Downloads · 30 days
0
phiwi/Llama-3.3-70B-Instruct-bnb-4bit_regulatome_ppi_lora
Llama-3.3-70B-Instruct-bnb-4bit_regulatome_ppi_lora is a text generation model from phiwi. Use it when you need the model to write or continue text. The card lists the license as mit.
Fine-tuned Llama 3.3 70B Instruct (4-bit QLoRA) for protein–protein interaction extraction from biomedical abstracts.
Downloads · 30 days
0
Access
Public
Updated Jul 17, 2026
Repo size
—
Likes
0
Public
Click a slice to open those files.
.md3.7 KB · 71%
From the Hugging Face model README
Fine-tuned Llama 3.3 70B Instruct (4-bit QLoRA) for protein–protein interaction extraction from biomedical abstracts.
| Property | Value |
|---|---|
| Base model | unsloth/Llama-3.3-70B-Instruct-bnb-4bit |
| Fine-tuning method | LoRA (r=16, alpha=16, dropout=0) |
| Training quantization | 4-bit (NF4) |
| Inference quantization | Q4_K_M (~40 GB VRAM) |
| Chat template | ChatML (`< |
| Task | PPI triple extraction (INTERACTS_WITH) |
| Parameter | Value |
|---|---|
| Epochs | 5 |
| Per-device batch size | 2 |
| Gradient accumulation | 4 steps |
| Effective batch size | 8 |
| Learning rate | 2e-4 (linear schedule) |
| Warmup steps | 5 |
| Optimizer | AdamW 8-bit |
| Max sequence length | 120,000 |
| Hardware | 1x NVIDIA H100 (80 GB), ~40 GB VRAM |
| Training time | ~2 hours |
The model was fine-tuned using the ChatML template. For inference, format your prompt as:
<|im_start|>system
You are an expert molecular biologist specializing in protein-protein interactions. Your TASK is to extract protein-protein interactions from scientific texts with high precision. You understand the difference between direct physical interactions, functional relationships, and regulatory effects. When extracting relationships, focus on evidence-based direct interactions rather than indirect associations.<|im_end|>
<|im_start|>system
TEXT: [your biomedical abstract here]<|im_end|>
<|im_start|>system
Use the following OUTPUT FORMAT:
{
triples: [
{
head: string,
relation: "INTERACTS_WITH",
tail: string,
}
],
}<|im_end|>
<|im_start|>user
TASK: Extract all the protein-protein interactions from the TEXT. Focus on direct physical interactions where proteins bind to each other, modify each other, or form complexes.<|im_end|>
<|im_start|>assistant
The model responds with a JSON object containing extracted triples.
ollama run llama3.3:70b-regulatome-ppi
from unsloth import FastLanguageModel
model, tokenizer = FastLanguageModel.from_pretrained(
model_name="phiwi/Llama-3.3-70B-Instruct-bnb-4bit_regulatome_ppi_lora",
max_seq_length=120000,
load_in_4bit=True,
)
python extract.py --model llama33regu --data regulatome --target ppi
Evaluated on the RegulaTome test set (312 abstracts, direct extraction). See LLM-PPI-Extraction for full results.
A smaller 8B variant is also available: phiwi/Meta-Llama-3.1-8B-Instruct-bnb-4bit_regulatome_ppi_lora
@software{LLM-PPI-Extraction,
author = {Philipp Wiesenbach},
title = {LLM-PPI-Extraction: LLM-based extraction of protein-protein interactions},
year = {2026},
url = {https://github.com/dieterich-lab/LLM-PPI-Extraction}
}