lmind_hotpot_train1000_eval500_v1_recite_qa_gpt2-xl

This model is a fine-tuned version of gpt2-xl on the tyzhu/lmind_hotpot_train1000_eval500_v1_recite_qa dataset. It achieves the following results on the evaluation set:

• Loss: 0.4181
• Accuracy: 0.7496

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	Validation Loss	Accuracy
1.9823	1.0	248	1.6149	0.6398
1.3868	2.0	496	1.1929	0.6700
0.9102	3.0	744	0.8513	0.6992
0.5869	4.0	992	0.6181	0.7218
0.364	5.0	1240	0.4963	0.7352
0.2706	6.0	1488	0.4413	0.7420
0.1906	7.0	1736	0.4085	0.7458
0.1474	8.0	1984	0.4010	0.7477
0.1264	9.0	2232	0.3979	0.7484
0.1091	10.0	2480	0.4060	0.7487
0.1072	11.0	2728	0.4050	0.7490
0.0969	12.0	2976	0.4081	0.7492
0.0935	13.0	3224	0.4145	0.7495
0.0932	14.0	3472	0.4078	0.7494
0.0929	15.0	3720	0.4140	0.7494
0.0951	16.0	3968	0.4145	0.7495
0.0926	17.0	4216	0.4134	0.7495
0.0946	18.0	4464	0.4257	0.7493
0.0897	19.0	4712	0.4164	0.7496
0.092	20.0	4960	0.4181	0.7496

Framework versions

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