albert-base-v2-finetuned-squad-v2

This model is a fine-tuned version of albert-base-v2 on the squad_v2 dataset. It achieves the following results on the evaluation set:

• Loss: 0.9645

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: 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: 3

Training results

Training Loss	Epoch	Step	Validation Loss
0.864	1.0	8248	0.8698
0.6246	2.0	16496	0.8351
0.4359	3.0	24744	0.9645

Performance

• 'exact': 78.36267160784975,
• 'f1': 81.72483834090231,
• 'total': 11873,
• 'HasAns_exact': 74.527665317139,
• 'HasAns_f1': 81.26164062441536,
• 'HasAns_total': 5928,
• 'NoAns_exact': 82.18671152228764,
• 'NoAns_f1': 82.18671152228764,
• 'NoAns_total': 5945,
• 'best_exact': 78.36267160784975,
• 'best_exact_thresh': 0.9990501403808594,
• 'best_f1': 81.72483834090268,
• 'best_f1_thresh': 0.9990501403808594,
• 'total_time_in_seconds': 224.37217425400013,
• 'samples_per_second': 52.9165438605555,
• 'latency_in_seconds': 0.018897681651983505

Framework versions

• Transformers 4.30.2
• Pytorch 2.0.1+cu118
• Datasets 2.13.1
• Tokenizers 0.13.3