Shijia's picture
End of training
1eb72c1 verified
metadata
license: mit
base_model: FacebookAI/xlm-roberta-base
tags:
  - generated_from_trainer
model-index:
  - name: xlm-roberta-base_kin_corr_2e-05
    results: []

xlm-roberta-base_kin_corr_2e-05

This model is a fine-tuned version of FacebookAI/xlm-roberta-base on the None dataset. It achieves the following results on the evaluation set:

  • Loss: 0.0213
  • Spearman Corr: 0.7790

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: 32
  • eval_batch_size: 128
  • seed: 42
  • gradient_accumulation_steps: 2
  • total_train_batch_size: 64
  • optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
  • lr_scheduler_type: linear
  • num_epochs: 30
  • mixed_precision_training: Native AMP

Training results

Training Loss Epoch Step Validation Loss Spearman Corr
No log 0.89 200 0.0210 0.7800
No log 1.78 400 0.0210 0.7820
0.0006 2.67 600 0.0208 0.7800
0.0006 3.56 800 0.0209 0.7822
0.0006 4.45 1000 0.0212 0.7801
0.0006 5.35 1200 0.0214 0.7792
0.0005 6.24 1400 0.0213 0.7768
0.0005 7.13 1600 0.0211 0.7803
0.0005 8.02 1800 0.0210 0.7785
0.0005 8.91 2000 0.0213 0.7783
0.0005 9.8 2200 0.0211 0.7809
0.0005 10.69 2400 0.0213 0.7790

Framework versions

  • Transformers 4.37.2
  • Pytorch 2.2.0+cu121
  • Datasets 2.17.0
  • Tokenizers 0.15.2