DOWNLOAD.md

📥 Download RoBERTa Joint NER+RE Model

The large model files are hosted on HuggingFace Hub for better performance and reliability.

🤗 Direct Download from HuggingFace Hub

Option 1: Using Transformers Library (Recommended)

from transformers import AutoTokenizer, AutoModelForTokenClassification

# Automatically downloads and caches the model
tokenizer = AutoTokenizer.from_pretrained("lemkin-ai/roberta-joint-ner-re")
model = AutoModelForTokenClassification.from_pretrained("lemkin-ai/roberta-joint-ner-re")

Option 2: Manual Download via CLI

# Install HuggingFace Hub CLI
pip install huggingface_hub

# Download all model files
huggingface-cli download lemkin-ai/roberta-joint-ner-re --local-dir ./models/roberta-joint-ner-re/

Option 3: Git Clone from HuggingFace

# Clone the model repository
git clone https://huggingface.co/lemkin-ai/roberta-joint-ner-re

📊 Model Files Available on HuggingFace Hub

File	Size	Description
pytorch_model.bin	2.1GB	PyTorch model weights
config.json	2KB	Model configuration
tokenizer_config.json	1KB	Tokenizer configuration
vocab.json	779KB	Vocabulary file
merges.txt	446KB	BPE merges
tokenizer.json	2.0MB	Fast tokenizer
special_tokens_map.json	1KB	Special tokens mapping

🌐 Model Hub URL

https://huggingface.co/lemkin-ai/roberta-joint-ner-re

⚡ Quick Start

import torch
from transformers import AutoTokenizer, AutoModelForTokenClassification

# Load model
tokenizer = AutoTokenizer.from_pretrained("lemkin-ai/roberta-joint-ner-re")
model = AutoModelForTokenClassification.from_pretrained("lemkin-ai/roberta-joint-ner-re")

# Example usage
text = "The International Criminal Court issued a warrant for the general's arrest."
inputs = tokenizer(text, return_tensors="pt", padding=True, truncation=True)

with torch.no_grad():
    outputs = model(**inputs)
    predictions = torch.argmax(outputs.logits, dim=-1)

# Process predictions for NER + RE tasks
print("Entity predictions:", predictions)

---

For complete documentation, see the main repository README and model card on HuggingFace Hub.