distilbert_add_pre-training-dim-96

This model is a fine-tuned version of distilbert-base-uncased on the wikitext wikitext-103-raw-v1 dataset. It achieves the following results on the evaluation set:

• Loss: 6.6092
• Accuracy: 0.1494

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: 5e-05
• train_batch_size: 64
• eval_batch_size: 64
• seed: 10
• distributed_type: multi-GPU
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 100
• num_epochs: 15
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
14.685	1.0	3573	9.3922	0.1240
8.0255	2.0	7146	7.1510	0.1315
7.0152	3.0	10719	6.7861	0.1482
6.8127	4.0	14292	6.7053	0.1493
6.74	5.0	17865	6.6695	0.1474
6.7067	6.0	21438	6.6431	0.1491
6.6871	7.0	25011	6.6204	0.1483
6.6748	8.0	28584	6.6250	0.1473
6.6649	9.0	32157	6.6108	0.1486
6.6596	10.0	35730	6.6140	0.1497
6.6536	11.0	39303	6.6067	0.1493
6.6483	12.0	42876	6.6140	0.1489
6.6463	13.0	46449	6.6096	0.1484
6.6434	14.0	50022	6.5570	0.1526
6.6414	15.0	53595	6.5836	0.1526

Framework versions

• Transformers 4.26.0
• Pytorch 1.14.0a0+410ce96
• Datasets 2.9.0
• Tokenizers 0.13.2