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