Falcon-H1-Tiny-90M-Instruct-math-aggressive

🎯 MATH-optimized | 📦 Aggressive pruning | ⚡ 3% weights pruned

This model is a aggressively pruned version of tiiuae/Falcon-H1-Tiny-90M-Instruct.

Performance Comparison

Category	Original	Pruned	Change
Python	0.0%	0.0%	→
Html	50.0%	33.3%	↓ 16.7%
Trivia	33.3%	16.7%	↓ 16.7%
Math	50.0%	33.3% ⭐	↓ 16.7%
Reasoning	16.7%	0.0%	↓ 16.7%
Medical	0.0%	0.0%	→
Linux	16.7%	16.7%	→
Writing	16.7%	16.7%	→

Average: 22.9% → 14.6% (-8.3%)

Math Retention: 66.7%

Quick Start

from transformers import AutoModelForCausalLM, AutoTokenizer

model = AutoModelForCausalLM.from_pretrained("CompactAI/Falcon-H1-Tiny-90M-Instruct-math-aggressive")
tokenizer = AutoTokenizer.from_pretrained("CompactAI/Falcon-H1-Tiny-90M-Instruct-math-aggressive")

inputs = tokenizer("Your prompt here", return_tensors="pt")
outputs = model.generate(**inputs, max_new_tokens=100)
print(tokenizer.decode(outputs[0], skip_special_tokens=True))

Technical Details

Property	Value
Base Model	tiiuae/Falcon-H1-Tiny-90M-Instruct
Specialization	Math
Prune Mode	Aggressive
Weight Reduction	3% weights pruned

License

This model inherits the license from the base model.