---
tags: [esg, greenwashing, phobert, vietnamese, sustainability]
license: mit
language: vi
---

# ESG Greenwashing Detection Model

Multi-task PhoBERT model for Vietnamese ESG content analysis.

## Model Architecture

4-task learning:
1. **Greenwashing Classification** (Legitimate/Greenwashing/Uncertain)
2. **ESG Pillar Classification** (Environmental/Social/Governance/General)
3. **Content Quality Scoring** (0-100)
4. **ESG Score Prediction** (0-100)

## Training

- **Base Model:** vinai/phobert-base
- **Strategy:** stratified_group_kfold_2
- **Folds:** 1
- **Total Samples:** 21

## Performance

### Greenwashing Detection
- F1 Score: 0.118
- Precision: 0.111
- Recall: 0.125

### Pillar Classification
- Accuracy: 0.000
- F1 Macro: 0.000

### Quality Scoring
- MAE: 35.345
- R²: -19.549

### ESG Score Prediction
- MAE: 33.567
- R²: -4.038
- Correlation: 0.131

## Usage

```python
from transformers import AutoTokenizer, AutoModel
import torch

tokenizer = AutoTokenizer.from_pretrained("hiennthp/esg-bank-model-v4")
# Load model architecture then weights
# model = MultiTaskPhoBERT(config)
# model.load_state_dict(torch.load("best_model_fold0.pt"))
```

## Files

- `best_model_fold0.pt` - Fold 0 model weights
- `step5_metrics.json` - Detailed metrics with per-fold breakdown
- `tokenizer/` - PhoBERT tokenizer files

## Citation

```bibtex
@software{esg_greenwashing_model,
  author = {ESG Research Team},
  title = {Vietnamese ESG Greenwashing Detection Model},
  year = {2026},
  url = {https://huggingface.co/hiennthp/esg-bank-model-v4}
}
```