recipe-distilbert-s

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: 1.0321

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
1.8594	1.0	844	1.4751
1.4763	2.0	1688	1.3282
1.3664	3.0	2532	1.2553
1.2975	4.0	3376	1.2093
1.2543	5.0	4220	1.1667
1.2189	6.0	5064	1.1472
1.1944	7.0	5908	1.1251
1.1737	8.0	6752	1.1018
1.1549	9.0	7596	1.0950
1.1387	10.0	8440	1.0796
1.1295	11.0	9284	1.0713
1.1166	12.0	10128	1.0639
1.1078	13.0	10972	1.0485
1.099	14.0	11816	1.0431
1.0951	15.0	12660	1.0425
1.0874	16.0	13504	1.0323
1.0828	17.0	14348	1.0368
1.0802	18.0	15192	1.0339
1.0798	19.0	16036	1.0247
1.0758	20.0	16880	1.0321

Framework versions

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