JeremiahZ's picture
Add evaluation results on the mrpc config and validation split of glue (#1)
47ac4af
|
raw
history blame
2.9 kB
metadata
language:
  - en
license: apache-2.0
tags:
  - generated_from_trainer
datasets:
  - glue
metrics:
  - accuracy
  - f1
model-index:
  - name: bert-base-uncased-mrpc
    results:
      - task:
          name: Text Classification
          type: text-classification
        dataset:
          name: GLUE MRPC
          type: glue
          args: mrpc
        metrics:
          - name: Accuracy
            type: accuracy
            value: 0.8578431372549019
          - name: F1
            type: f1
            value: 0.9023569023569024
      - task:
          type: natural-language-inference
          name: Natural Language Inference
        dataset:
          name: glue
          type: glue
          config: mrpc
          split: validation
        metrics:
          - name: Accuracy
            type: accuracy
            value: 0.8578431372549019
            verified: true
          - name: Precision
            type: precision
            value: 0.8507936507936508
            verified: true
          - name: Recall
            type: recall
            value: 0.9605734767025089
            verified: true
          - name: AUC
            type: auc
            value: 0.8931260592926008
            verified: true
          - name: F1
            type: f1
            value: 0.9023569023569024
            verified: true
          - name: loss
            type: loss
            value: 0.5572634935379028
            verified: true

bert-base-uncased-mrpc

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

  • Loss: 0.5572
  • Accuracy: 0.8578
  • F1: 0.9024
  • Combined Score: 0.8801

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: 2e-05
  • train_batch_size: 16
  • eval_batch_size: 8
  • seed: 42
  • optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
  • lr_scheduler_type: linear
  • lr_scheduler_warmup_ratio: 0.06
  • num_epochs: 5.0

Training results

Training Loss Epoch Step Validation Loss Accuracy F1 Combined Score
No log 1.0 230 0.4111 0.8088 0.8704 0.8396
No log 2.0 460 0.3762 0.8480 0.8942 0.8711
0.4287 3.0 690 0.5572 0.8578 0.9024 0.8801
0.4287 4.0 920 0.6087 0.8554 0.8977 0.8766
0.1172 5.0 1150 0.6524 0.8456 0.8901 0.8678

Framework versions

  • Transformers 4.20.0.dev0
  • Pytorch 1.11.0+cu113
  • Datasets 2.1.0
  • Tokenizers 0.12.1