| --- |
| language: |
| - en |
| license: mit |
| base_model: microsoft/Phi-3.5-mini-instruct |
| tags: |
| - lora |
| - adapter |
| - text-generation |
| - knowledge-graph |
| - information-extraction |
| - rdf |
| - fine-tuned |
| - peft |
| - trustworthy-ai |
| - hallucination-detection |
| datasets: |
| - BSVGK/Text_to_KG_Construction_Dataset |
| pipeline_tag: text-generation |
| --- |
| |
| # Phi-3.5 Mini Instruct — LoRA Adapter (Text-to-KG) |
|
|
| ## Model Summary |
|
|
| This is the **LoRA adapter** for the Phi-3.5 Mini Instruct model fine-tuned to extract structured **RDF knowledge graph triples** from UK government procurement contract text. |
|
|
| > For the full merged model ready for inference, use: |
| > 👉 [BSVGK/phi35-mini-lora-text2kg-merged](https://huggingface.co/BSVGK/phi35-mini-lora-text2kg-merged) |
|
|
| ## Key Results |
|
|
| | Metric | Score | |
| |--------|-------| |
| | F1 Score | **0.9954** | |
| | BERTScore F1 | **0.9997** | |
| | Hallucination Rate | **0.00% (Zero)** | |
| | Test Contracts | 1,387 unseen contracts | |
|
|
| ## Model Details |
|
|
| - **Base Model:** microsoft/Phi-3.5-mini-instruct |
| - **Adapter Type:** LoRA (Low-Rank Adaptation) |
| - **Task:** Text-to-KG — RDF triple extraction from contract text |
| - **Domain:** UK Government Procurement Contracts |
| - **Training Dataset:** 9,244 verified UK contracts |
| - **Hardware:** NVIDIA A100 |
| - **Framework:** PyTorch, Hugging Face PEFT, TRL, SFTTrainer |
|
|
| ## Usage |
|
|
| ```python |
| from transformers import AutoTokenizer, AutoModelForCausalLM |
| from peft import PeftModel |
| |
| # Load base model |
| base_model = AutoModelForCausalLM.from_pretrained( |
| "microsoft/Phi-3.5-mini-instruct" |
| ) |
| tokenizer = AutoTokenizer.from_pretrained( |
| "microsoft/Phi-3.5-mini-instruct" |
| ) |
| |
| # Load LoRA adapter |
| model = PeftModel.from_pretrained( |
| base_model, |
| "BSVGK/phi35-mini-lora-text2kg-adapter" |
| ) |
| |
| prompt = """Extract RDF triples from the following UK government contract: |
| |
| Contract: [paste your contract text here] |
| |
| RDF Triples:""" |
| |
| inputs = tokenizer(prompt, return_tensors="pt") |
| outputs = model.generate(**inputs, max_new_tokens=256) |
| print(tokenizer.decode(outputs[0], skip_special_tokens=True)) |
| |
