willherbert27/electra-finetuned-combo-with-context

electra-finetuned-combo-textbook

This model is a fine-tuned version of google/electra-small-generator on the None dataset. It achieves the following results on the evaluation set:

• Loss: 2.0465

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
3.1458	1.0	2468	2.8891
2.8707	2.0	4936	2.7016
2.7021	3.0	7404	2.5199
2.6068	4.0	9872	2.4902
2.5111	5.0	12340	2.4149
2.4193	6.0	14808	2.3743
2.383	7.0	17276	2.3333
2.3082	8.0	19744	2.2770
2.2723	9.0	22212	2.2606
2.2151	10.0	24680	2.2444
2.1869	11.0	27148	2.1859
2.1588	12.0	29616	2.1959
2.1188	13.0	32084	2.1490
2.0947	14.0	34552	2.1259
2.06	15.0	37020	2.1211
1.9976	16.0	39488	2.0701
1.9975	17.0	41956	2.1048
1.9479	18.0	44424	2.1072
1.9215	19.0	46892	2.0794
1.9426	20.0	49360	2.0465

Framework versions

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