lewtun's picture
lewtun HF staff
update model card README.md
3d9b0c9
---
license: mit
tags:
- generated_from_trainer
metrics:
- accuracy
model-index:
- name: deberta-v3-large__sst2__train-8-9
results: []
---
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
should probably proofread and complete it, then remove this comment. -->
# deberta-v3-large__sst2__train-8-9
This model is a fine-tuned version of [microsoft/deberta-v3-large](https://huggingface.co/microsoft/deberta-v3-large) on the None dataset.
It achieves the following results on the evaluation set:
- Loss: 0.6013
- Accuracy: 0.7210
## Model description
More information needed
## Intended uses & limitations
More information needed
## Training and evaluation data
More information needed
## Training procedure
### Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 2e-05
- train_batch_size: 4
- eval_batch_size: 4
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 50
- mixed_precision_training: Native AMP
### Training results
| Training Loss | Epoch | Step | Validation Loss | Accuracy |
|:-------------:|:-----:|:----:|:---------------:|:--------:|
| 0.6757 | 1.0 | 3 | 0.7810 | 0.25 |
| 0.6506 | 2.0 | 6 | 0.8102 | 0.25 |
| 0.6463 | 3.0 | 9 | 0.8313 | 0.25 |
| 0.5813 | 4.0 | 12 | 0.8858 | 0.25 |
| 0.4635 | 5.0 | 15 | 0.8220 | 0.25 |
| 0.3992 | 6.0 | 18 | 0.7226 | 0.5 |
| 0.3281 | 7.0 | 21 | 0.6707 | 0.75 |
| 0.2276 | 8.0 | 24 | 0.7515 | 0.75 |
| 0.1674 | 9.0 | 27 | 0.6971 | 0.75 |
| 0.0873 | 10.0 | 30 | 0.5419 | 0.75 |
| 0.0525 | 11.0 | 33 | 0.5025 | 0.75 |
| 0.0286 | 12.0 | 36 | 0.5229 | 0.75 |
| 0.0149 | 13.0 | 39 | 0.5660 | 0.75 |
| 0.0082 | 14.0 | 42 | 0.6954 | 0.75 |
| 0.006 | 15.0 | 45 | 0.8649 | 0.75 |
| 0.0043 | 16.0 | 48 | 1.0011 | 0.75 |
| 0.0035 | 17.0 | 51 | 1.0909 | 0.75 |
| 0.0021 | 18.0 | 54 | 1.1615 | 0.75 |
| 0.0017 | 19.0 | 57 | 1.2147 | 0.75 |
| 0.0013 | 20.0 | 60 | 1.2585 | 0.75 |
| 0.0016 | 21.0 | 63 | 1.2917 | 0.75 |
### Framework versions
- Transformers 4.15.0
- Pytorch 1.10.2+cu102
- Datasets 1.18.2
- Tokenizers 0.10.3