|
--- |
|
tags: |
|
- generated_from_trainer |
|
datasets: |
|
- squad |
|
model-index: |
|
- name: ft_electra-small-turkish-uncased-discriminator_lr-2e-1_epochs-1 |
|
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. --> |
|
|
|
This model is a fine-tuned version of [loodos/electra-small-turkish-uncased-discriminator](https://huggingface.co/loodos/electra-small-turkish-uncased-discriminator) on the squad dataset. |
|
It achieves the following results on the evaluation set: |
|
- Loss: 5.9506 |
|
|
|
## 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: 0.2 |
|
- train_batch_size: 16 |
|
- eval_batch_size: 16 |
|
- seed: 42 |
|
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08 |
|
- lr_scheduler_type: linear |
|
- num_epochs: 1 |
|
|
|
### Training results |
|
|
|
| Training Loss | Epoch | Step | Validation Loss | |
|
|:-------------:|:-----:|:----:|:---------------:| |
|
| 5.951 | 1.0 | 5818 | 5.9506 | |
|
|
|
|
|
### Framework versions |
|
|
|
- Transformers 4.15.0 |
|
- Pytorch 1.10.0+cu111 |
|
- Datasets 1.17.0 |
|
- Tokenizers 0.10.3 |
|
|
