rama9agg/distilbert-sentiment-tone-analyzer

DistilBERT Sentiment & Tone Analyzer

A lightweight, fine-tuned DistilBERT model designed for dual-head sequence classification. It simultaneously predicts core sentiment (Positive/Negative) and conversational tone across text inputs such as product reviews, user comments, social posts, and quotes.

---

🎭 Model Details & Classes

The model utilizes a shared distilbert-base-uncased backbone with two custom classification heads built on top of the pooled [CLS] embeddings:

1. Sentiment Classification (2 Classes)

   • 0: NEGATIVE
   • 1: POSITIVE

2. Tone Classification (4 Classes)

   • 0: MOTIVATIONAL
   • 1: SARCASTIC
   • 2: FUNNY
   • 3: LOVE QUOTE

---

⚡ Quickstart & Usage

Since this model uses a custom multi-task architecture, define the architecture class and load the weights using safetensors and huggingface_hub.

Prerequisites

pip install torch transformers safetensors huggingface_hub