DepNeCTI-LSTM: Dependency-based Nested Compound Type Identification for Sanskrit

This repository contains the DepNeCTI-LSTM model checkpoint and configuration files trained for nested compound type identification in Sanskrit using a dependency-based LSTM encoder.

---

Model Details

• Model type: LSTM-based dependency parser
• Task: Nested Compound Type Identification (NeCTI) in Sanskrit
• Architecture: LSTM encoder with word, POS, and character embeddings
• Trained on: DepNeCTI dataset (from Sandhan et al., 2023)
• Pretrained embeddings used: FastText embeddings (cc.NeCTIS.300.txt)
• Framework: PyTorch 1.13.0
• CUDA version: 11.7

---

Files

• domain_san.pt — Pretrained model state for the Domain-SAN model.

• domain_san.arg.json — JSON file containing model hyperparameters and configuration settings.

• README.md — Instructions for setup, usage, and reproduction of results.

• requirements.txt — List of Python dependencies required to run the model.

• LICENSE — Apache License 2.0 — grants broad usage rights with conditions for attribution and inclusion of the license when redistributing.

---

Usage

This model and arguments(json format) were obtained after running the training script. To reproduce the model in accordance to your needs refer to the original paper

Paper

@misc{sandhan2023depnecti,
      title={DepNeCTI: Dependency-based Nested Compound Type Identification for Sanskrit}, 
      author={Jivnesh Sandhan and Yaswanth Narsupalli and Sreevatsa Muppirala and Sriram Krishnan and Pavankumar Satuluri and Amba Kulkarni and Pawan Goyal},
      year={2023},
      eprint={2310.09501},
      archivePrefix={arXiv},
      primaryClass={cs.CL}
}

Original paper DepNeCTI: Dependency-based Nested Compound Type Identification for Sanskrit

Github Repository of DepNeCTI