L3Cube-MahaEmotions

L3Cube-MahaEmotions

MH Specific

L3Cube-MahaEmotions dataset for language nlp.

Build an emotion-aware Marathi chatbot for mental health support that adapts its responses based on detected user emotions.
HomepageGitHub

Quick Start

from datasets import load_dataset
ds = load_dataset('l3cube-pune/MahaEmotions')
from collections import Counter
labels = [ex['label'] for ex in ds['train']]
print(f"Emotion distribution: {dict(Counter(labels).most_common())}")
Modality
text
Size
15,000 samples, 11 emotion labels
License
Format
CSV
Language
mr
Update Frequency
static
Organization
L3Cube, Pune
Category
language-nlp

Schema

FieldTypeDescription
textstringMarathi text sample
labelstringEmotion category (one of 11 labels: happy, sad, angry, fearful, surprised, etc.)

Build With This

Create a student wellbeing monitor for Marathi-medium schools that detects distress signals in written assignments
Develop a customer experience analytics tool that goes beyond sentiment to identify specific emotions in Marathi feedback
Build an emotion-trend dashboard that tracks the emotional pulse of Marathi social media during festivals, elections, and crises

AI Use Cases

Emotion recognition
Last verified: 2026-03-07