metadata
license: mit
tags:
- generated_from_trainer
datasets:
- ag_news
metrics:
- accuracy
- f1
- precision
- recall
base_model: prajjwal1/bert-tiny
model-index:
- name: results
results:
- task:
type: text-classification
name: Text Classification
dataset:
name: ag_news
type: ag_news
config: default
split: train[:40000]
args: default
metrics:
- type: accuracy
value: 0.8951
name: Accuracy
- type: f1
value: 0.8964447542636089
name: F1
- type: precision
value: 0.8978261707981314
name: Precision
- type: recall
value: 0.896474840596734
name: Recall
results
This model is a fine-tuned version of prajjwal1/bert-tiny on the ag_news dataset. It achieves the following results on the evaluation set:
- Loss: 0.3320
- Accuracy: 0.8951
- F1: 0.8964
- Precision: 0.8978
- Recall: 0.8965
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: 0.0003
- train_batch_size: 64
- eval_batch_size: 64
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 3
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Accuracy | F1 | Precision | Recall |
|---|---|---|---|---|---|---|---|
| 0.2783 | 1.0 | 625 | 0.3046 | 0.8949 | 0.8960 | 0.8970 | 0.8963 |
| 0.1878 | 2.0 | 1250 | 0.3139 | 0.8954 | 0.8971 | 0.8995 | 0.8965 |
| 0.1311 | 3.0 | 1875 | 0.3320 | 0.8951 | 0.8964 | 0.8978 | 0.8965 |
Framework versions
- Transformers 4.22.0
- Pytorch 1.12.1+cu113
- Datasets 2.4.0
- Tokenizers 0.12.1