End of training
Browse files
README.md
CHANGED
@@ -1,19 +1,50 @@
|
|
1 |
---
|
|
|
|
|
|
|
2 |
license: apache-2.0
|
3 |
base_model: openai/whisper-small
|
4 |
tags:
|
5 |
- generated_from_trainer
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
6 |
model-index:
|
7 |
-
- name:
|
8 |
-
results:
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
9 |
---
|
10 |
|
11 |
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
|
12 |
should probably proofread and complete it, then remove this comment. -->
|
13 |
|
14 |
-
#
|
15 |
|
16 |
-
This model is a fine-tuned version of [openai/whisper-small](https://huggingface.co/openai/whisper-small) on
|
|
|
|
|
|
|
|
|
|
|
17 |
|
18 |
## Model description
|
19 |
|
@@ -42,6 +73,22 @@ The following hyperparameters were used during training:
|
|
42 |
- training_steps: 1000
|
43 |
- mixed_precision_training: Native AMP
|
44 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
45 |
### Framework versions
|
46 |
|
47 |
- Transformers 4.39.3
|
|
|
1 |
---
|
2 |
+
language:
|
3 |
+
- ga
|
4 |
+
- en
|
5 |
license: apache-2.0
|
6 |
base_model: openai/whisper-small
|
7 |
tags:
|
8 |
- generated_from_trainer
|
9 |
+
datasets:
|
10 |
+
- ymoslem/IWSLT2023-GA-EN
|
11 |
+
- ymoslem/FLEURS-GA-EN
|
12 |
+
- ymoslem/BitesizeIrish-GA-EN
|
13 |
+
- ymoslem/SpokenWords-GA-EN-MTed
|
14 |
+
- ymoslem/Tatoeba-Speech-Irish
|
15 |
+
- ymoslem/Wikimedia-Speech-Irish
|
16 |
+
metrics:
|
17 |
+
- bleu
|
18 |
+
- wer
|
19 |
model-index:
|
20 |
+
- name: Whisper Small GA-EN Speech Translation
|
21 |
+
results:
|
22 |
+
- task:
|
23 |
+
name: Automatic Speech Recognition
|
24 |
+
type: automatic-speech-recognition
|
25 |
+
dataset:
|
26 |
+
name: IWSLT-2023, FLEURS, BiteSize, SpokenWords, Tatoeba, and Wikimedia
|
27 |
+
type: ymoslem/IWSLT2023-GA-EN
|
28 |
+
metrics:
|
29 |
+
- name: Bleu
|
30 |
+
type: bleu
|
31 |
+
value: 25.68
|
32 |
+
- name: Wer
|
33 |
+
type: wer
|
34 |
+
value: 71.04907699234579
|
35 |
---
|
36 |
|
37 |
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
|
38 |
should probably proofread and complete it, then remove this comment. -->
|
39 |
|
40 |
+
# Whisper Small GA-EN Speech Translation
|
41 |
|
42 |
+
This model is a fine-tuned version of [openai/whisper-small](https://huggingface.co/openai/whisper-small) on the IWSLT-2023, FLEURS, BiteSize, SpokenWords, Tatoeba, and Wikimedia dataset.
|
43 |
+
It achieves the following results on the evaluation set:
|
44 |
+
- Loss: 1.1571
|
45 |
+
- Bleu: 25.68
|
46 |
+
- Chrf: 45.53
|
47 |
+
- Wer: 71.0491
|
48 |
|
49 |
## Model description
|
50 |
|
|
|
73 |
- training_steps: 1000
|
74 |
- mixed_precision_training: Native AMP
|
75 |
|
76 |
+
### Training results
|
77 |
+
|
78 |
+
| Training Loss | Epoch | Step | Validation Loss | Bleu | Chrf | Wer |
|
79 |
+
|:-------------:|:-----:|:----:|:---------------:|:-----:|:-----:|:--------:|
|
80 |
+
| 2.6685 | 0.07 | 100 | 2.0544 | 5.05 | 20.18 | 139.8919 |
|
81 |
+
| 2.4028 | 0.13 | 200 | 1.7367 | 12.29 | 29.72 | 95.5425 |
|
82 |
+
| 2.1231 | 0.2 | 300 | 1.6141 | 14.33 | 30.77 | 101.3958 |
|
83 |
+
| 1.9192 | 0.26 | 400 | 1.4778 | 16.86 | 35.65 | 91.0851 |
|
84 |
+
| 1.7129 | 0.33 | 500 | 1.3811 | 16.77 | 37.53 | 93.8766 |
|
85 |
+
| 1.5398 | 0.39 | 600 | 1.3427 | 18.85 | 39.0 | 90.2296 |
|
86 |
+
| 1.4257 | 0.46 | 700 | 1.2784 | 25.73 | 43.3 | 70.3287 |
|
87 |
+
| 1.3044 | 0.53 | 800 | 1.2274 | 25.43 | 44.33 | 72.3548 |
|
88 |
+
| 1.2626 | 0.59 | 900 | 1.1875 | 25.09 | 44.62 | 72.6249 |
|
89 |
+
| 1.2801 | 0.66 | 1000 | 1.1571 | 25.68 | 45.53 | 71.0491 |
|
90 |
+
|
91 |
+
|
92 |
### Framework versions
|
93 |
|
94 |
- Transformers 4.39.3
|