Rubert Base Cased Russian Sentiment

🚀 RuBERT for Russian Sentiment Classification

This project presents a fine - tuned RuBERT model designed for the sentiment classification of short Russian texts. It addresses a multi - class classification task with the following labels:

0: neutral
1: positive
2: negative

The mapping from English labels to Russian labels is as follows:

neutral: нейтральный
positive: позитивный
negative: негативный

🚀 Quick Start

💻 Usage Examples

Basic Usage

from transformers import pipeline
model = pipeline(model="seara/rubert-base-cased-russian-sentiment")
model("Привет, ты мне нравишься!")
# [{'label': 'positive', 'score': 0.9818321466445923}]

📚 Documentation

📦 Datasets

This model was trained on the union of the following datasets:

• Kaggle Russian News Dataset
• Linis Crowd 2015
• Linis Crowd 2016
• RuReviews
• RuSentiment

An overview of the training data can be found on S. Smetanin Github repository.

Download links for all Russian sentiment datasets collected by Smetanin can be found in this repository.

🔧 Training Details

Training was conducted in this project with the following parameters:

tokenizer.max_length: 256
batch_size: 32
optimizer: adam
lr: 0.00001
weight_decay: 0
epochs: 2

The Train/validation/test splits are 80%/10%/10%.

📊 Eval results (on test split)

📄 License

This project is released under the MIT license.