lmind_hotpot_train5000_eval5000_v1_recite_qa_gpt2-xl

This model is a fine-tuned version of gpt2-xl on the tyzhu/lmind_hotpot_train5000_eval5000_v1_recite_qa dataset. It achieves the following results on the evaluation set:

• Loss: 0.4522
• Accuracy: 0.7661

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: 16
• eval_batch_size: 16
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: constant
• num_epochs: 20.0

Training results

Training Loss	Epoch	Step	Accuracy	Validation Loss
1.9686	1.0	1452	0.6681	1.5662
1.4604	2.0	2904	0.6915	1.2069
1.0522	3.0	4356	0.7149	0.9046
0.7455	4.0	5808	0.7341	0.6844
0.5307	5.0	7260	0.7475	0.5420
0.3796	6.0	8712	0.7560	0.4609
0.2912	7.0	10164	0.7604	0.4311
0.2282	8.0	11616	0.7627	0.4184
0.1905	9.0	13068	0.7640	0.4136
0.1687	10.0	14520	0.7648	0.4175
0.1553	11.0	15972	0.7651	0.4212
0.1447	12.0	17424	0.7653	0.4283
0.1388	13.0	18876	0.7656	0.4287
0.1329	14.0	20328	0.7657	0.4349
0.1292	15.0	21780	0.7657	0.4353
0.1267	16.0	23232	0.7659	0.4383
0.1251	17.0	24684	0.4416	0.7661
0.1201	18.0	26136	0.4467	0.7659
0.1186	19.0	27588	0.4508	0.7660
0.1176	20.0	29040	0.4522	0.7661

Framework versions

• Transformers 4.34.0
• Pytorch 2.1.0+cu121
• Datasets 2.14.5
• Tokenizers 0.14.1