tinybert_base_train_kd

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

• Loss: 62.6034
• Accuracy: 0.5511

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: 0.0001
• train_batch_size: 96
• eval_batch_size: 96
• seed: 10
• optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 10000
• num_epochs: 25

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
447.3158	4.1982	10000	428.8420	0.1658
151.7919	8.3963	20000	135.7859	0.4816
95.9257	12.5945	30000	84.8065	0.5308
78.7736	16.7926	40000	70.8755	0.5468
70.9704	20.9908	50000	63.0908	0.5510

Framework versions

• Transformers 4.51.2
• Pytorch 2.6.0+cu126
• Datasets 3.5.0
• Tokenizers 0.21.1