Kaludi/Reviews-Sentiment-Analysis

Reviews Sentiment Analysis

A tool that analyzes the overall sentiment of customer reviews for a specific product or service, whether it’s positive or negative. This analysis is performed by using natural language processing algorithms and machine learning from the model ‘Reviews-Sentiment-Analysis’ trained by Kaludi, allowing businesses to gain valuable insights into customer satisfaction and improve their products and services accordingly.

Training Procedure

• learning_rate = 1e-5
• batch_size = 32
• warmup = 600
• max_seq_length = 128
• num_train_epochs = 10.0

Validation Metrics

• Loss: 0.159
• Accuracy: 0.952
• Precision: 0.965
• Recall: 0.938
• AUC: 0.988
• F1: 0.951

Usage

You can use cURL to access this model:

$ curl -X POST -H "Authorization: Bearer YOUR_API_KEY" -H "Content-Type: application/json" -d '{"inputs": "I don't feel like you trust me to do my job."}' https://api-inference.huggingface.co/models/Kaludi/Reviews-Sentiment-Analysis

Or Python API:

from transformers import AutoModelForSequenceClassification, AutoTokenizer

model = AutoModelForSequenceClassification.from_pretrained("Kaludi/Reviews-Sentiment-Analysis", use_auth_token=True)

tokenizer = AutoTokenizer.from_pretrained("Kaludi/Reviews-Sentiment-Analysis", use_auth_token=True)

inputs = tokenizer("I don't feel like you trust me to do my job.", return_tensors="pt")

outputs = model(**inputs)