wav2vec2-attempt2

README.md

@@ -3,6 +3,9 @@ license: apache-2.0
 base_model: facebook/wav2vec2-base
 tags:
 - generated_from_trainer
+metrics:
+- accuracy
+- f1
 model-index:
 - name: wav2vec2-attempt2
   results: []
@@ -15,14 +18,9 @@ should probably proofread and complete it, then remove this comment. -->
 
 This model is a fine-tuned version of [facebook/wav2vec2-base](https://huggingface.co/facebook/wav2vec2-base) on the None dataset.
 It achieves the following results on the evaluation set:
-- eval_loss: 0.2901
-- eval_accuracy: 0.9246
-- eval_f1: 0.9248
-- eval_runtime: 1457.2717
-- eval_samples_per_second: 9.588
-- eval_steps_per_second: 1.066
-- epoch: 5.0
-- step: 3234
+- Loss: 0.2685
+- Accuracy: 0.9373
+- F1: 0.9372
 
 ## Model description
 
@@ -53,6 +51,18 @@ The following hyperparameters were used during training:
 - num_epochs: 6
 - mixed_precision_training: Native AMP
 
+### Training results
+
+| Training Loss | Epoch | Step | Validation Loss | Accuracy | F1     |
+|:-------------:|:-----:|:----:|:---------------:|:--------:|:------:|
+| 1.0285        | 1.0   | 646  | 0.9877          | 0.7123   | 0.7029 |
+| 0.6367        | 2.0   | 1293 | 0.5708          | 0.8314   | 0.8333 |
+| 0.3915        | 3.0   | 1940 | 0.4389          | 0.8745   | 0.8730 |
+| 0.2359        | 4.0   | 2587 | 0.3361          | 0.9077   | 0.9082 |
+| 0.0987        | 5.0   | 3234 | 0.2901          | 0.9246   | 0.9248 |
+| 0.0705        | 5.99  | 3876 | 0.2685          | 0.9373   | 0.9372 |
+
+
 ### Framework versions
 
 - Transformers 4.36.2