phi2-autism-support-1000

Autism Support AI - Phi-2 Fine-tuned

Model Description

This is a specialized AI model fine-tuned to provide evidence-based autism support and guidance for parents and caregivers of autistic children.

Base Model: microsoft/phi-2 (2.7B parameters)
Fine-tuning: LoRA (Low-Rank Adaptation)
Training Data: 1000 autism-specific examples (500 parent conversations + 500 child conversations)
Training Loss: 0.789 (professional grade)

Intended Use

Primary Use Cases:

• Providing autism-specific parenting advice
• Answering questions about autism therapy and support
• Offering evidence-based recommendations for common autism challenges
• Supporting parents through their autism journey

Example Queries:

• "How can I help my child with meltdowns?"
• "What should I know about sensory processing issues?"
• "How do I prepare for an IEP meeting?"
• "My child refuses to eat anything but chicken nuggets"

Training Details

• Training Duration: 1 hour
• Dataset Size: 1000 examples
• Epochs: 10
• Final Loss: 0.789
• Hardware: Google Colab T4 GPU

Usage

from transformers import AutoModelForCausalLM, AutoTokenizer
from peft import PeftModel

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

# Load LoRA adapter
model = PeftModel.from_pretrained(base_model, "Dhibe/phi12-autism-support-1000")
tokenizer = AutoTokenizer.from_pretrained("Dhibe/phi12-autism-support-1000")

# Generate response
prompt = "How can I help my autistic child with transitions?"
inputs = tokenizer(prompt, return_tensors="pt")
outputs = model.generate(**inputs, max_new_tokens=300)
response = tokenizer.decode(outputs[0], skip_special_tokens=True)
print(response)

API Usage

import requests

API_URL = "https://router.huggingface.co/models/Dhibe/phi12-autism-support-1000"
headers = {"Authorization": "Bearer YOUR_HF_TOKEN"}

def query(text):
    response = requests.post(API_URL, headers=headers, json={"inputs": text})
    return response.json()

result = query("How do I handle school refusal in autistic children?")
print(result)

Limitations

• This model provides general autism support information and should not replace professional medical advice
• Responses are based on training data and may not cover all autism-related scenarios
• Always consult with healthcare professionals for personalized medical guidance

Model Card Authors

Shakti Dheerays S (Dhibe)

Citation

@misc{dhibe2026autism,
  author = {Dheerays, Shakti},
  title = {Autism Support AI - Phi-2 Fine-tuned},
  year = {2026},
  publisher = {HuggingFace},
  url = {https://huggingface.co/Dhibe/phi12-autism-support-1000}
}