deberta-base-squad / README.md
jimypbr's picture
Update README.md
a9d25bf
|
raw
history blame
1.25 kB
metadata
license: apache2.0
tags:
  - generated_from_trainer
datasets:
  - squad
model-index:
  - name: deberta-base-squad
    results: []

deberta-base-squad

This model is a fine-tuned version of microsoft/deberta-base on the squad dataset.

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: 6e-05
  • train_batch_size: 1
  • eval_batch_size: 1
  • seed: 1984
  • distributed_type: IPU
  • gradient_accumulation_steps: 32
  • total_train_batch_size: 128
  • total_eval_batch_size: 16
  • optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
  • lr_scheduler_type: linear
  • lr_scheduler_warmup_ratio: 0.25
  • num_epochs: 2.0
  • training precision: Mixed Precision

Training results

Framework versions

  • Transformers 4.19.0.dev0
  • Pytorch 1.10.0+cpu
  • Datasets 2.0.0
  • Tokenizers 0.11.6