Indic Sentence Bert Nli

🚀 IndicSBERT

IndicSBERT is a model based on MuRIL (google/muril-base-cased), trained on the NLI dataset of ten major Indian languages. It supports multiple languages including English, Hindi, Marathi, etc., and has cross - lingual capabilities.

📋 Model Information

🔍 Example Widgets

• Monolingual - Marathi
  • Source Sentence: दिवाळी आपण मोठ्या उत्साहाने साजरी करतो
  • Sentences:
    • दिवाळी आपण आनंदाने साजरी करतो
    • दिवाळी हा दिव्यांचा सण आहे
• Monolingual - Hindi
  • Source Sentence: हम दीपावली उत्साह के साथ मनाते हैं
  • Sentences:
    • हम दीपावली खुशियों से मनाते हैं
    • दिवाली रोशनी का त्योहार है
• Monolingual - Gujarati
  • Source Sentence: અમે ઉત્સાહથી દિવાળી ઉજવીએ છીએ
  • Sentences:
    • દિવાળી આપણે ખુશીઓથી ઉજવીએ છીએ
    • દિવાળી એ રોશનીનો તહેવાર છે
• Cross - lingual 1
  • Source Sentence: आम्हाला भारतीय असल्याचा अभिमान आहे
  • Sentences:
    • हमें भारतीय होने पर गर्व है
    • భారతీయులమైనందుకు గర్విస్తున్నాం
    • અમને ભારતીય હોવાનો ગર્વ છે
• Cross - lingual 2
  • Source Sentence: ਬਾਰਿਸ਼ ਤੋਂ ਬਾਅਦ ਬਗੀਚਾ ਸੁੰਦਰ ਦਿਖਾਈ ਦਿੰਦਾ ਹੈ
  • Sentences:
    • മഴയ്ക്ക് ശേഷം പൂന്തോട്ടം മനോഹരമായി കാണപ്പെടുന്നു
    • ବର୍ଷା ପରେ ବଗିଚା ସୁନ୍ଦର ଦେଖାଯାଏ |
    • बारिश के बाद बगीचा सुंदर दिखता है

📄 Related Links

• This model is released as a part of project MahaNLP: [https://github.com/l3cube - pune/MarathiNLP](https://github.com/l3cube - pune/MarathiNLP)
• A better sentence similarity model (fine - tuned version of this model) is shared here: [https://huggingface.co/l3cube - pune/indic - sentence - similarity - sbert](https://huggingface.co/l3cube - pune/indic - sentence - similarity - sbert)
• More details on the dataset, models, and baseline results can be found in our paper

📚 Citations

@article{deode2023l3cube,
  title={L3Cube - IndicSBERT: A simple approach for learning cross - lingual sentence representations using multilingual BERT},
  author={Deode, Samruddhi and Gadre, Janhavi and Kajale, Aditi and Joshi, Ananya and Joshi, Raviraj},
  journal={arXiv preprint arXiv:2304.11434},
  year={2023}
}

🔗 Other Related Models

• Monolingual Indic sentence BERT models:
  • [Marathi SBERT](https://huggingface.co/l3cube - pune/marathi - sentence - bert - nli)
  • [Hindi SBERT](https://huggingface.co/l3cube - pune/hindi - sentence - bert - nli)
  • [Kannada SBERT](https://huggingface.co/l3cube - pune/kannada - sentence - bert - nli)
  • [Telugu SBERT](https://huggingface.co/l3cube - pune/telugu - sentence - bert - nli)
  • [Malayalam SBERT](https://huggingface.co/l3cube - pune/malayalam - sentence - bert - nli)
  • [Tamil SBERT](https://huggingface.co/l3cube - pune/tamil - sentence - bert - nli)
  • [Gujarati SBERT](https://huggingface.co/l3cube - pune/gujarati - sentence - bert - nli)
  • [Oriya SBERT](https://huggingface.co/l3cube - pune/odia - sentence - bert - nli)
  • [Bengali SBERT](https://huggingface.co/l3cube - pune/bengali - sentence - bert - nli)
  • [Punjabi SBERT](https://huggingface.co/l3cube - pune/punjabi - sentence - bert - nli)
  • [Indic SBERT (multilingual)](https://huggingface.co/l3cube - pune/indic - sentence - bert - nli)
• Monolingual similarity models:
  • [Marathi Similarity](https://huggingface.co/l3cube - pune/marathi - sentence - similarity - sbert)
  • [Hindi Similarity](https://huggingface.co/l3cube - pune/hindi - sentence - similarity - sbert)
  • [Kannada Similarity](https://huggingface.co/l3cube - pune/kannada - sentence - similarity - sbert)
  • [Telugu Similarity](https://huggingface.co/l3cube - pune/telugu - sentence - similarity - sbert)
  • [Malayalam Similarity](https://huggingface.co/l3cube - pune/malayalam - sentence - similarity - sbert)
  • [Tamil Similarity](https://huggingface.co/l3cube - pune/tamil - sentence - similarity - sbert)
  • [Gujarati Similarity](https://huggingface.co/l3cube - pune/gujarati - sentence - similarity - sbert)
  • [Oriya Similarity](https://huggingface.co/l3cube - pune/odia - sentence - similarity - sbert)
  • [Bengali Similarity](https://huggingface.co/l3cube - pune/bengali - sentence - similarity - sbert)
  • [Punjabi Similarity](https://huggingface.co/l3cube - pune/punjabi - sentence - similarity - sbert)
  • [Indic Similarity (multilingual)](https://huggingface.co/l3cube - pune/indic - sentence - similarity - sbert)

🚀 Quick Start

💻 Usage Examples

🔧 Using Sentence - Transformers

Using this model becomes easy when you have sentence - transformers installed.

pip install -U sentence - transformers

Then you can use the model like this:

from sentence_transformers import SentenceTransformer
sentences = ["This is an example sentence", "Each sentence is converted"]

model = SentenceTransformer('{MODEL_NAME}')
embeddings = model.encode(sentences)
print(embeddings)

🔧 Using HuggingFace Transformers

Without sentence - transformers, you can use the model like this: First, you pass your input through the transformer model, then you have to apply the right pooling - operation on - top of the contextualized word embeddings.

from transformers import AutoTokenizer, AutoModel
import torch


#Mean Pooling - Take attention mask into account for correct averaging
def mean_pooling(model_output, attention_mask):
    token_embeddings = model_output[0] #First element of model_output contains all token embeddings
    input_mask_expanded = attention_mask.unsqueeze(-1).expand(token_embeddings.size()).float()
    return torch.sum(token_embeddings * input_mask_expanded, 1) / torch.clamp(input_mask_expanded.sum(1), min=1e-9)


# Sentences we want sentence embeddings for
sentences = ['This is an example sentence', 'Each sentence is converted']

# Load model from HuggingFace Hub
tokenizer = AutoTokenizer.from_pretrained('{MODEL_NAME}')
model = AutoModel.from_pretrained('{MODEL_NAME}')

# Tokenize sentences
encoded_input = tokenizer(sentences, padding=True, truncation=True, return_tensors='pt')

# Compute token embeddings
with torch.no_grad():
    model_output = model(**encoded_input)

# Perform pooling. In this case, mean pooling.
sentence_embeddings = mean_pooling(model_output, encoded_input['attention_mask'])

print("Sentence embeddings:")
print(sentence_embeddings)