Commit
•
2b1fb87
1
Parent(s):
178557a
update model card README.md
Browse files
README.md
CHANGED
@@ -14,6 +14,9 @@ should probably proofread and complete it, then remove this comment. -->
|
|
14 |
#
|
15 |
|
16 |
This model was trained from scratch on the librispeech_asr dataset.
|
|
|
|
|
|
|
17 |
|
18 |
## Model description
|
19 |
|
@@ -32,12 +35,12 @@ More information needed
|
|
32 |
### Training hyperparameters
|
33 |
|
34 |
The following hyperparameters were used during training:
|
35 |
-
- learning_rate:
|
36 |
-
- train_batch_size:
|
37 |
-
- eval_batch_size:
|
38 |
- seed: 42
|
39 |
-
- gradient_accumulation_steps:
|
40 |
-
- total_train_batch_size:
|
41 |
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
|
42 |
- lr_scheduler_type: linear
|
43 |
- lr_scheduler_warmup_steps: 500
|
@@ -46,6 +49,11 @@ The following hyperparameters were used during training:
|
|
46 |
|
47 |
### Training results
|
48 |
|
|
|
|
|
|
|
|
|
|
|
49 |
|
50 |
|
51 |
### Framework versions
|
|
|
14 |
#
|
15 |
|
16 |
This model was trained from scratch on the librispeech_asr dataset.
|
17 |
+
It achieves the following results on the evaluation set:
|
18 |
+
- Loss: 4.1930
|
19 |
+
- Wer: 1.8951
|
20 |
|
21 |
## Model description
|
22 |
|
|
|
35 |
### Training hyperparameters
|
36 |
|
37 |
The following hyperparameters were used during training:
|
38 |
+
- learning_rate: 3e-05
|
39 |
+
- train_batch_size: 8
|
40 |
+
- eval_batch_size: 8
|
41 |
- seed: 42
|
42 |
+
- gradient_accumulation_steps: 2
|
43 |
+
- total_train_batch_size: 16
|
44 |
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
|
45 |
- lr_scheduler_type: linear
|
46 |
- lr_scheduler_warmup_steps: 500
|
|
|
49 |
|
50 |
### Training results
|
51 |
|
52 |
+
| Training Loss | Epoch | Step | Validation Loss | Wer |
|
53 |
+
|:-------------:|:-----:|:----:|:---------------:|:------:|
|
54 |
+
| 4.1429 | 0.28 | 500 | 4.3255 | 1.9596 |
|
55 |
+
| 4.4493 | 0.56 | 1000 | 4.2641 | 1.9047 |
|
56 |
+
| 4.2715 | 0.84 | 1500 | 4.1930 | 1.8951 |
|
57 |
|
58 |
|
59 |
### Framework versions
|