Odia Sentiment Classifier v4

Fine-tuned google/muril-base-cased for Odia language sentiment analysis.

Built with ❤️ for 45 million Odia speakers 🇮🇳 Dedicated to Lord Jagannath and Lord Lingaraj (We are from old town :) Parakaran Sahi) and the people of Odisha 🙏

Results

Metric Score
Accuracy 0.838 (83.8%)
F1 Score 0.834
Samples 37 test samples

Previous versions

Version Accuracy
v1 ~33%
v2 ~30% (auto-label bug)
v3 77.8%
v4 83.8%

Usage

from transformers import pipeline
pipe = pipeline("text-classification", model="Baps24/odia-sentiment-muril-v4")
pipe("ଏହି ଉତ୍ପାଦ ଭଲ")   # Positive
pipe("ଏ ଜିନିଷ ଖରାପ")     # Negative
pipe("ଠିକ ଅଛି")           # Neutral

Community

This model is part of the Odias in ML initiative. https://odisha-ml.github.io

Downloads last month
27
Safetensors
Model size
0.2B params
Tensor type
F32
·
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support

Model tree for Baps24/odia-sentiment-muril-v4

Finetuned
(50)
this model