Model Card for llm7-graph-270m-it-ft-20251009

This model, llm7-graph-270m-it-ft-20251009, is a specialized version of LLM that has been fine-tuned to extract and generate knowledge graphs from unstructured text. 🧠

Its primary purpose is to identify entities and the relationships between them in a given piece of text and present this information in a structured graph format. For instance, if you provide the input "The capital of France is Paris," the model will output a structured relationship like [{"from": "Paris", "relationship": "located_in", "to": "France"}].

The model was trained using Supervised Fine-Tuning (SFT) and is multilingual, supporting the following languages:

• English (en)
• German (de)
• French (fr)
• Spanish (es)
• Italian (it)
• Russian (ru)
• Chinese (zh)
• Romanian (ro)

Quick start

from transformers import pipeline

question = "The capital of France is Paris."

generator = pipeline("text-generation", model="EugeneEvstafev/llm7-graph-270m-it-ft-20251009", device="cpu")
output = generator([{"role": "user", "content": question}], max_new_tokens=128, return_full_text=False)[0]

print(output["generated_text"])  # [{"from": "Paris", "relationship": "located_in", "to": "France"}] 

Training procedure

This model was trained with SFT.

Framework versions

• TRL: 0.23.1
• Transformers: 4.56.2
• Pytorch: 2.8.0+cu126
• Datasets: 4.0.0
• Tokenizers: 0.22.1

Citations

Cite llm7-graph-270m-it-ft-20251009 as:

@misc{EugeneEvstafevLLM7,
    title        = {{llm7-graph-270m-it-ft-20251009}},
    author       = {Eugene Evstafev{\'e}dec},
    year         = 2025
}