File size: 4,689 Bytes
a090a8c 7067fdc 0c51a03 a090a8c | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 | ---
license: mit
language:
- hi
metrics:
- exact_match
base_model:
- ai4bharat/IndicBERTv2-MLM-only
pipeline_tag: token-classification
library_name: transformers
tags:
- word-grouping
- indic-nlp
- hindi
- token-classification
- local-word-groups
- bio-tagging
---
# WG-IndicBERT
A token classification model fine-tuned from [IndicBERT v2](https://huggingface.co/ai4bharat/IndicBERTv2-MLM-only) for **Indic Word Grouping** (Local Word Group identification). Developed as part of the BHASHA 2025 Shared Task 2: IndicWG.
- **Developed by:** Manav Dhamecha, Gaurav Damor, Sunil Choudhary, Pruthwik Mishra
- **License:** MIT
- **Base model:** [ai4bharat/IndicBERTv2-MLM-only](https://huggingface.co/ai4bharat/IndicBERTv2-MLM-only)
- **Paper:** [Team Horizon at BHASHA Task 2](https://aclanthology.org/2025.bhasha-1.18/)
- **Repository:** [manavdhamecha77/IndicWG2025](https://github.com/manavdhamecha77/IndicWG2025)
- **GitHub.io:** [Indic Word Grouping](https://manavdhamecha77.github.io/wg/)
---
## What it does
Given an input sentence in Hindi, the model identifies **Local Word Groups (LWGs)** — semantically cohesive sequences of words that convey a single complete meaning (e.g., noun compounds, postpositional groups, verb groups with auxiliaries, light verb constructions).
The task is modeled as **BIO token classification** with three labels: `B` (beginning of a group), `I` (inside a group), `O` (outside / delimiter). The output is reconstructed into grouped sentences using `__` as the group boundary separator.
**Exact Match Accuracy: 52.73% on the official test set**
---
## Quick Start
```python
from transformers import AutoTokenizer, AutoModelForTokenClassification
import torch
model_name = "manavdhamecha77/WG-IndicBERT"
tokenizer = AutoTokenizer.from_pretrained(model_name)
model = AutoModelForTokenClassification.from_pretrained(model_name)
device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
model.to(device)
sentence = "राम ने बाजार से सब्जियां खरीदीं।"
inputs = tokenizer(sentence, return_tensors="pt", truncation=True, max_length=128).to(device)
with torch.no_grad():
outputs = model(**inputs)
predictions = torch.argmax(outputs.logits, dim=-1)[0].tolist()
tokens = tokenizer.convert_ids_to_tokens(inputs["input_ids"][0])
# Label map: {0: "B", 1: "I", 2: "O"}
id2label = {0: "B", 1: "I", 2: "O"}
for token, pred in zip(tokens, predictions):
if token not in ["[CLS]", "[SEP]", "[PAD]"]:
print(f"{token:20s} {id2label[pred]}")
```
---
## Training Details
The model is fine-tuned using `AutoModelForTokenClassification` with a class-weighted cross-entropy loss to address the dominant `O`-label imbalance. Labels are aligned to subword tokens using the tokenizer's `word_ids()` helper; only the first subword of each word is labeled, with subsequent subwords set to `-100`.
| Parameter | Value |
|---|---|
| Optimizer | AdamW |
| Learning Rate | 3×10⁻⁵ |
| Batch Size | 8 (train/eval) |
| Epochs | 20 |
| Weight Decay | 0.01 |
| Label Map | B:0, I:1, O:2 |
| Hardware | H100 GPU (94GB) |
**Training data:** Official BHASHA/IndicWG Hindi dataset (550 train / 100 dev / 226 test sentences), supplemented with 5K augmented sentences from a rule-based LWG finder applied to IndicCorp.
---
## Evaluation
| Model | Dev EM (%) | Test EM (%) |
|---|---|---|
| MuRIL | 46.58 | 58.18 |
| XLM-Roberta | 39.00 | 53.36 |
| **IndicBERT v2 (this model)** | **35.40** | **52.73** |
Evaluation metric: **Exact Match Accuracy** — a prediction is correct only if the entire reconstructed grouped sentence matches the gold output exactly.
---
## Limitations
- Trained and evaluated on Hindi only; generalization to other Indic languages is not guaranteed.
- Performance degrades on longer sentences (>40 words).
- IndicBERT v2 was pretrained with MLM only, without task-specific fine-tuning on sequence labeling, which may explain its slightly lower performance compared to MuRIL on this task.
- Gold annotations contain inconsistencies in multiword expressions and light-verb constructions, which caps achievable exact-match accuracy.
---
## Citation
```bibtex
@inproceedings{dhamecha2025horizonwg,
title = {Team Horizon at {BHASHA} Task 2: Fine-tuning Multilingual Transformers for Indic Word Grouping},
author = {Dhamecha, Manav and Damor, Gaurav and Choudhary, Sunil and Mishra, Pruthwik},
booktitle = {Proceedings of the 1st Workshop on Benchmarks, Harmonization, Annotation, and Standardization for Human-Centric AI in Indian Languages (BHASHA 2025)},
year = {2025},
url = {https://aclanthology.org/2025.bhasha-1.18/}
}
```
|