Modelpublic

alaamostafa/Microsoft-Phi-2

sourceHugging Facemitupdated 2y agoView on Hugging Face

0likes7downloads

Model Card

Neuroscience Fine-tuned Phi-2 Model

Model Description

This is a fine-tuned version of Microsoft's Phi-2 model, adapted specifically for neuroscience domain content.

Training Procedure

—Base Model: Microsoft Phi-2 (2.7B parameters)
—Training Type: LoRA fine-tuning
—Training Dataset: BrainGPT/trainvalidsplitpmcneuroscience2002-2022filtered_subset
—Training Duration: 3+ epochs
—Parameter-Efficient Fine-Tuning: Used LoRA with r=16, alpha=32

Usage

python

from transformers import AutoModelForCausalLM, AutoTokenizer
from peft import PeftModel

# Load base model
base_model = AutoModelForCausalLM.from_pretrained("microsoft/phi-2", torch_dtype="auto")

# Load adapter
model = PeftModel.from_pretrained(base_model, "alaamostafa/Microsoft-Phi-2")

# Load tokenizer
tokenizer = AutoTokenizer.from_pretrained("microsoft/phi-2")

# Generate text
input_text = "Recent advances in neuroscience suggest that"
inputs = tokenizer(input_text, return_tensors="pt")
outputs = model.generate(**inputs, max_length=100)
print(tokenizer.decode(outputs[0], skip_special_tokens=True))