README.md

---
license: bigscience-openrail-m
pipeline_tag: text-classification

widget:
  - example_title: "Very Positive"
    text: "This product is absolutely amazing!"
  - example_title: "Positive"
    text: "I like the features of this app."
  - example_title: "Somewhat Positive"
    text: "The service is pretty good."
  - example_title: "Neutral"
    text: "It meets my expectations."
  - example_title: "Somewhat Negative"
    text: "The experience was not as bad."
  - example_title: "Negative"
    text: "I'm disappointed with the quality."
  - example_title: "Very Negative"
    text: "This is the worst purchase I've ever made."

  - example_title: "Very Positive"
    text: "I recently bought a new sports jersey from this store, and it's absolutely amazing! The fabric is high quality, the fit is perfect, and the design looks even better in person. I'm thrilled with my purchase and highly recommend this store to anyone looking for custom sports apparel."

  - example_title: "Positive"
    text: "I tried out the new fitness app, and I like its features. It has a user-friendly interface, plenty of workout plans, and tracks my progress effectively. It has definitely added value to my fitness routine, and I find it motivating to use daily."

  - example_title: "Somewhat Positive"
    text: "The customer service at this cafe was pretty good. They were polite and attended to my needs promptly. The coffee tasted nice, and the atmosphere was cozy. I enjoyed my visit and would consider coming back, although there is room for improvement in the menu variety."

  - example_title: "Neutral"
    text: "The smartphone I purchased meets my expectations. It has all the features I need and performs adequately. The battery life is decent, and the camera quality is average. Overall, it serves its purpose well without any standout qualities or significant drawbacks."

  - example_title: "Somewhat Negative"
    text: "The hotel stay was not as bad as I had feared, but there were some issues. The room was clean, but the service was slow, and the amenities were limited. It wasn't a terrible experience, but I expected more for the price I paid."

  - example_title: "Negative"
    text: "I'm disappointed with the quality of this product. It feels flimsy and cheaply made. After only a few uses, it started showing signs of wear and tear. I expected better durability and performance based on the reviews. I wouldn't recommend this to others."

  - example_title: "Very Negative"
    text: "This is the worst purchase I've ever made. The item arrived broken, and the customer service was unresponsive. It was a complete waste of money, and I regret buying it. The overall experience was frustrating and disappointing. Avoid this product at all costs."


---
Multi-label sentiment classification model developed by [Dejan Marketing](https://dejanmarketing.com/).

The model is designed to be deployed in an automated pipeline capable of classifying text sentiment for thousands (or even millions) of text chunks or as a part of a scraping pipeline.

This is a demo model which may occassionally misclasify some texts. In a typical commercial project, a larger model is deployed for the task, and in special cases, a domain-specific model is developed for the client.

# Engage Our Team
Interested in using this in an automated pipeline for bulk query processing?

Please [book an appointment](https://dejanmarketing.com/conference/) to discuss your needs.

# Base Model

albert/albert-base-v2


## Labels

sentiment_labels = {
    0: "very positive",
    1: "positive",
    2: "somewhat positive",
    3: "neutral",
    4: "somewhat negative",
    5: "negative",
    6: "very negative"
}

# Sources of Training Data

Synthetic. Llama3.