Commit
•
368754a
1
Parent(s):
f14d5dd
End of training
Browse files
README.md
CHANGED
@@ -22,7 +22,7 @@ model-index:
|
|
22 |
metrics:
|
23 |
- name: Wer
|
24 |
type: wer
|
25 |
-
value: 0.
|
26 |
---
|
27 |
|
28 |
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
|
@@ -32,9 +32,9 @@ should probably proofread and complete it, then remove this comment. -->
|
|
32 |
|
33 |
This model is a fine-tuned version of [openai/whisper-tiny](https://huggingface.co/openai/whisper-tiny) on the PolyAI/minds14 dataset.
|
34 |
It achieves the following results on the evaluation set:
|
35 |
-
- Loss: 0.
|
36 |
-
- Wer Ortho: 0.
|
37 |
-
- Wer: 0.
|
38 |
|
39 |
## Model description
|
40 |
|
@@ -53,32 +53,24 @@ More information needed
|
|
53 |
### Training hyperparameters
|
54 |
|
55 |
The following hyperparameters were used during training:
|
56 |
-
- learning_rate:
|
57 |
- train_batch_size: 16
|
58 |
- eval_batch_size: 16
|
59 |
- seed: 42
|
60 |
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
|
61 |
- lr_scheduler_type: constant_with_warmup
|
62 |
- lr_scheduler_warmup_steps: 50
|
63 |
-
- training_steps:
|
64 |
- mixed_precision_training: Native AMP
|
65 |
|
66 |
### Training results
|
67 |
|
68 |
| Training Loss | Epoch | Step | Validation Loss | Wer Ortho | Wer |
|
69 |
|:-------------:|:-----:|:----:|:---------------:|:---------:|:------:|
|
70 |
-
|
|
71 |
-
| 0.
|
72 |
-
| 0.
|
73 |
-
| 0.
|
74 |
-
| 0.0912 | 8.93 | 250 | 0.5966 | 0.4399 | 0.3985 |
|
75 |
-
| 0.0385 | 10.71 | 300 | 0.6222 | 0.4324 | 0.3996 |
|
76 |
-
| 0.0203 | 12.5 | 350 | 0.6532 | 0.4861 | 0.4616 |
|
77 |
-
| 0.0079 | 14.29 | 400 | 0.6867 | 0.5503 | 0.5331 |
|
78 |
-
| 0.0039 | 16.07 | 450 | 0.7002 | 0.5713 | 0.5555 |
|
79 |
-
| 0.0026 | 17.86 | 500 | 0.7144 | 0.6428 | 0.6275 |
|
80 |
-
| 0.0021 | 19.64 | 550 | 0.7275 | 0.6619 | 0.6499 |
|
81 |
-
| 0.0017 | 21.43 | 600 | 0.7357 | 0.6860 | 0.6736 |
|
82 |
|
83 |
|
84 |
### Framework versions
|
|
|
22 |
metrics:
|
23 |
- name: Wer
|
24 |
type: wer
|
25 |
+
value: 0.4244391971664699
|
26 |
---
|
27 |
|
28 |
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
|
|
|
32 |
|
33 |
This model is a fine-tuned version of [openai/whisper-tiny](https://huggingface.co/openai/whisper-tiny) on the PolyAI/minds14 dataset.
|
34 |
It achieves the following results on the evaluation set:
|
35 |
+
- Loss: 0.6467
|
36 |
+
- Wer Ortho: 0.4534
|
37 |
+
- Wer: 0.4244
|
38 |
|
39 |
## Model description
|
40 |
|
|
|
53 |
### Training hyperparameters
|
54 |
|
55 |
The following hyperparameters were used during training:
|
56 |
+
- learning_rate: 4e-06
|
57 |
- train_batch_size: 16
|
58 |
- eval_batch_size: 16
|
59 |
- seed: 42
|
60 |
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
|
61 |
- lr_scheduler_type: constant_with_warmup
|
62 |
- lr_scheduler_warmup_steps: 50
|
63 |
+
- training_steps: 200
|
64 |
- mixed_precision_training: Native AMP
|
65 |
|
66 |
### Training results
|
67 |
|
68 |
| Training Loss | Epoch | Step | Validation Loss | Wer Ortho | Wer |
|
69 |
|:-------------:|:-----:|:----:|:---------------:|:---------:|:------:|
|
70 |
+
| 0.1212 | 1.79 | 50 | 0.5863 | 0.4584 | 0.4085 |
|
71 |
+
| 0.0775 | 3.57 | 100 | 0.5957 | 0.4442 | 0.3991 |
|
72 |
+
| 0.0415 | 5.36 | 150 | 0.6280 | 0.4399 | 0.4091 |
|
73 |
+
| 0.0208 | 7.14 | 200 | 0.6467 | 0.4534 | 0.4244 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
74 |
|
75 |
|
76 |
### Framework versions
|
runs/Feb04_23-06-31_MSI/events.out.tfevents.1707087993.MSI
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
-
size
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:03e18889bd5332d00a1da2dfac40ddece034a06ec989192dfced8d8ec714166f
|
3 |
+
size 9190
|