ViT-chess-V4 / README.md
Migga's picture
update model card README.md
42c638f
---
tags:
- generated_from_trainer
metrics:
- accuracy
model-index:
- name: ViT-chess-V4
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. -->
# ViT-chess-V4
This model is a fine-tuned version of [](https://huggingface.co/) on the None dataset.
It achieves the following results on the evaluation set:
- Loss: 4.2867
- Accuracy: 0.1942
## 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: 10
- eval_batch_size: 4
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 5
### Training results
| Training Loss | Epoch | Step | Validation Loss | Accuracy |
|:-------------:|:-----:|:------:|:---------------:|:--------:|
| 5.4877 | 1.0 | 45000 | 5.4554 | 0.1044 |
| 4.9794 | 2.0 | 90000 | 5.0001 | 0.1371 |
| 4.5956 | 3.0 | 135000 | 4.6720 | 0.1596 |
| 4.3402 | 4.0 | 180000 | 4.4082 | 0.1834 |
| 4.097 | 5.0 | 225000 | 4.2867 | 0.1942 |
### Framework versions
- Transformers 4.18.0
- Pytorch 1.11.0
- Datasets 2.1.0
- Tokenizers 0.12.1