bert-finetuned-japanese-sentiment
This model is a fine-tuned version of cl-tohoku/bert-base-japanese-v2 on product amazon reviews japanese dataset.
Model Train for amazon reviews Japanese sentence sentiments.
Sentiment analysis is a common task in natural language processing. It consists of classifying the polarity of a given text at the sentence or document level. For instance, the sentence "The food is good" has a positive sentiment, while the sentence "The food is bad" has a negative sentiment.
In this model, we fine-tuned a BERT model on a Japanese sentiment analysis dataset. The dataset contains 20,000 sentences extracted from Amazon reviews. Each sentence is labeled as positive, neutral, or negative. The model was trained for 5 epochs with a batch size of 16.
- Epochs: 6 - Training Loss: 0.087600 - Validation Loss: 1.028876 - Accuracy: 0.813202 - Precision: 0.712440 - Recall: 0.756031 - F1: 0.728455
The following hyperparameters were used during training:
- learningrate: 2e-05 - trainbatchsize: 16 - evalbatchsize: 16 - seed: 0 - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08 - lrschedulertype: linear - numepochs: 6
- Transformers 4.27.4 - Pytorch 2.0.0+cu118 - Tokenizers 0.13.2