IMDb_data_subset-MLM_with-custom_collator-distilbert-base-uncased

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

• Loss: 3.2538
• Model Preparation Time: 0.0042

Model description

distilbert-base-uncased

Intended uses & limitations

Mask filling

Training and evaluation data

The IMDb dataset is tokenized, and words are masked with 0.2 probability. The resulting dataset is downsampled, resulting in 10,000 training samples and 1,000 validation samples.

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

• learning_rate: 2e-05
• train_batch_size: 64
• eval_batch_size: 64
• seed: 42
• optimizer: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: linear
• num_epochs: 5
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Model Preparation Time
3.5579	1.0	157	3.3058	0.0042
3.3945	2.0	314	3.2732	0.0042
3.3487	3.0	471	3.2542	0.0042
3.3088	4.0	628	3.2237	0.0042
3.2961	5.0	785	3.2538	0.0042

Framework versions

• Transformers 4.50.0
• Pytorch 2.6.0+cu124
• Datasets 3.4.1
• Tokenizers 0.21.1