hf-distilbert-imdb-mlm-cosine

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: 2.0939

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: 64
• eval_batch_size: 64
• seed: 420
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: cosine
• num_epochs: 20.0

Training results

Training Loss	Epoch	Step	Validation Loss
2.5054	1.0	383	2.3043
2.3885	2.0	766	2.2447
2.3418	3.0	1149	2.2319
2.3045	4.0	1532	2.1883
2.2772	5.0	1915	2.1893
2.2543	6.0	2298	2.1683
2.2308	7.0	2681	2.1454
2.2139	8.0	3064	2.1403
2.2008	9.0	3447	2.1165
2.1937	10.0	3830	2.1281
2.1778	11.0	4213	2.1189
2.1742	12.0	4596	2.1218
2.1611	13.0	4979	2.0996
2.1562	14.0	5362	2.0992
2.1508	15.0	5745	2.1013
2.1469	16.0	6128	2.0945
2.1437	17.0	6511	2.0899
2.1436	18.0	6894	2.0827
2.1443	19.0	7277	2.0935
2.1389	20.0	7660	2.0939

Framework versions

• Transformers 4.30.2
• Pytorch 2.0.1+cu117
• Datasets 2.13.1
• Tokenizers 0.13.3