VaisakhKrishna/Emotional_Sentiment_Analysis

Emotional Sentiment Analysis Dataset for LLaMA-2 Fine-tuning (The formatted version can be directly used for fine tuning which contain only the formatted text, while the dataset.csv contain all the text, emotion, response and the formatted text)

This dataset contains conversational data for training and fine-tuning language models for emotional sentiment analysis and response generation. The dataset includes user inputs, their corresponding emotional states, and tailored chatbot responses generated by the Gemma2 model, formatted for training LLaMA-2 models.

Dataset Overview

The dataset consists of four columns:

Text: User input, which is a statement or query representing the user's emotions, sourced from the GoEmotions dataset. Emotion: The emotional state of the user, which is classified into various categories such as sadness, joy, anger, etc. Response: A tailored response generated by the Gemma2 model, addressing the user's input based on their identified emotion. Formatted: A combination of the "Text", "Emotion", and "Response" columns, formatted in a structured way to make it suitable for fine-tuning the LLaMA-2 chat model.

Columns Explanation

1. 1.Text Description: The user’s statement or input. This column contains conversational text, which can represent a wide range of emotions and scenarios. Example: "I feel so alone and sad today."
2. 2.Emotion Description: The emotional state of the user inferred from the text. Emotions are classified into predefined categories such as sadness, happiness, anger, etc. Example: "sadness"
3. 3.Response Description: The response generated by the Gemma2 model, tailored according to the emotion identified in the "Emotion" column. This response is empathetic and context-aware. Example: "I'm really sorry you're feeling that way. It's tough to experience such emotions. Remember, you're not alone."
4. 4.Formatted Description: A combination of the "Text", "Emotion", and "Response" columns, formatted as per LLaMA-2’s fine-tuning requirements.

Purpose This dataset was created to train models for emotional sentiment analysis and response generation. By fine-tuning the LLaMA-2 chat model with this data, the model can generate emotionally-aware and contextually appropriate responses to user inputs.

Key Features: Emotion Classification: Each user input is associated with an emotion, allowing the model to understand the sentiment behind the input. Tailored Responses: Responses are generated by the Gemma2 model, ensuring they are relevant to the emotional context. LLaMA-2 Ready: The data is formatted in a way that is suitable for fine-tuning the LLaMA-2 chat model.

Usage You can use this dataset to fine-tune the LLaMA-2 chat model on Hugging Face. The "Formatted" column is designed for this purpose, and the dataset provides a comprehensive set of examples for training and evaluating emotional sentiment analysis models.

How to Load the Dataset You can load the dataset using the Hugging Face datasets library as shown below: python

from datasets import load_dataset

dataset = loaddataset('VaisakhKrishna/EmotionalSentiment_Analysis') print(dataset)

Intended Use Fine-tuning the LLaMA-2 model for emotional sentiment analysis and adaptive response generation. Building emotionally intelligent chatbots capable of understanding and responding to user emotions. Enhancing conversational agents to be more empathetic and contextually relevant.

Citation If you use this dataset in your work, please cite it as follows:

@dataset{Emotional Sentiment Analysis, author = {VaisakhKrishna}, title = {Emotional Sentiment Analysis Dataset for LLaMA-2 Fine-tuning}, year = {2024}, url = {https://huggingface.co/datasets/VaisakhKrishna/EmotionalSentimentAnalysis} }

License This dataset is available under the Llama2 license. Please refer to the LICENSE file for more details.