End of training
Browse files
README.md
CHANGED
|
@@ -2,14 +2,28 @@
|
|
| 2 |
language:
|
| 3 |
- de
|
| 4 |
license: apache-2.0
|
| 5 |
-
|
| 6 |
tags:
|
| 7 |
- generated_from_trainer
|
|
|
|
| 8 |
datasets:
|
| 9 |
-
- rmacek/whisper-large-v3
|
|
|
|
|
|
|
| 10 |
model-index:
|
| 11 |
- name: Whisper ORF Bundeslaender
|
| 12 |
-
results:
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 13 |
---
|
| 14 |
|
| 15 |
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
|
|
@@ -18,6 +32,9 @@ should probably proofread and complete it, then remove this comment. -->
|
|
| 18 |
# Whisper ORF Bundeslaender
|
| 19 |
|
| 20 |
This model is a fine-tuned version of [openai/whisper-large-v3](https://huggingface.co/openai/whisper-large-v3) on the ZIB2 Common Voice dataset.
|
|
|
|
|
|
|
|
|
|
| 21 |
|
| 22 |
## Model description
|
| 23 |
|
|
@@ -37,18 +54,29 @@ More information needed
|
|
| 37 |
|
| 38 |
The following hyperparameters were used during training:
|
| 39 |
- learning_rate: 1e-05
|
| 40 |
-
- train_batch_size:
|
| 41 |
- eval_batch_size: 8
|
| 42 |
- seed: 42
|
| 43 |
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
|
| 44 |
- lr_scheduler_type: linear
|
| 45 |
- lr_scheduler_warmup_steps: 500
|
| 46 |
-
- training_steps:
|
| 47 |
- mixed_precision_training: Native AMP
|
| 48 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 49 |
### Framework versions
|
| 50 |
|
|
|
|
| 51 |
- Transformers 4.40.0
|
| 52 |
- Pytorch 2.2.1+cu121
|
| 53 |
- Datasets 2.19.0
|
| 54 |
-
- Tokenizers 0.19.1
|
|
|
|
| 2 |
language:
|
| 3 |
- de
|
| 4 |
license: apache-2.0
|
| 5 |
+
library_name: peft
|
| 6 |
tags:
|
| 7 |
- generated_from_trainer
|
| 8 |
+
base_model: openai/whisper-large-v3
|
| 9 |
datasets:
|
| 10 |
+
- rmacek/ORF-whisper-large-v3
|
| 11 |
+
metrics:
|
| 12 |
+
- wer
|
| 13 |
model-index:
|
| 14 |
- name: Whisper ORF Bundeslaender
|
| 15 |
+
results:
|
| 16 |
+
- task:
|
| 17 |
+
type: automatic-speech-recognition
|
| 18 |
+
name: Automatic Speech Recognition
|
| 19 |
+
dataset:
|
| 20 |
+
name: ZIB2 Common Voice
|
| 21 |
+
type: rmacek/ORF-whisper-large-v3
|
| 22 |
+
args: 'config: de, split: test'
|
| 23 |
+
metrics:
|
| 24 |
+
- type: wer
|
| 25 |
+
value: 17.29558995956067
|
| 26 |
+
name: Wer
|
| 27 |
---
|
| 28 |
|
| 29 |
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
|
|
|
|
| 32 |
# Whisper ORF Bundeslaender
|
| 33 |
|
| 34 |
This model is a fine-tuned version of [openai/whisper-large-v3](https://huggingface.co/openai/whisper-large-v3) on the ZIB2 Common Voice dataset.
|
| 35 |
+
It achieves the following results on the evaluation set:
|
| 36 |
+
- Loss: 0.3878
|
| 37 |
+
- Wer: 17.2956
|
| 38 |
|
| 39 |
## Model description
|
| 40 |
|
|
|
|
| 54 |
|
| 55 |
The following hyperparameters were used during training:
|
| 56 |
- learning_rate: 1e-05
|
| 57 |
+
- train_batch_size: 16
|
| 58 |
- eval_batch_size: 8
|
| 59 |
- seed: 42
|
| 60 |
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
|
| 61 |
- lr_scheduler_type: linear
|
| 62 |
- lr_scheduler_warmup_steps: 500
|
| 63 |
+
- training_steps: 4000
|
| 64 |
- mixed_precision_training: Native AMP
|
| 65 |
|
| 66 |
+
### Training results
|
| 67 |
+
|
| 68 |
+
| Training Loss | Epoch | Step | Validation Loss | Wer |
|
| 69 |
+
|:-------------:|:------:|:----:|:---------------:|:-------:|
|
| 70 |
+
| 0.3943 | 1.7153 | 1000 | 0.4072 | 17.5540 |
|
| 71 |
+
| 0.3431 | 3.4305 | 2000 | 0.3922 | 17.3458 |
|
| 72 |
+
| 0.3961 | 5.1458 | 3000 | 0.3885 | 17.3506 |
|
| 73 |
+
| 0.3548 | 6.8611 | 4000 | 0.3878 | 17.2956 |
|
| 74 |
+
|
| 75 |
+
|
| 76 |
### Framework versions
|
| 77 |
|
| 78 |
+
- PEFT 0.10.1.dev0
|
| 79 |
- Transformers 4.40.0
|
| 80 |
- Pytorch 2.2.1+cu121
|
| 81 |
- Datasets 2.19.0
|
| 82 |
+
- Tokenizers 0.19.1
|