Spaces:
Runtime error
Runtime error
| """ | |
| Estella Explainer — HuggingFace Spaces / Lightning AI Demo | |
| A math reading coach for grades 3–8, based on Usable Math (usablemath.org) | |
| Developed by Sai Gattupalli, PhD | |
| College of Education, University of Massachusetts Amherst | |
| Principal Scientist, Society and AI Research Group | |
| Model: sgattup/EstellaExplainerLLM (Gemma 4 E4B fine-tuned) | |
| """ | |
| import gradio as gr | |
| import torch | |
| from transformers import AutoModelForCausalLM, AutoTokenizer | |
| MODEL_ID = "sgattup/EstellaExplainerLLM" | |
| SYSTEM_PROMPT = """You are Estella Explainer, a math reading and language coach for young learners in grades 3 through 8. Your motto is: "My job is to explain math questions clearly so you know what you are supposed to do to solve the problem." | |
| Your rules: | |
| - Give only ONE hint per problem. | |
| - Do NOT solve the problem or show the answer. | |
| - Do NOT show calculation steps or compute anything. | |
| - Help the student understand what the problem is ASKING. | |
| - Use very simple words and short sentences (Flesch Reading Ease 90-100). | |
| - Identify what is known and what is being asked. | |
| - Use phrases like "We know...", "We are looking for...", or "Think about...". | |
| - Be friendly and encouraging.""" | |
| EXAMPLES = [ | |
| "A rectangle has a length of 12 cm and a width of 5 cm. What is the area?", | |
| "Sam had 3/4 of a pizza. He ate 1/4. How much is left?", | |
| "A store sells apples for $0.75 each. Jake buys 6 apples. How much does he spend?", | |
| "Round 4,567 to the nearest hundred.", | |
| "There are 48 students put into equal groups of 6. How many groups are there?", | |
| "A bag has 5 red marbles and 3 blue marbles. What fraction of the marbles are red?", | |
| "A train leaves at 10:15 AM and arrives at 1:45 PM. How long is the trip?", | |
| "The perimeter of a square is 36 inches. What is the length of one side?", | |
| ] | |
| print(f"Loading {MODEL_ID}...") | |
| tokenizer = AutoTokenizer.from_pretrained(MODEL_ID) | |
| model = AutoModelForCausalLM.from_pretrained( | |
| MODEL_ID, | |
| torch_dtype=torch.float16, | |
| device_map="auto" | |
| ) | |
| model.eval() | |
| print("Estella is ready.") | |
| def ask_estella(message: str, history: list) -> str: | |
| """Chat function — history is kept for UI display only; Estella gives one hint per problem.""" | |
| if not message.strip(): | |
| return "Please type a math word problem and I'll help explain what it's asking!" | |
| messages = [ | |
| {"role": "user", "content": SYSTEM_PROMPT + f"\n\n### Math Problem:\n{message.strip()}\n\n### Estella's Hint:"} | |
| ] | |
| text = tokenizer.apply_chat_template( | |
| messages, | |
| add_generation_prompt=True, | |
| tokenize=False | |
| ) | |
| inputs = tokenizer(text=[text], return_tensors="pt") | |
| inputs = {k: v.to(model.device) for k, v in inputs.items()} | |
| input_len = inputs["input_ids"].shape[1] | |
| with torch.no_grad(): | |
| outputs = model.generate( | |
| **inputs, | |
| max_new_tokens=150, | |
| temperature=0.7, | |
| do_sample=True, | |
| repetition_penalty=1.1, | |
| ) | |
| hint = tokenizer.decode(outputs[0][input_len:], skip_special_tokens=True).strip() | |
| return hint | |
| # Gradio ChatInterface | |
| demo = gr.ChatInterface( | |
| fn=ask_estella, | |
| title="📚 Estella Explainer — Math Reading Coach", | |
| description="""**"My job is to explain math questions clearly so you know what you are supposed to do to solve the problem."** | |
| Type any math word problem (grades 3–8) and Estella will help you understand what it's asking — without solving it. | |
| *Developed by [Sai Gattupalli, PhD](https://huggingface.co/sgattup) · UMass Amherst College of Education · [Society and AI Research Group](https://societyandai.org) · Based on [Usable Math](https://usablemath.org)*""", | |
| examples=EXAMPLES, | |
| cache_examples=False, | |
| type="messages", | |
| ) | |
| demo.launch() | |