recipe-distilbert-upper-Is

This model is a fine-tuned version of distilbert-base-uncased on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.8565

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: 2e-05
• train_batch_size: 256
• eval_batch_size: 256
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 20
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss
1.6309	1.0	1305	1.2607
1.2639	2.0	2610	1.1291
1.1592	3.0	3915	1.0605
1.0987	4.0	5220	1.0128
1.0569	5.0	6525	0.9796
1.0262	6.0	7830	0.9592
1.0032	7.0	9135	0.9352
0.9815	8.0	10440	0.9186
0.967	9.0	11745	0.9086
0.9532	10.0	13050	0.8973
0.9436	11.0	14355	0.8888
0.9318	12.0	15660	0.8835
0.9243	13.0	16965	0.8748
0.9169	14.0	18270	0.8673
0.9117	15.0	19575	0.8610
0.9066	16.0	20880	0.8562
0.9028	17.0	22185	0.8566
0.901	18.0	23490	0.8583
0.8988	19.0	24795	0.8557
0.8958	20.0	26100	0.8565

Framework versions

• Transformers 4.19.0.dev0
• Pytorch 1.11.0+cu102
• Datasets 2.3.2
• Tokenizers 0.12.1