Created README
Browse filesAdded results and whisper tag
README.md
ADDED
|
@@ -0,0 +1,35 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
language:
|
| 3 |
+
- ar
|
| 4 |
+
tags:
|
| 5 |
+
- whisper-event
|
| 6 |
+
- generated_from_trainer
|
| 7 |
+
datasets:
|
| 8 |
+
- mozilla-foundation/common_voice_11_0
|
| 9 |
+
metrics:
|
| 10 |
+
- wer
|
| 11 |
+
model-index:
|
| 12 |
+
- name: "Whisper Medium AR - gerryc"
|
| 13 |
+
results:
|
| 14 |
+
- task:
|
| 15 |
+
name: Automatic Speech Recognition
|
| 16 |
+
type: automatic-speech-recognition
|
| 17 |
+
dataset:
|
| 18 |
+
name: FLEURS
|
| 19 |
+
type: google/fleurs
|
| 20 |
+
config: ar_eg
|
| 21 |
+
split: test
|
| 22 |
+
args: ar
|
| 23 |
+
metrics:
|
| 24 |
+
- name: Wer
|
| 25 |
+
type: wer
|
| 26 |
+
value: 52.8
|
| 27 |
+
---
|
| 28 |
+
|
| 29 |
+
# Whisper Medium Ar - gerryc
|
| 30 |
+
|
| 31 |
+
Model was trained on CommonVoice Train. Tensorboard eval is on 256 samples of CommonVoice Dev.
|
| 32 |
+
|
| 33 |
+
Not normalized or lowercased. Evaluated and saved every 2500 steps.
|
| 34 |
+
This model was overtrained and is overfitted. About 2000 to 4000 steps is around the best spot to stop training.
|
| 35 |
+
|