zainulhakim
commited on
Commit
•
e3010c3
1
Parent(s):
71e9741
240624-wav2vec2-ASR-Arab
Browse files
README.md
CHANGED
@@ -1,6 +1,6 @@
|
|
1 |
---
|
2 |
license: apache-2.0
|
3 |
-
base_model: zainulhakim/
|
4 |
tags:
|
5 |
- generated_from_trainer
|
6 |
metrics:
|
@@ -61,12 +61,13 @@ should probably proofread and complete it, then remove this comment. -->
|
|
61 |
[<img src="https://raw.githubusercontent.com/wandb/assets/main/wandb-github-badge-28.svg" alt="Visualize in Weights & Biases" width="200" height="32"/>](https://wandb.ai/projektarbeit_dalim/huggingface/runs/ogklm4ck)
|
62 |
[<img src="https://raw.githubusercontent.com/wandb/assets/main/wandb-github-badge-28.svg" alt="Visualize in Weights & Biases" width="200" height="32"/>](https://wandb.ai/projektarbeit_dalim/huggingface/runs/ogklm4ck)
|
63 |
[<img src="https://raw.githubusercontent.com/wandb/assets/main/wandb-github-badge-28.svg" alt="Visualize in Weights & Biases" width="200" height="32"/>](https://wandb.ai/projektarbeit_dalim/huggingface/runs/ogklm4ck)
|
|
|
64 |
# 240624-wav2vec2-ASR-Arab
|
65 |
|
66 |
-
This model is a fine-tuned version of [zainulhakim/
|
67 |
It achieves the following results on the evaluation set:
|
68 |
-
- Loss: 1.
|
69 |
-
- Wer: 0.
|
70 |
|
71 |
## Model description
|
72 |
|
@@ -92,17 +93,16 @@ The following hyperparameters were used during training:
|
|
92 |
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
|
93 |
- lr_scheduler_type: linear
|
94 |
- lr_scheduler_warmup_steps: 500
|
95 |
-
- num_epochs:
|
96 |
- mixed_precision_training: Native AMP
|
97 |
|
98 |
### Training results
|
99 |
|
100 |
| Training Loss | Epoch | Step | Validation Loss | Wer |
|
101 |
|:-------------:|:-----:|:----:|:---------------:|:------:|
|
102 |
-
| No log | 6.25 | 100 |
|
103 |
-
| No log | 12.5 | 200 | 1.
|
104 |
-
| No log | 18.75 | 300 | 1.
|
105 |
-
| No log | 25.0 | 400 | 1.5319 | 0.4545 |
|
106 |
|
107 |
|
108 |
### Framework versions
|
|
|
1 |
---
|
2 |
license: apache-2.0
|
3 |
+
base_model: zainulhakim/240615-wav2vec2-ASR-Arabs
|
4 |
tags:
|
5 |
- generated_from_trainer
|
6 |
metrics:
|
|
|
61 |
[<img src="https://raw.githubusercontent.com/wandb/assets/main/wandb-github-badge-28.svg" alt="Visualize in Weights & Biases" width="200" height="32"/>](https://wandb.ai/projektarbeit_dalim/huggingface/runs/ogklm4ck)
|
62 |
[<img src="https://raw.githubusercontent.com/wandb/assets/main/wandb-github-badge-28.svg" alt="Visualize in Weights & Biases" width="200" height="32"/>](https://wandb.ai/projektarbeit_dalim/huggingface/runs/ogklm4ck)
|
63 |
[<img src="https://raw.githubusercontent.com/wandb/assets/main/wandb-github-badge-28.svg" alt="Visualize in Weights & Biases" width="200" height="32"/>](https://wandb.ai/projektarbeit_dalim/huggingface/runs/ogklm4ck)
|
64 |
+
[<img src="https://raw.githubusercontent.com/wandb/assets/main/wandb-github-badge-28.svg" alt="Visualize in Weights & Biases" width="200" height="32"/>](https://wandb.ai/projektarbeit_dalim/dalim/runs/7r04i5y6)
|
65 |
# 240624-wav2vec2-ASR-Arab
|
66 |
|
67 |
+
This model is a fine-tuned version of [zainulhakim/240615-wav2vec2-ASR-Arabs](https://huggingface.co/zainulhakim/240615-wav2vec2-ASR-Arabs) on an unknown dataset.
|
68 |
It achieves the following results on the evaluation set:
|
69 |
+
- Loss: 1.6706
|
70 |
+
- Wer: 0.9697
|
71 |
|
72 |
## Model description
|
73 |
|
|
|
93 |
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
|
94 |
- lr_scheduler_type: linear
|
95 |
- lr_scheduler_warmup_steps: 500
|
96 |
+
- num_epochs: 20
|
97 |
- mixed_precision_training: Native AMP
|
98 |
|
99 |
### Training results
|
100 |
|
101 |
| Training Loss | Epoch | Step | Validation Loss | Wer |
|
102 |
|:-------------:|:-----:|:----:|:---------------:|:------:|
|
103 |
+
| No log | 6.25 | 100 | 2.5459 | 1.0 |
|
104 |
+
| No log | 12.5 | 200 | 1.4102 | 1.0 |
|
105 |
+
| No log | 18.75 | 300 | 1.6706 | 0.9697 |
|
|
|
106 |
|
107 |
|
108 |
### Framework versions
|