gpt2-finetuned-mcqa-sciq2-safety

This model is a fine-tuned version of openai-community/gpt2 on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 1.7824
• Bleu: 0.0816
• Bertscore Precision: 0.4316
• Bertscore Recall: 0.4553
• Bertscore F1: 0.4431

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: 3e-05
• train_batch_size: 8
• eval_batch_size: 8
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 16
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 10

Training results

Training Loss	Epoch	Step	Validation Loss	Bleu	Bertscore Precision	Bertscore Recall	Bertscore F1
1.917	1.0	1460	1.8258	0.0856	0.4279	0.4541	0.4406
1.7299	2.0	2920	1.7928	0.0789	0.4311	0.4549	0.4427
1.6342	3.0	4380	1.7870	0.0858	0.4315	0.4542	0.4425
1.5534	4.0	5840	1.7824	0.0816	0.4316	0.4553	0.4431
1.4878	5.0	7300	1.7909	0.0797	0.4315	0.4556	0.4432
1.4554	6.0	8760	1.8021	0.0772	0.4312	0.4557	0.4431
1.3981	7.0	10220	1.8103	0.0767	0.4314	0.4558	0.4433
1.3712	8.0	11680	1.8242	0.0427	0.4315	0.4561	0.4434
1.3431	9.0	13140	1.8338	0.0416	0.4315	0.4560	0.4434
1.3354	10.0	14600	1.8390	0.0420	0.4315	0.4560	0.4434

Framework versions

• Transformers 4.40.2
• Pytorch 2.3.0+cu121
• Datasets 2.19.1
• Tokenizers 0.19.1