End of training
Browse files
README.md
CHANGED
@@ -10,7 +10,7 @@ datasets:
|
|
10 |
metrics:
|
11 |
- wer
|
12 |
model-index:
|
13 |
-
- name:
|
14 |
results:
|
15 |
- task:
|
16 |
name: Automatic Speech Recognition
|
@@ -19,23 +19,23 @@ model-index:
|
|
19 |
name: Common Voice 11.0
|
20 |
type: mozilla-foundation/common_voice_11_0
|
21 |
config: hi
|
22 |
-
split:
|
23 |
args: 'config: hi, split: test'
|
24 |
metrics:
|
25 |
- name: Wer
|
26 |
type: wer
|
27 |
-
value:
|
28 |
---
|
29 |
|
30 |
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
|
31 |
should probably proofread and complete it, then remove this comment. -->
|
32 |
|
33 |
-
#
|
34 |
|
35 |
This model is a fine-tuned version of [openai/whisper-small](https://huggingface.co/openai/whisper-small) on the Common Voice 11.0 dataset.
|
36 |
It achieves the following results on the evaluation set:
|
37 |
-
- Loss: 0.
|
38 |
-
- Wer:
|
39 |
|
40 |
## Model description
|
41 |
|
@@ -61,20 +61,19 @@ The following hyperparameters were used during training:
|
|
61 |
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
|
62 |
- lr_scheduler_type: linear
|
63 |
- lr_scheduler_warmup_steps: 500
|
64 |
-
- training_steps:
|
65 |
- mixed_precision_training: Native AMP
|
66 |
|
67 |
### Training results
|
68 |
|
69 |
| Training Loss | Epoch | Step | Validation Loss | Wer |
|
70 |
|:-------------:|:------:|:----:|:---------------:|:-------:|
|
71 |
-
| 0.
|
72 |
-
| 0.0175 | 4.8900 | 2000 | 0.3303 | 33.1288 |
|
73 |
|
74 |
|
75 |
### Framework versions
|
76 |
|
77 |
- Transformers 4.41.2
|
78 |
-
- Pytorch 2.
|
79 |
- Datasets 2.20.0
|
80 |
- Tokenizers 0.19.1
|
|
|
10 |
metrics:
|
11 |
- wer
|
12 |
model-index:
|
13 |
+
- name: Hindi Whisper Model
|
14 |
results:
|
15 |
- task:
|
16 |
name: Automatic Speech Recognition
|
|
|
19 |
name: Common Voice 11.0
|
20 |
type: mozilla-foundation/common_voice_11_0
|
21 |
config: hi
|
22 |
+
split: test
|
23 |
args: 'config: hi, split: test'
|
24 |
metrics:
|
25 |
- name: Wer
|
26 |
type: wer
|
27 |
+
value: 34.38584610175231
|
28 |
---
|
29 |
|
30 |
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
|
31 |
should probably proofread and complete it, then remove this comment. -->
|
32 |
|
33 |
+
# Hindi Whisper Model
|
34 |
|
35 |
This model is a fine-tuned version of [openai/whisper-small](https://huggingface.co/openai/whisper-small) on the Common Voice 11.0 dataset.
|
36 |
It achieves the following results on the evaluation set:
|
37 |
+
- Loss: 0.2859
|
38 |
+
- Wer: 34.3858
|
39 |
|
40 |
## Model description
|
41 |
|
|
|
61 |
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
|
62 |
- lr_scheduler_type: linear
|
63 |
- lr_scheduler_warmup_steps: 500
|
64 |
+
- training_steps: 1000
|
65 |
- mixed_precision_training: Native AMP
|
66 |
|
67 |
### Training results
|
68 |
|
69 |
| Training Loss | Epoch | Step | Validation Loss | Wer |
|
70 |
|:-------------:|:------:|:----:|:---------------:|:-------:|
|
71 |
+
| 0.0818 | 2.4450 | 1000 | 0.2859 | 34.3858 |
|
|
|
72 |
|
73 |
|
74 |
### Framework versions
|
75 |
|
76 |
- Transformers 4.41.2
|
77 |
+
- Pytorch 2.1.2
|
78 |
- Datasets 2.20.0
|
79 |
- Tokenizers 0.19.1
|