README.md

metadata

license: mit
base_model: microsoft/deberta-v3-xsmall
tags:
  - generated_from_trainer
datasets:
  - FineWebSentences
metrics:
  - accuracy
model-index:
  - name: Deberta-FineWebEdu
    results:
      - task:
          name: Masked Language Modeling
          type: fill-mask
        dataset:
          name: FineWebSentences
          type: FineWebSentences
        metrics:
          - name: Accuracy
            type: accuracy
            value: 0.4905470376215008

Deberta-FineWebEdu

This model is a fine-tuned version of microsoft/deberta-v3-xsmall on the FineWebSentences dataset. It achieves the following results on the evaluation set:

• Loss: 3.4314
• Accuracy: 0.4905

Model description

Finetuned on sentences from randomly chosen HuggingFaceFW/fineweb-edu entries.

Intended uses & limitations

To be finetuned on more tasks involving English sentences.

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

• learning_rate: 5e-05
• train_batch_size: 8
• eval_batch_size: 8
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 3.0

Training results

The evaluation and training losses were similar indicating no overfitting.

Framework versions

• Transformers 4.39.3
• Pytorch 2.3.0+cu121
• Datasets 2.18.0
• Tokenizers 0.15.2