<!-- This model card has been generated automatically according to the information the Trainer had access to. You should probably proofread and complete it, then remove this comment. -->

results-yelp

This model is a fine-tuned version of textattack/bert-base-uncased-yelp-polarity on a filtered and manually reviewed Yelp dataset containing restaurant reviews only. It achieves the following results on the evaluation set:

• Loss: 0.3563
• Accuracy: 0.9302
• Precision: 0.9461
• Recall: 0.9608
• F1: 0.9534

Note: to use this tokenizer, please use the following code to load all the required files:

tokenizer = AutoTokenizer.from_pretrained("bert-base-uncased", config=AutoConfig.from_pretrained("potatobunny/results-yelp"))

Model description

This model is fine-tuned on a Yelp dataset with labelled data containing a restaurant review (text) and whether it has a positive (1) or negative (0) sentiment.

Intended uses & limitations

This is intended to perform text classification, specifically sentiment analysis, on text data obtained from restaurant reviews to determine if the particular review is positive or negative.

Training and evaluation data

The training and evaluation data were both obtained from the same Yelp dataset. The data was split into 70% training and 30% validation.

<!-- ## Training procedure -->

Training hyperparameters

The following hyperparameters were used during training:

• learning_rate: 5e-05
• train_batch_size: 16
• eval_batch_size: 64
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 500
• num_epochs: 3

Training results

The training loss obtained was 0.265741667.

Framework versions

• Transformers 4.18.0
• Pytorch 1.10.0+cu111
• Datasets 2.0.0
• Tokenizers 0.12.1