lmind_nq_train600_eval300_v1_docidx_gpt2-xl

This model is a fine-tuned version of gpt2-xl on the tyzhu/lmind_nq_train600_eval300_v1_docidx dataset. It achieves the following results on the evaluation set:

• Loss: 0.3697
• Accuracy: 0.8462

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: 10.0

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
2.4931	0.5	28	2.3904	0.5749
2.4699	1.0	56	2.1014	0.6050
1.9006	1.5	84	1.8079	0.6393
1.9317	2.0	112	1.5510	0.6722
1.3984	2.5	140	1.2850	0.7075
1.3662	3.0	168	1.0900	0.7374
0.9041	3.5	196	0.8909	0.7670
0.9056	4.0	224	0.7502	0.7903
0.6131	4.5	252	0.6304	0.8067
0.6101	5.0	280	0.5429	0.8215
0.3772	5.5	308	0.4872	0.8287
0.4265	6.0	336	0.4437	0.8357
0.2552	6.5	364	0.4226	0.8389
0.2875	7.0	392	0.4019	0.8418
0.1874	7.5	420	0.3965	0.8430
0.1958	8.0	448	0.3812	0.8441
0.1443	8.5	476	0.3852	0.8450
0.1535	9.0	504	0.3791	0.8456
0.1236	9.5	532	0.3849	0.8456
0.1221	10.0	560	0.3697	0.8462

Framework versions

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