bert-finetuned-combo-textbook-no-context

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

• Loss: 1.6241

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: 16
• eval_batch_size: 8
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 20

Training results

Training Loss	Epoch	Step	Validation Loss
2.4652	1.0	2444	2.2878
2.2439	2.0	4888	2.2005
2.1422	3.0	7332	2.1331
2.0527	4.0	9776	2.0734
1.9734	5.0	12220	2.0488
1.8752	6.0	14664	1.9844
1.7757	7.0	17108	1.9257
1.7478	8.0	19552	1.8901
1.6469	9.0	21996	1.8797
1.5894	10.0	24440	1.8198
1.5341	11.0	26884	1.7803
1.4563	12.0	29328	1.7602
1.4385	13.0	31772	1.7663
1.3534	14.0	34216	1.7007
1.3007	15.0	36660	1.6705
1.2689	16.0	39104	1.6996
1.2268	17.0	41548	1.6319
1.2152	18.0	43992	1.6599
1.1317	19.0	46436	1.6433
1.1133	20.0	48880	1.6241

Framework versions

• Transformers 4.38.2
• Pytorch 1.13.1+cu116
• Datasets 2.18.0
• Tokenizers 0.15.2