698630136986302,
|
| 1749 |
+
"eval_accuracy_label_economy, business, and finance": 0.6893203883495146,
|
| 1750 |
+
"eval_accuracy_label_environment": 0.4482758620689655,
|
| 1751 |
+
"eval_accuracy_label_health": 0.7222222222222222,
|
| 1752 |
+
"eval_accuracy_label_human interest": 0.3181818181818182,
|
| 1753 |
"eval_accuracy_label_labour": 0.5,
|
| 1754 |
+
"eval_accuracy_label_lifestyle and leisure": 0.5555555555555556,
|
| 1755 |
+
"eval_accuracy_label_politics": 0.7909407665505227,
|
| 1756 |
"eval_accuracy_label_religion": 0.0,
|
| 1757 |
+
"eval_accuracy_label_science and technology": 0.4583333333333333,
|
| 1758 |
+
"eval_accuracy_label_society": 0.35384615384615387,
|
| 1759 |
"eval_accuracy_label_sport": 0.9615384615384616,
|
| 1760 |
+
"eval_accuracy_label_weather": 0.0,
|
| 1761 |
+
"eval_f1": 0.747448709907019,
|
| 1762 |
+
"eval_loss": 0.8030127882957458,
|
| 1763 |
+
"eval_precision": 0.7695117491320402,
|
| 1764 |
+
"eval_recall": 0.743125,
|
| 1765 |
+
"eval_runtime": 12.5023,
|
| 1766 |
+
"eval_samples_per_second": 127.976,
|
| 1767 |
+
"eval_steps_per_second": 7.999,
|
| 1768 |
+
"step": 2064
|
| 1769 |
}
|
| 1770 |
],
|
| 1771 |
"logging_steps": 10,
|
| 1772 |
+
"max_steps": 2064,
|
| 1773 |
"num_input_tokens_seen": 0,
|
| 1774 |
"num_train_epochs": 3,
|
| 1775 |
"save_steps": 1000,
|
| 1776 |
+
"total_flos": 8687308686630912.0,
|
| 1777 |
"train_batch_size": 16,
|
| 1778 |
"trial_name": null,
|
| 1779 |
"trial_params": null
|
training_args.bin
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 4984
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:731425a181dc26a314ae5d1639930934164b694586b113e63d14cc1c77f6c544
|
| 3 |
size 4984
|