Model Card for distilbert-base-uncased-finetuned-amazon-reviews

Table of Contents

• Model Card for distilbert-base-uncased-finetuned-amazon-reviews
• Table of Contents
• Model Details
• Uses
• Fine-tuning hyperparameters
• Evaluation
• Framework versions

Model Details

Model Description

This model is a fine-tuned version of distilbert-base-uncased on amazon_reviews_multi dataset. This model reaches an accuracy of xxx on the dev set.

• Model type: Language model
• Language(s) (NLP): en
• License: apache-2.0
• Parent Model: For more details about DistilBERT, check out this model card.
• Resources for more information:
  • Model Documentation

Uses

You can use this model directly with a pipeline for text classification.

from transformers import pipeline

checkpoint = "amir7d0/distilbert-base-uncased-finetuned-amazon-reviews"
classifier = pipeline("text-classification", model=checkpoint)
classifier(["Replace me by any text you'd like."])

and in TensorFlow:

from transformers import AutoTokenizer, TFAutoModelForSequenceClassification

checkpoint = "amir7d0/distilbert-base-uncased-finetuned-amazon-reviews"
tokenizer = AutoTokenizer.from_pretrained(checkpoint)
model = TFAutoModelForSequenceClassification.from_pretrained(checkpoint)

text = "Replace me by any text you'd like."
encoded_input = tokenizer(text, return_tensors='tf')
output = model(encoded_input)

Training Details

Training and Evaluation Data

Here is the raw dataset (amazon_reviews_multi) we used for finetuning the model. The dataset contains 200,000, 5,000, and 5,000 reviews in the training, dev, and test sets respectively.

Fine-tuning hyperparameters

The following hyperparameters were used during training:

• learning_rate: 2e-05
• train_batch_size: 16
• eval_batch_size: 16
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 5

Accuracy

The fine-tuned model was evaluated on the test set of amazon_reviews_multi.

• Accuracy (exact) is the exact match of the number of stars.
• Accuracy (off-by-1) is the percentage of reviews where the number of stars the model predicts differs by a maximum of 1 from the number given by the human reviewer.

Split	Accuracy (exact)	Accuracy (off-by-1)
Dev set	56.96%	85.50%
Test set	57.36%	85.58%

Framework versions

• Transformers 4.26.1
• TensorFlow 2.11.0
• Datasets 2.1.0
• Tokenizers 0.13.2