rob-base-superqa1 / README.md
nbroad's picture
nbroad HF staff
Add evaluation results on the default config of quoref (#3)
e49ab07
metadata
license: mit
tags:
  - generated_from_trainer
model-index:
  - name: rob-base-superqa
    results:
      - task:
          type: question-answering
          name: Question Answering
        dataset:
          name: adversarial_qa
          type: adversarial_qa
          config: adversarialQA
          split: validation
        metrics:
          - name: Exact Match
            type: exact_match
            value: 43.8667
            verified: true
          - name: F1
            type: f1
            value: 55.135
            verified: true
      - task:
          type: question-answering
          name: Question Answering
        dataset:
          name: squad_v2
          type: squad_v2
          config: squad_v2
          split: validation
        metrics:
          - name: Exact Match
            type: exact_match
            value: 79.2432
            verified: true
          - name: F1
            type: f1
            value: 82.336
            verified: true
      - task:
          type: question-answering
          name: Question Answering
        dataset:
          name: quoref
          type: quoref
          config: default
          split: validation
        metrics:
          - name: Exact Match
            type: exact_match
            value: 78.8581
            verified: true
          - name: F1
            type: f1
            value: 82.8261
            verified: true
task:
  - question-answering
datasets:
  - squad_v2
  - quoref
  - adversarial_qa
  - duorc

rob-base-superqa

This model is a fine-tuned version of roberta-base on the None 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: 7e-05
  • train_batch_size: 32
  • eval_batch_size: 32
  • seed: 42
  • distributed_type: multi-GPU
  • num_devices: 8
  • total_train_batch_size: 256
  • total_eval_batch_size: 256
  • optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-06
  • lr_scheduler_type: linear
  • num_epochs: 3.0

Training results

Framework versions

  • Transformers 4.21.1
  • Pytorch 1.11.0a0+gita4c10ee
  • Datasets 2.4.0
  • Tokenizers 0.12.1