Masaki Hattori
commited on
Commit
•
992bfd6
1
Parent(s):
8691953
Model save
Browse files- README.md +73 -0
- trainer_state.json +608 -0
README.md
ADDED
|
@@ -0,0 +1,73 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
license: apache-2.0
|
| 3 |
+
base_model: rinna/japanese-hubert-base
|
| 4 |
+
tags:
|
| 5 |
+
- generated_from_trainer
|
| 6 |
+
metrics:
|
| 7 |
+
- wer
|
| 8 |
+
model-index:
|
| 9 |
+
- name: hubert-rinnna-jp-jdrtsp-fw07sp-12
|
| 10 |
+
results: []
|
| 11 |
+
---
|
| 12 |
+
|
| 13 |
+
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
|
| 14 |
+
should probably proofread and complete it, then remove this comment. -->
|
| 15 |
+
|
| 16 |
+
# hubert-rinnna-jp-jdrtsp-fw07sp-12
|
| 17 |
+
|
| 18 |
+
This model is a fine-tuned version of [rinna/japanese-hubert-base](https://huggingface.co/rinna/japanese-hubert-base) on the None dataset.
|
| 19 |
+
It achieves the following results on the evaluation set:
|
| 20 |
+
- Loss: 1.1989
|
| 21 |
+
- Wer: 0.6801
|
| 22 |
+
- Cer: 0.5794
|
| 23 |
+
|
| 24 |
+
## Model description
|
| 25 |
+
|
| 26 |
+
More information needed
|
| 27 |
+
|
| 28 |
+
## Intended uses & limitations
|
| 29 |
+
|
| 30 |
+
More information needed
|
| 31 |
+
|
| 32 |
+
## Training and evaluation data
|
| 33 |
+
|
| 34 |
+
More information needed
|
| 35 |
+
|
| 36 |
+
## Training procedure
|
| 37 |
+
|
| 38 |
+
### Training hyperparameters
|
| 39 |
+
|
| 40 |
+
The following hyperparameters were used during training:
|
| 41 |
+
- learning_rate: 1e-05
|
| 42 |
+
- train_batch_size: 32
|
| 43 |
+
- eval_batch_size: 16
|
| 44 |
+
- seed: 42
|
| 45 |
+
- gradient_accumulation_steps: 2
|
| 46 |
+
- total_train_batch_size: 64
|
| 47 |
+
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
|
| 48 |
+
- lr_scheduler_type: linear
|
| 49 |
+
- lr_scheduler_warmup_steps: 1000
|
| 50 |
+
- num_epochs: 10
|
| 51 |
+
|
| 52 |
+
### Training results
|
| 53 |
+
|
| 54 |
+
| Training Loss | Epoch | Step | Validation Loss | Wer | Cer |
|
| 55 |
+
|:-------------:|:-----:|:----:|:---------------:|:------:|:------:|
|
| 56 |
+
| 5.0318 | 1.0 | 404 | 4.2999 | 0.9798 | 0.9889 |
|
| 57 |
+
| 3.5113 | 2.0 | 808 | 3.3289 | 0.9798 | 0.9889 |
|
| 58 |
+
| 2.7536 | 3.0 | 1212 | 2.7007 | 0.9798 | 0.9889 |
|
| 59 |
+
| 2.4826 | 4.0 | 1616 | 2.3732 | 0.9798 | 0.9889 |
|
| 60 |
+
| 2.0642 | 5.0 | 2020 | 1.9165 | 0.9798 | 0.9888 |
|
| 61 |
+
| 1.834 | 6.0 | 2424 | 1.6739 | 0.9504 | 0.9464 |
|
| 62 |
+
| 1.6869 | 7.0 | 2828 | 1.4651 | 0.8239 | 0.7865 |
|
| 63 |
+
| 1.5734 | 8.0 | 3232 | 1.3267 | 0.7440 | 0.6939 |
|
| 64 |
+
| 1.5052 | 9.0 | 3636 | 1.2331 | 0.7045 | 0.6231 |
|
| 65 |
+
| 1.4573 | 10.0 | 4040 | 1.1989 | 0.6801 | 0.5794 |
|
| 66 |
+
|
| 67 |
+
|
| 68 |
+
### Framework versions
|
| 69 |
+
|
| 70 |
+
- Transformers 4.34.0.dev0
|
| 71 |
+
- Pytorch 2.0.1+cu118
|
| 72 |
+
- Datasets 2.14.5
|
| 73 |
+
- Tokenizers 0.13.3
|
trainer_state.json
ADDED
|
@@ -0,0 +1,608 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 10.0,
|
| 5 |
+
"eval_steps": 500,
|
| 6 |
+
"global_step": 4040,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.12,
|
| 13 |
+
"learning_rate": 5.000000000000001e-07,
|
| 14 |
+
"loss": 10.9669,
|
| 15 |
+
"step": 50
|
| 16 |
+
},
|
| 17 |
+
{
|
| 18 |
+
"epoch": 0.25,
|
| 19 |
+
"learning_rate": 1.0000000000000002e-06,
|
| 20 |
+
"loss": 10.7574,
|
| 21 |
+
"step": 100
|
| 22 |
+
},
|
| 23 |
+
{
|
| 24 |
+
"epoch": 0.37,
|
| 25 |
+
"learning_rate": 1.5e-06,
|
| 26 |
+
"loss": 10.4847,
|
| 27 |
+
"step": 150
|
| 28 |
+
},
|
| 29 |
+
{
|
| 30 |
+
"epoch": 0.5,
|
| 31 |
+
"learning_rate": 2.0000000000000003e-06,
|
| 32 |
+
"loss": 9.3795,
|
| 33 |
+
"step": 200
|
| 34 |
+
},
|
| 35 |
+
{
|
| 36 |
+
"epoch": 0.62,
|
| 37 |
+
"learning_rate": 2.5e-06,
|
| 38 |
+
"loss": 7.4952,
|
| 39 |
+
"step": 250
|
| 40 |
+
},
|
| 41 |
+
{
|
| 42 |
+
"epoch": 0.74,
|
| 43 |
+
"learning_rate": 3e-06,
|
| 44 |
+
"loss": 6.3533,
|
| 45 |
+
"step": 300
|
| 46 |
+
},
|
| 47 |
+
{
|
| 48 |
+
"epoch": 0.87,
|
| 49 |
+
"learning_rate": 3.5e-06,
|
| 50 |
+
"loss": 5.356,
|
| 51 |
+
"step": 350
|
| 52 |
+
},
|
| 53 |
+
{
|
| 54 |
+
"epoch": 0.99,
|
| 55 |
+
"learning_rate": 4.000000000000001e-06,
|
| 56 |
+
"loss": 5.0318,
|
| 57 |
+
"step": 400
|
| 58 |
+
},
|
| 59 |
+
{
|
| 60 |
+
"epoch": 1.0,
|
| 61 |
+
"eval_cer": 0.9888512643814729,
|
| 62 |
+
"eval_loss": 4.29986047744751,
|
| 63 |
+
"eval_runtime": 38.3788,
|
| 64 |
+
"eval_samples_per_second": 168.426,
|
| 65 |
+
"eval_steps_per_second": 10.527,
|
| 66 |
+
"eval_wer": 0.9797794777861581,
|
| 67 |
+
"step": 404
|
| 68 |
+
},
|
| 69 |
+
{
|
| 70 |
+
"epoch": 1.11,
|
| 71 |
+
"learning_rate": 4.5e-06,
|
| 72 |
+
"loss": 4.7526,
|
| 73 |
+
"step": 450
|
| 74 |
+
},
|
| 75 |
+
{
|
| 76 |
+
"epoch": 1.24,
|
| 77 |
+
"learning_rate": 5e-06,
|
| 78 |
+
"loss": 4.5027,
|
| 79 |
+
"step": 500
|
| 80 |
+
},
|
| 81 |
+
{
|
| 82 |
+
"epoch": 1.36,
|
| 83 |
+
"learning_rate": 5.500000000000001e-06,
|
| 84 |
+
"loss": 4.3223,
|
| 85 |
+
"step": 550
|
| 86 |
+
},
|
| 87 |
+
{
|
| 88 |
+
"epoch": 1.49,
|
| 89 |
+
"learning_rate": 6e-06,
|
| 90 |
+
"loss": 4.0849,
|
| 91 |
+
"step": 600
|
| 92 |
+
},
|
| 93 |
+
{
|
| 94 |
+
"epoch": 1.61,
|
| 95 |
+
"learning_rate": 6.5000000000000004e-06,
|
| 96 |
+
"loss": 3.9545,
|
| 97 |
+
"step": 650
|
| 98 |
+
},
|
| 99 |
+
{
|
| 100 |
+
"epoch": 1.73,
|
| 101 |
+
"learning_rate": 7e-06,
|
| 102 |
+
"loss": 3.7431,
|
| 103 |
+
"step": 700
|
| 104 |
+
},
|
| 105 |
+
{
|
| 106 |
+
"epoch": 1.86,
|
| 107 |
+
"learning_rate": 7.500000000000001e-06,
|
| 108 |
+
"loss": 3.6413,
|
| 109 |
+
"step": 750
|
| 110 |
+
},
|
| 111 |
+
{
|
| 112 |
+
"epoch": 1.98,
|
| 113 |
+
"learning_rate": 8.000000000000001e-06,
|
| 114 |
+
"loss": 3.5113,
|
| 115 |
+
"step": 800
|
| 116 |
+
},
|
| 117 |
+
{
|
| 118 |
+
"epoch": 2.0,
|
| 119 |
+
"eval_cer": 0.9888512643814729,
|
| 120 |
+
"eval_loss": 3.3288848400115967,
|
| 121 |
+
"eval_runtime": 39.0341,
|
| 122 |
+
"eval_samples_per_second": 165.599,
|
| 123 |
+
"eval_steps_per_second": 10.35,
|
| 124 |
+
"eval_wer": 0.9797794777861581,
|
| 125 |
+
"step": 808
|
| 126 |
+
},
|
| 127 |
+
{
|
| 128 |
+
"epoch": 2.1,
|
| 129 |
+
"learning_rate": 8.5e-06,
|
| 130 |
+
"loss": 3.3907,
|
| 131 |
+
"step": 850
|
| 132 |
+
},
|
| 133 |
+
{
|
| 134 |
+
"epoch": 2.23,
|
| 135 |
+
"learning_rate": 9e-06,
|
| 136 |
+
"loss": 3.2836,
|
| 137 |
+
"step": 900
|
| 138 |
+
},
|
| 139 |
+
{
|
| 140 |
+
"epoch": 2.35,
|
| 141 |
+
"learning_rate": 9.5e-06,
|
| 142 |
+
"loss": 3.1635,
|
| 143 |
+
"step": 950
|
| 144 |
+
},
|
| 145 |
+
{
|
| 146 |
+
"epoch": 2.48,
|
| 147 |
+
"learning_rate": 1e-05,
|
| 148 |
+
"loss": 3.0437,
|
| 149 |
+
"step": 1000
|
| 150 |
+
},
|
| 151 |
+
{
|
| 152 |
+
"epoch": 2.6,
|
| 153 |
+
"learning_rate": 9.835526315789474e-06,
|
| 154 |
+
"loss": 2.9567,
|
| 155 |
+
"step": 1050
|
| 156 |
+
},
|
| 157 |
+
{
|
| 158 |
+
"epoch": 2.72,
|
| 159 |
+
"learning_rate": 9.671052631578948e-06,
|
| 160 |
+
"loss": 2.8632,
|
| 161 |
+
"step": 1100
|
| 162 |
+
},
|
| 163 |
+
{
|
| 164 |
+
"epoch": 2.85,
|
| 165 |
+
"learning_rate": 9.506578947368423e-06,
|
| 166 |
+
"loss": 2.8078,
|
| 167 |
+
"step": 1150
|
| 168 |
+
},
|
| 169 |
+
{
|
| 170 |
+
"epoch": 2.97,
|
| 171 |
+
"learning_rate": 9.342105263157895e-06,
|
| 172 |
+
"loss": 2.7536,
|
| 173 |
+
"step": 1200
|
| 174 |
+
},
|
| 175 |
+
{
|
| 176 |
+
"epoch": 3.0,
|
| 177 |
+
"eval_cer": 0.9888512643814729,
|
| 178 |
+
"eval_loss": 2.700655460357666,
|
| 179 |
+
"eval_runtime": 38.3318,
|
| 180 |
+
"eval_samples_per_second": 168.633,
|
| 181 |
+
"eval_steps_per_second": 10.54,
|
| 182 |
+
"eval_wer": 0.9797794777861581,
|
| 183 |
+
"step": 1212
|
| 184 |
+
},
|
| 185 |
+
{
|
| 186 |
+
"epoch": 3.09,
|
| 187 |
+
"learning_rate": 9.17763157894737e-06,
|
| 188 |
+
"loss": 2.7127,
|
| 189 |
+
"step": 1250
|
| 190 |
+
},
|
| 191 |
+
{
|
| 192 |
+
"epoch": 3.22,
|
| 193 |
+
"learning_rate": 9.013157894736843e-06,
|
| 194 |
+
"loss": 2.6749,
|
| 195 |
+
"step": 1300
|
| 196 |
+
},
|
| 197 |
+
{
|
| 198 |
+
"epoch": 3.34,
|
| 199 |
+
"learning_rate": 8.848684210526316e-06,
|
| 200 |
+
"loss": 2.6572,
|
| 201 |
+
"step": 1350
|
| 202 |
+
},
|
| 203 |
+
{
|
| 204 |
+
"epoch": 3.47,
|
| 205 |
+
"learning_rate": 8.68421052631579e-06,
|
| 206 |
+
"loss": 2.6236,
|
| 207 |
+
"step": 1400
|
| 208 |
+
},
|
| 209 |
+
{
|
| 210 |
+
"epoch": 3.59,
|
| 211 |
+
"learning_rate": 8.519736842105265e-06,
|
| 212 |
+
"loss": 2.6007,
|
| 213 |
+
"step": 1450
|
| 214 |
+
},
|
| 215 |
+
{
|
| 216 |
+
"epoch": 3.71,
|
| 217 |
+
"learning_rate": 8.355263157894737e-06,
|
| 218 |
+
"loss": 2.5653,
|
| 219 |
+
"step": 1500
|
| 220 |
+
},
|
| 221 |
+
{
|
| 222 |
+
"epoch": 3.84,
|
| 223 |
+
"learning_rate": 8.19078947368421e-06,
|
| 224 |
+
"loss": 2.5333,
|
| 225 |
+
"step": 1550
|
| 226 |
+
},
|
| 227 |
+
{
|
| 228 |
+
"epoch": 3.96,
|
| 229 |
+
"learning_rate": 8.026315789473685e-06,
|
| 230 |
+
"loss": 2.4826,
|
| 231 |
+
"step": 1600
|
| 232 |
+
},
|
| 233 |
+
{
|
| 234 |
+
"epoch": 4.0,
|
| 235 |
+
"eval_cer": 0.9888512643814729,
|
| 236 |
+
"eval_loss": 2.373181104660034,
|
| 237 |
+
"eval_runtime": 37.8617,
|
| 238 |
+
"eval_samples_per_second": 170.727,
|
| 239 |
+
"eval_steps_per_second": 10.67,
|
| 240 |
+
"eval_wer": 0.9797794777861581,
|
| 241 |
+
"step": 1616
|
| 242 |
+
},
|
| 243 |
+
{
|
| 244 |
+
"epoch": 4.08,
|
| 245 |
+
"learning_rate": 7.86184210526316e-06,
|
| 246 |
+
"loss": 2.4176,
|
| 247 |
+
"step": 1650
|
| 248 |
+
},
|
| 249 |
+
{
|
| 250 |
+
"epoch": 4.21,
|
| 251 |
+
"learning_rate": 7.697368421052632e-06,
|
| 252 |
+
"loss": 2.371,
|
| 253 |
+
"step": 1700
|
| 254 |
+
},
|
| 255 |
+
{
|
| 256 |
+
"epoch": 4.33,
|
| 257 |
+
"learning_rate": 7.532894736842106e-06,
|
| 258 |
+
"loss": 2.302,
|
| 259 |
+
"step": 1750
|
| 260 |
+
},
|
| 261 |
+
{
|
| 262 |
+
"epoch": 4.46,
|
| 263 |
+
"learning_rate": 7.368421052631579e-06,
|
| 264 |
+
"loss": 2.2307,
|
| 265 |
+
"step": 1800
|
| 266 |
+
},
|
| 267 |
+
{
|
| 268 |
+
"epoch": 4.58,
|
| 269 |
+
"learning_rate": 7.203947368421054e-06,
|
| 270 |
+
"loss": 2.1823,
|
| 271 |
+
"step": 1850
|
| 272 |
+
},
|
| 273 |
+
{
|
| 274 |
+
"epoch": 4.7,
|
| 275 |
+
"learning_rate": 7.0394736842105274e-06,
|
| 276 |
+
"loss": 2.157,
|
| 277 |
+
"step": 1900
|
| 278 |
+
},
|
| 279 |
+
{
|
| 280 |
+
"epoch": 4.83,
|
| 281 |
+
"learning_rate": 6.875e-06,
|
| 282 |
+
"loss": 2.1087,
|
| 283 |
+
"step": 1950
|
| 284 |
+
},
|
| 285 |
+
{
|
| 286 |
+
"epoch": 4.95,
|
| 287 |
+
"learning_rate": 6.710526315789474e-06,
|
| 288 |
+
"loss": 2.0642,
|
| 289 |
+
"step": 2000
|
| 290 |
+
},
|
| 291 |
+
{
|
| 292 |
+
"epoch": 5.0,
|
| 293 |
+
"eval_cer": 0.9888314620091487,
|
| 294 |
+
"eval_loss": 1.9164613485336304,
|
| 295 |
+
"eval_runtime": 38.0086,
|
| 296 |
+
"eval_samples_per_second": 170.067,
|
| 297 |
+
"eval_steps_per_second": 10.629,
|
| 298 |
+
"eval_wer": 0.9797794777861581,
|
| 299 |
+
"step": 2020
|
| 300 |
+
},
|
| 301 |
+
{
|
| 302 |
+
"epoch": 5.07,
|
| 303 |
+
"learning_rate": 6.5460526315789476e-06,
|
| 304 |
+
"loss": 2.0318,
|
| 305 |
+
"step": 2050
|
| 306 |
+
},
|
| 307 |
+
{
|
| 308 |
+
"epoch": 5.2,
|
| 309 |
+
"learning_rate": 6.381578947368422e-06,
|
| 310 |
+
"loss": 2.0039,
|
| 311 |
+
"step": 2100
|
| 312 |
+
},
|
| 313 |
+
{
|
| 314 |
+
"epoch": 5.32,
|
| 315 |
+
"learning_rate": 6.217105263157896e-06,
|
| 316 |
+
"loss": 1.97,
|
| 317 |
+
"step": 2150
|
| 318 |
+
},
|
| 319 |
+
{
|
| 320 |
+
"epoch": 5.45,
|
| 321 |
+
"learning_rate": 6.0526315789473685e-06,
|
| 322 |
+
"loss": 1.9502,
|
| 323 |
+
"step": 2200
|
| 324 |
+
},
|
| 325 |
+
{
|
| 326 |
+
"epoch": 5.57,
|
| 327 |
+
"learning_rate": 5.888157894736842e-06,
|
| 328 |
+
"loss": 1.9235,
|
| 329 |
+
"step": 2250
|
| 330 |
+
},
|
| 331 |
+
{
|
| 332 |
+
"epoch": 5.69,
|
| 333 |
+
"learning_rate": 5.723684210526316e-06,
|
| 334 |
+
"loss": 1.8816,
|
| 335 |
+
"step": 2300
|
| 336 |
+
},
|
| 337 |
+
{
|
| 338 |
+
"epoch": 5.82,
|
| 339 |
+
"learning_rate": 5.55921052631579e-06,
|
| 340 |
+
"loss": 1.8675,
|
| 341 |
+
"step": 2350
|
| 342 |
+
},
|
| 343 |
+
{
|
| 344 |
+
"epoch": 5.94,
|
| 345 |
+
"learning_rate": 5.394736842105264e-06,
|
| 346 |
+
"loss": 1.834,
|
| 347 |
+
"step": 2400
|
| 348 |
+
},
|
| 349 |
+
{
|
| 350 |
+
"epoch": 6.0,
|
| 351 |
+
"eval_cer": 0.9463949781183786,
|
| 352 |
+
"eval_loss": 1.6738649606704712,
|
| 353 |
+
"eval_runtime": 38.6983,
|
| 354 |
+
"eval_samples_per_second": 167.036,
|
| 355 |
+
"eval_steps_per_second": 10.44,
|
| 356 |
+
"eval_wer": 0.9504004597205761,
|
| 357 |
+
"step": 2424
|
| 358 |
+
},
|
| 359 |
+
{
|
| 360 |
+
"epoch": 6.06,
|
| 361 |
+
"learning_rate": 5.230263157894737e-06,
|
| 362 |
+
"loss": 1.8211,
|
| 363 |
+
"step": 2450
|
| 364 |
+
},
|
| 365 |
+
{
|
| 366 |
+
"epoch": 6.19,
|
| 367 |
+
"learning_rate": 5.0657894736842104e-06,
|
| 368 |
+
"loss": 1.7972,
|
| 369 |
+
"step": 2500
|
| 370 |
+
},
|
| 371 |
+
{
|
| 372 |
+
"epoch": 6.31,
|
| 373 |
+
"learning_rate": 4.901315789473685e-06,
|
| 374 |
+
"loss": 1.7812,
|
| 375 |
+
"step": 2550
|
| 376 |
+
},
|
| 377 |
+
{
|
| 378 |
+
"epoch": 6.44,
|
| 379 |
+
"learning_rate": 4.736842105263158e-06,
|
| 380 |
+
"loss": 1.7627,
|
| 381 |
+
"step": 2600
|
| 382 |
+
},
|
| 383 |
+
{
|
| 384 |
+
"epoch": 6.56,
|
| 385 |
+
"learning_rate": 4.572368421052632e-06,
|
| 386 |
+
"loss": 1.7391,
|
| 387 |
+
"step": 2650
|
| 388 |
+
},
|
| 389 |
+
{
|
| 390 |
+
"epoch": 6.68,
|
| 391 |
+
"learning_rate": 4.407894736842105e-06,
|
| 392 |
+
"loss": 1.7241,
|
| 393 |
+
"step": 2700
|
| 394 |
+
},
|
| 395 |
+
{
|
| 396 |
+
"epoch": 6.81,
|
| 397 |
+
"learning_rate": 4.2434210526315796e-06,
|
| 398 |
+
"loss": 1.7089,
|
| 399 |
+
"step": 2750
|
| 400 |
+
},
|
| 401 |
+
{
|
| 402 |
+
"epoch": 6.93,
|
| 403 |
+
"learning_rate": 4.078947368421053e-06,
|
| 404 |
+
"loss": 1.6869,
|
| 405 |
+
"step": 2800
|
| 406 |
+
},
|
| 407 |
+
{
|
| 408 |
+
"epoch": 7.0,
|
| 409 |
+
"eval_cer": 0.7864710192281035,
|
| 410 |
+
"eval_loss": 1.4651445150375366,
|
| 411 |
+
"eval_runtime": 38.8273,
|
| 412 |
+
"eval_samples_per_second": 166.481,
|
| 413 |
+
"eval_steps_per_second": 10.405,
|
| 414 |
+
"eval_wer": 0.8239413856265488,
|
| 415 |
+
"step": 2828
|
| 416 |
+
},
|
| 417 |
+
{
|
| 418 |
+
"epoch": 7.05,
|
| 419 |
+
"learning_rate": 3.914473684210527e-06,
|
| 420 |
+
"loss": 1.6732,
|
| 421 |
+
"step": 2850
|
| 422 |
+
},
|
| 423 |
+
{
|
| 424 |
+
"epoch": 7.18,
|
| 425 |
+
"learning_rate": 3.7500000000000005e-06,
|
| 426 |
+
"loss": 1.6575,
|
| 427 |
+
"step": 2900
|
| 428 |
+
},
|
| 429 |
+
{
|
| 430 |
+
"epoch": 7.3,
|
| 431 |
+
"learning_rate": 3.5855263157894737e-06,
|
| 432 |
+
"loss": 1.6262,
|
| 433 |
+
"step": 2950
|
| 434 |
+
},
|
| 435 |
+
{
|
| 436 |
+
"epoch": 7.43,
|
| 437 |
+
"learning_rate": 3.421052631578948e-06,
|
| 438 |
+
"loss": 1.635,
|
| 439 |
+
"step": 3000
|
| 440 |
+
},
|
| 441 |
+
{
|
| 442 |
+
"epoch": 7.55,
|
| 443 |
+
"learning_rate": 3.256578947368421e-06,
|
| 444 |
+
"loss": 1.6008,
|
| 445 |
+
"step": 3050
|
| 446 |
+
},
|
| 447 |
+
{
|
| 448 |
+
"epoch": 7.67,
|
| 449 |
+
"learning_rate": 3.092105263157895e-06,
|
| 450 |
+
"loss": 1.5957,
|
| 451 |
+
"step": 3100
|
| 452 |
+
},
|
| 453 |
+
{
|
| 454 |
+
"epoch": 7.8,
|
| 455 |
+
"learning_rate": 2.927631578947369e-06,
|
| 456 |
+
"loss": 1.5809,
|
| 457 |
+
"step": 3150
|
| 458 |
+
},
|
| 459 |
+
{
|
| 460 |
+
"epoch": 7.92,
|
| 461 |
+
"learning_rate": 2.7631578947368424e-06,
|
| 462 |
+
"loss": 1.5734,
|
| 463 |
+
"step": 3200
|
| 464 |
+
},
|
| 465 |
+
{
|
| 466 |
+
"epoch": 8.0,
|
| 467 |
+
"eval_cer": 0.693914730984772,
|
| 468 |
+
"eval_loss": 1.3266735076904297,
|
| 469 |
+
"eval_runtime": 37.9957,
|
| 470 |
+
"eval_samples_per_second": 170.125,
|
| 471 |
+
"eval_steps_per_second": 10.633,
|
| 472 |
+
"eval_wer": 0.7439571885213518,
|
| 473 |
+
"step": 3232
|
| 474 |
+
},
|
| 475 |
+
{
|
| 476 |
+
"epoch": 8.04,
|
| 477 |
+
"learning_rate": 2.598684210526316e-06,
|
| 478 |
+
"loss": 1.5604,
|
| 479 |
+
"step": 3250
|
| 480 |
+
},
|
| 481 |
+
{
|
| 482 |
+
"epoch": 8.17,
|
| 483 |
+
"learning_rate": 2.4342105263157898e-06,
|
| 484 |
+
"loss": 1.5457,
|
| 485 |
+
"step": 3300
|
| 486 |
+
},
|
| 487 |
+
{
|
| 488 |
+
"epoch": 8.29,
|
| 489 |
+
"learning_rate": 2.2697368421052634e-06,
|
| 490 |
+
"loss": 1.5277,
|
| 491 |
+
"step": 3350
|
| 492 |
+
},
|
| 493 |
+
{
|
| 494 |
+
"epoch": 8.42,
|
| 495 |
+
"learning_rate": 2.105263157894737e-06,
|
| 496 |
+
"loss": 1.53,
|
| 497 |
+
"step": 3400
|
| 498 |
+
},
|
| 499 |
+
{
|
| 500 |
+
"epoch": 8.54,
|
| 501 |
+
"learning_rate": 1.9407894736842107e-06,
|
| 502 |
+
"loss": 1.5365,
|
| 503 |
+
"step": 3450
|
| 504 |
+
},
|
| 505 |
+
{
|
| 506 |
+
"epoch": 8.66,
|
| 507 |
+
"learning_rate": 1.7763157894736844e-06,
|
| 508 |
+
"loss": 1.5237,
|
| 509 |
+
"step": 3500
|
| 510 |
+
},
|
| 511 |
+
{
|
| 512 |
+
"epoch": 8.79,
|
| 513 |
+
"learning_rate": 1.611842105263158e-06,
|
| 514 |
+
"loss": 1.5106,
|
| 515 |
+
"step": 3550
|
| 516 |
+
},
|
| 517 |
+
{
|
| 518 |
+
"epoch": 8.91,
|
| 519 |
+
"learning_rate": 1.4473684210526317e-06,
|
| 520 |
+
"loss": 1.5052,
|
| 521 |
+
"step": 3600
|
| 522 |
+
},
|
| 523 |
+
{
|
| 524 |
+
"epoch": 9.0,
|
| 525 |
+
"eval_cer": 0.6230816451810927,
|
| 526 |
+
"eval_loss": 1.233142614364624,
|
| 527 |
+
"eval_runtime": 39.1396,
|
| 528 |
+
"eval_samples_per_second": 165.152,
|
| 529 |
+
"eval_steps_per_second": 10.322,
|
| 530 |
+
"eval_wer": 0.7045217828538591,
|
| 531 |
+
"step": 3636
|
| 532 |
+
},
|
| 533 |
+
{
|
| 534 |
+
"epoch": 9.03,
|
| 535 |
+
"learning_rate": 1.2828947368421055e-06,
|
| 536 |
+
"loss": 1.4932,
|
| 537 |
+
"step": 3650
|
| 538 |
+
},
|
| 539 |
+
{
|
| 540 |
+
"epoch": 9.16,
|
| 541 |
+
"learning_rate": 1.118421052631579e-06,
|
| 542 |
+
"loss": 1.4764,
|
| 543 |
+
"step": 3700
|
| 544 |
+
},
|
| 545 |
+
{
|
| 546 |
+
"epoch": 9.28,
|
| 547 |
+
"learning_rate": 9.539473684210528e-07,
|
| 548 |
+
"loss": 1.4474,
|
| 549 |
+
"step": 3750
|
| 550 |
+
},
|
| 551 |
+
{
|
| 552 |
+
"epoch": 9.41,
|
| 553 |
+
"learning_rate": 7.894736842105263e-07,
|
| 554 |
+
"loss": 1.4755,
|
| 555 |
+
"step": 3800
|
| 556 |
+
},
|
| 557 |
+
{
|
| 558 |
+
"epoch": 9.53,
|
| 559 |
+
"learning_rate": 6.25e-07,
|
| 560 |
+
"loss": 1.481,
|
| 561 |
+
"step": 3850
|
| 562 |
+
},
|
| 563 |
+
{
|
| 564 |
+
"epoch": 9.65,
|
| 565 |
+
"learning_rate": 4.605263157894737e-07,
|
| 566 |
+
"loss": 1.4635,
|
| 567 |
+
"step": 3900
|
| 568 |
+
},
|
| 569 |
+
{
|
| 570 |
+
"epoch": 9.78,
|
| 571 |
+
"learning_rate": 2.9605263157894736e-07,
|
| 572 |
+
"loss": 1.4576,
|
| 573 |
+
"step": 3950
|
| 574 |
+
},
|
| 575 |
+
{
|
| 576 |
+
"epoch": 9.9,
|
| 577 |
+
"learning_rate": 1.3157894736842107e-07,
|
| 578 |
+
"loss": 1.4573,
|
| 579 |
+
"step": 4000
|
| 580 |
+
},
|
| 581 |
+
{
|
| 582 |
+
"epoch": 10.0,
|
| 583 |
+
"eval_cer": 0.5793976118338977,
|
| 584 |
+
"eval_loss": 1.198919653892517,
|
| 585 |
+
"eval_runtime": 38.6365,
|
| 586 |
+
"eval_samples_per_second": 167.303,
|
| 587 |
+
"eval_steps_per_second": 10.456,
|
| 588 |
+
"eval_wer": 0.680063211579212,
|
| 589 |
+
"step": 4040
|
| 590 |
+
},
|
| 591 |
+
{
|
| 592 |
+
"epoch": 10.0,
|
| 593 |
+
"step": 4040,
|
| 594 |
+
"total_flos": 2.1546455399489285e+18,
|
| 595 |
+
"train_loss": 2.8324050827781755,
|
| 596 |
+
"train_runtime": 8680.4972,
|
| 597 |
+
"train_samples_per_second": 29.786,
|
| 598 |
+
"train_steps_per_second": 0.465
|
| 599 |
+
}
|
| 600 |
+
],
|
| 601 |
+
"logging_steps": 50,
|
| 602 |
+
"max_steps": 4040,
|
| 603 |
+
"num_train_epochs": 10,
|
| 604 |
+
"save_steps": 500,
|
| 605 |
+
"total_flos": 2.1546455399489285e+18,
|
| 606 |
+
"trial_name": null,
|
| 607 |
+
"trial_params": null
|
| 608 |
+
}
|