ymoslem commited on
Commit
33924d3
1 Parent(s): d13fa69

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +72 -11
README.md CHANGED
@@ -1,19 +1,41 @@
1
- ---
2
- license: apache-2.0
3
- base_model: openai/whisper-small
4
- tags:
5
- - generated_from_trainer
6
- model-index:
7
- - name: whisper-small-ga2en-v3.2-r
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
  # whisper-small-ga2en-v3.2-r
15
 
16
- This model is a fine-tuned version of [openai/whisper-small](https://huggingface.co/openai/whisper-small) on an unknown dataset.
 
 
 
 
 
 
 
17
 
18
  ## Model description
19
 
@@ -29,6 +51,10 @@ More information needed
29
 
30
  ## Training procedure
31
 
 
 
 
 
32
  ### Training hyperparameters
33
 
34
  The following hyperparameters were used during training:
@@ -41,9 +67,44 @@ The following hyperparameters were used during training:
41
  - training_steps: 3000
42
  - mixed_precision_training: Native AMP
43
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
44
  ### Framework versions
45
 
46
  - Transformers 4.40.2
47
  - Pytorch 2.2.0+cu121
48
  - Datasets 2.19.1
49
- - Tokenizers 0.19.1
 
1
+ ---
2
+ license: apache-2.0
3
+ base_model: openai/whisper-small
4
+ tags:
5
+ - generated_from_trainer
6
+ model-index:
7
+ - name: whisper-small-ga2en-v3.2-r
8
+ results: []
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
+ language:
17
+ - ga
18
+ - en
19
+ metrics:
20
+ - bleu
21
+ - wer
22
+ - chrf
23
+ library_name: transformers
24
+ ---
25
 
26
  <!-- This model card has been generated automatically according to the information the Trainer had access to. You
27
  should probably proofread and complete it, then remove this comment. -->
28
 
29
  # whisper-small-ga2en-v3.2-r
30
 
31
+ This model is a fine-tuned version of [openai/whisper-small](https://huggingface.co/openai/whisper-small)
32
+ on the IWSLT-2023, FLEURS, BiteSize, SpokenWords, Tatoeba, and Wikimedia datasets.
33
+ The best model checkpoint (this version) based on ChrF is at step 2700, epoch 3.5433, and
34
+ it achieves the following results on the evaluation set:
35
+ - Loss: 1.4313
36
+ - Bleu: 30.87
37
+ - Chrf: 47.72
38
+ - Wer: 64.2954
39
 
40
  ## Model description
41
 
 
51
 
52
  ## Training procedure
53
 
54
+ ### Hardware
55
+
56
+ 1 NVIDIA A100-SXM4-80GB
57
+
58
  ### Training hyperparameters
59
 
60
  The following hyperparameters were used during training:
 
67
  - training_steps: 3000
68
  - mixed_precision_training: Native AMP
69
 
70
+ ### Training results
71
+
72
+ |Step |Training Loss |Validation Loss |Bleu |Chrf |Wer|
73
+ |:-------------:|:------:|:----:|:---------------:|:-----:|:-----:|
74
+ |100 |2.311900 |1.773697 |9.20 |28.23 |120.486267|
75
+ |200 |1.870000 |1.479052 |16.90 |33.01 |83.701036|
76
+ |300 |1.627700 |1.372679 |20.33 |38.68 |84.061234|
77
+ |400 |1.460400 |1.309611 |24.52 |40.37 |74.696083|
78
+ |500 |1.336300 |1.283173 |20.35 |40.77 |86.537596|
79
+ |600 |1.193300 |1.255632 |20.63 |41.37 |95.632598|
80
+ |700 |1.015600 |1.251285 |21.24 |41.42 |82.170194|
81
+ |800 |0.518300 |1.292586 |28.08 |44.76 |66.951824|
82
+ |900 |0.467300 |1.329429 |25.16 |42.93 |76.316974|
83
+ |1000 |0.438700 |1.330984 |28.29 |46.08 |67.672220|
84
+ |1100 |0.403600 |1.300828 |27.43 |46.32 |68.977938|
85
+ |1200 |0.379500 |1.323791 |30.02 |45.48 |63.800090|
86
+ |1300 |0.337100 |1.327949 |30.40 |47.61 |61.999100|
87
+ |1400 |0.288000 |1.359497 |28.13 |44.60 |66.501576|
88
+ |1500 |0.265100 |1.355470 |26.58 |45.51 |71.319226|
89
+ |1600 |0.100800 |1.400149 |26.19 |46.02 |72.985142|
90
+ |1700 |0.092300 |1.383455 |24.83 |46.18 |77.532643|
91
+ |1800 |0.103900 |1.404863 |22.19 |43.19 |88.743809|
92
+ |1900 |0.090100 |1.402833 |29.73 |45.85 |66.186403|
93
+ |2000 |0.084200 |1.418717 |28.18 |45.29 |73.570464|
94
+ |2100 |0.074800 |1.461650 |26.58 |44.66 |74.020711|
95
+ |2200 |0.072000 |1.400547 |31.01 |47.30 |61.143629|
96
+ |2300 |0.042900 |1.424147 |28.72 |45.53 |65.511031|
97
+ |2400 |0.025200 |1.412174 |27.18 |47.19 |74.020711|
98
+ |2500 |0.026500 |1.438945 |30.01 |46.73 |65.105808|
99
+ |2600 |0.023300 |1.454140 |30.93 |46.65 |62.404322|
100
+ |2700 |0.021600 |1.431275 |30.87 |47.72 |64.295362|
101
+ |2800 |0.019200 |1.439022 |30.50 |46.98 |65.150833|
102
+ |2900 |0.018200 |1.439916 |31.09 |47.27 |63.529941|
103
+ |3000 |0.019000 |1.444545 |30.83 |47.35 |64.205313|
104
+
105
  ### Framework versions
106
 
107
  - Transformers 4.40.2
108
  - Pytorch 2.2.0+cu121
109
  - Datasets 2.19.1
110
+ - Tokenizers 0.19.1