kobigbird-pure45-62009834

This model is a fine-tuned version of monologg/kobigbird-bert-base on the custom_squad_v2 dataset. It achieves the following results on the evaluation set:

• Loss: 2.2563

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: 0.0005
• train_batch_size: 64
• eval_batch_size: 64
• seed: 45
• gradient_accumulation_steps: 8
• total_train_batch_size: 512
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: cosine
• num_epochs: 6
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss
No log	0.99	21	1.5632
No log	1.99	42	1.2859
No log	2.99	63	1.3649
No log	3.99	84	1.8906
No log	4.99	105	2.1430
No log	5.99	126	2.2563

Framework versions

• Transformers 4.25.1
• Pytorch 1.13.0+cu116
• Datasets 2.8.0
• Tokenizers 0.13.2