Commit
•
10736ea
1
Parent(s):
33d7498
End of training
Browse files
README.md
ADDED
|
@@ -0,0 +1,86 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
library_name: transformers
|
| 3 |
+
language:
|
| 4 |
+
- ne
|
| 5 |
+
license: mit
|
| 6 |
+
base_model: facebook/w2v-bert-2.0
|
| 7 |
+
tags:
|
| 8 |
+
- generated_from_trainer
|
| 9 |
+
datasets:
|
| 10 |
+
- kiranpantha/OpenSLR54-Balanced-Nepali
|
| 11 |
+
metrics:
|
| 12 |
+
- wer
|
| 13 |
+
model-index:
|
| 14 |
+
- name: Wave2Vec2-Bert2.0 - Kiran Pantha
|
| 15 |
+
results:
|
| 16 |
+
- task:
|
| 17 |
+
name: Automatic Speech Recognition
|
| 18 |
+
type: automatic-speech-recognition
|
| 19 |
+
dataset:
|
| 20 |
+
name: kiranpantha/OpenSLR54-Balanced-Nepali
|
| 21 |
+
type: kiranpantha/OpenSLR54-Balanced-Nepali
|
| 22 |
+
args: 'config: ne, split: train,test'
|
| 23 |
+
metrics:
|
| 24 |
+
- name: Wer
|
| 25 |
+
type: wer
|
| 26 |
+
value: 0.45372112917023094
|
| 27 |
+
---
|
| 28 |
+
|
| 29 |
+
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
|
| 30 |
+
should probably proofread and complete it, then remove this comment. -->
|
| 31 |
+
|
| 32 |
+
# Wave2Vec2-Bert2.0 - Kiran Pantha
|
| 33 |
+
|
| 34 |
+
This model is a fine-tuned version of [facebook/w2v-bert-2.0](https://huggingface.co/facebook/w2v-bert-2.0) on the kiranpantha/OpenSLR54-Balanced-Nepali dataset.
|
| 35 |
+
It achieves the following results on the evaluation set:
|
| 36 |
+
- Loss: 0.5146
|
| 37 |
+
- Wer: 0.4537
|
| 38 |
+
- Cer: 0.1137
|
| 39 |
+
|
| 40 |
+
## Model description
|
| 41 |
+
|
| 42 |
+
More information needed
|
| 43 |
+
|
| 44 |
+
## Intended uses & limitations
|
| 45 |
+
|
| 46 |
+
More information needed
|
| 47 |
+
|
| 48 |
+
## Training and evaluation data
|
| 49 |
+
|
| 50 |
+
More information needed
|
| 51 |
+
|
| 52 |
+
## Training procedure
|
| 53 |
+
|
| 54 |
+
### Training hyperparameters
|
| 55 |
+
|
| 56 |
+
The following hyperparameters were used during training:
|
| 57 |
+
- learning_rate: 5e-05
|
| 58 |
+
- train_batch_size: 8
|
| 59 |
+
- eval_batch_size: 8
|
| 60 |
+
- seed: 42
|
| 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 |
+
- num_epochs: 2
|
| 65 |
+
- mixed_precision_training: Native AMP
|
| 66 |
+
|
| 67 |
+
### Training results
|
| 68 |
+
|
| 69 |
+
| Training Loss | Epoch | Step | Validation Loss | Wer | Cer |
|
| 70 |
+
|:-------------:|:------:|:----:|:---------------:|:------:|:------:|
|
| 71 |
+
| 0.3129 | 0.24 | 300 | 0.5021 | 0.4484 | 0.1119 |
|
| 72 |
+
| 0.3868 | 0.48 | 600 | 0.5117 | 0.4686 | 0.1193 |
|
| 73 |
+
| 0.368 | 0.72 | 900 | 0.5399 | 0.4674 | 0.1291 |
|
| 74 |
+
| 0.3462 | 0.96 | 1200 | 0.4893 | 0.4506 | 0.1131 |
|
| 75 |
+
| 0.3009 | 1.2 | 1500 | 0.5081 | 0.4505 | 0.1134 |
|
| 76 |
+
| 0.2721 | 1.44 | 1800 | 0.5146 | 0.4681 | 0.1159 |
|
| 77 |
+
| 0.2499 | 1.6800 | 2100 | 0.5128 | 0.4549 | 0.1128 |
|
| 78 |
+
| 0.2366 | 1.92 | 2400 | 0.5146 | 0.4537 | 0.1137 |
|
| 79 |
+
|
| 80 |
+
|
| 81 |
+
### Framework versions
|
| 82 |
+
|
| 83 |
+
- Transformers 4.44.2
|
| 84 |
+
- Pytorch 2.4.1+cu121
|
| 85 |
+
- Datasets 3.0.1
|
| 86 |
+
- Tokenizers 0.19.1
|