qamodel_distilbert

This model is a fine-tuned version of distilbert-base-cased-distilled-squad on the subjqa dataset. It achieves the following results on the evaluation set:

• Loss: 1.7950

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: 1e-06
• 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: 18

Training results

Training Loss	Epoch	Step	Validation Loss
3.1563	1.0	81	2.6200
2.0431	2.0	162	2.1380
1.8432	3.0	243	2.0108
1.7601	4.0	324	1.9526
1.6957	5.0	405	1.9126
1.6477	6.0	486	1.8846
1.6173	7.0	567	1.8699
1.5799	8.0	648	1.8527
1.5749	9.0	729	1.8367
1.5422	10.0	810	1.8281
1.5353	11.0	891	1.8208
1.529	12.0	972	1.8116
1.5101	13.0	1053	1.8049
1.5005	14.0	1134	1.8018
1.4932	15.0	1215	1.8008
1.4895	16.0	1296	1.7976
1.4817	17.0	1377	1.7957
1.4695	18.0	1458	1.7950

Framework versions

• Transformers 4.26.1
• Pytorch 1.13.1+cu116
• Datasets 2.10.1
• Tokenizers 0.13.2