Palak's picture
Update README.md
81c283c
metadata
license: mit
tags:
  - generated_from_trainer
datasets:
  - squad
model-index:
  - name: xlm-roberta-base_squad
    results: []

xlm-roberta-base_squad

This model is a fine-tuned version of xlm-roberta-base on the squad dataset.

  • "eval_exact_match": 82.69631031220435
  • "eval_f1": 89.4562841806503
  • "eval_samples": 10918

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: 3e-05
  • train_batch_size: 32
  • eval_batch_size: 32
  • seed: 42
  • optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
  • lr_scheduler_type: linear
  • num_epochs: 3.0

Training results

Framework versions

  • Transformers 4.14.1
  • Pytorch 1.9.0
  • Datasets 1.16.1
  • Tokenizers 0.10.3