bert-qwantz-coherent

This model is a fine-tuned version of google-bert/bert-base-uncased on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.6861
• Accuracy: 0.8240

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: 0.0001
• train_batch_size: 24
• eval_batch_size: 24
• 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	Accuracy
0.4695	1.0	339	0.4547	0.7956
0.2521	2.0	678	0.4364	0.8131
0.0627	3.0	1017	0.6861	0.8240

Can save 90% of coherent strings by discarding 80% of dp strings (cutoff is 57.403409481048584)
Can save 95% of coherent strings by discarding 63% of dp strings (cutoff is -83.01011323928833)
Can save 98% of coherent strings by discarding 44% of dp strings (cutoff is -97.15004563331604)
Can save 99% of coherent strings by discarding 33% of dp strings (cutoff is -98.31664562225342)

Framework versions

• Transformers 4.40.1
• Pytorch 2.2.1+cu121
• Datasets 2.19.0
• Tokenizers 0.19.1