readme
Browse files
README.md
CHANGED
|
@@ -1,3 +1,72 @@
|
|
| 1 |
---
|
|
|
|
|
|
|
| 2 |
license: apache-2.0
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 3 |
---
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
---
|
| 2 |
+
language:
|
| 3 |
+
- no
|
| 4 |
license: apache-2.0
|
| 5 |
+
tags:
|
| 6 |
+
- whisper-event
|
| 7 |
+
- norwegian
|
| 8 |
+
datasets:
|
| 9 |
+
- NbAiLab/NCC_S
|
| 10 |
+
- NbAiLab/NPSC
|
| 11 |
+
- NbAiLab/NST
|
| 12 |
+
- google/fleurs
|
| 13 |
+
metrics:
|
| 14 |
+
- wer
|
| 15 |
+
model-index:
|
| 16 |
+
- name: Whisper Tiny Norwegian Bokmål
|
| 17 |
+
results:
|
| 18 |
+
- task:
|
| 19 |
+
name: Automatic Speech Recognition
|
| 20 |
+
type: automatic-speech-recognition
|
| 21 |
+
dataset:
|
| 22 |
+
name: FLEURS
|
| 23 |
+
type: google/fleurs
|
| 24 |
+
config: nb_no
|
| 25 |
+
split: test
|
| 26 |
+
args: nb_no
|
| 27 |
+
metrics:
|
| 28 |
+
- name: Wer
|
| 29 |
+
type: wer
|
| 30 |
+
value: 47.08
|
| 31 |
---
|
| 32 |
+
|
| 33 |
+
# Whisper Tiny Norwegian Bokmål
|
| 34 |
+
|
| 35 |
+
This model is a fine-tuned version of [openai/whisper-medium](https://huggingface.co/openai/whisper-medium) trained on several datasets.
|
| 36 |
+
|
| 37 |
+
It is currently in the middle of a large trainingi. Currently achieves the following results on the evaluation set:
|
| 38 |
+
- Loss: 1.464
|
| 39 |
+
- Wer: 47.08
|
| 40 |
+
|
| 41 |
+
## Model description
|
| 42 |
+
|
| 43 |
+
The model is trained on a large corpus of roughly 5.000 hours of voice. The sources are subtitles from the Norwegian broadcaster NRK, transcribed speeches from the Norwegian parliament and voice recordings from Norsk Språkteknologi.
|
| 44 |
+
|
| 45 |
+
## Intended uses & limitations
|
| 46 |
+
|
| 47 |
+
The model will be free for everyone to use when it is finished.
|
| 48 |
+
|
| 49 |
+
## Training and evaluation data
|
| 50 |
+
|
| 51 |
+
More information needed
|
| 52 |
+
|
| 53 |
+
## Training procedure
|
| 54 |
+
|
| 55 |
+
### Training hyperparameters
|
| 56 |
+
|
| 57 |
+
The following hyperparameters were used during training:
|
| 58 |
+
- learning_rate: 3e-06
|
| 59 |
+
- train_batch_size: 128
|
| 60 |
+
- eval_batch_size: 32
|
| 61 |
+
- seed: 42
|
| 62 |
+
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
|
| 63 |
+
- lr_scheduler_type: linear
|
| 64 |
+
- lr_scheduler_warmup_steps: 500
|
| 65 |
+
- training_steps: 100.000 (currently 4.000)
|
| 66 |
+
- mixed_precision_training: fp16
|
| 67 |
+
|
| 68 |
+
### Training results
|
| 69 |
+
See [Tensorboad Metrics](https://huggingface.co/NbAiLab/whisper-tiny-nob/tensorboard)
|
| 70 |
+
|
| 71 |
+
|
| 72 |
+
|
runs/Dec15_07-29-26_ficino/events.out.tfevents.1671085786.ficino.1546868.0
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
-
size
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:6eb8ce19f504fe6cc167f1f5e500d8e5a7ac04e2130444bc5017c0bb6b76b081
|
| 3 |
+
size 20784
|