Falcon-H1-Tiny-90M-Base-math-safe

🎯 MATH-optimized | 📦 Safe pruning | ⚡ 1% weights pruned

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

Performance Comparison

Category	Original	Pruned	Change
Python	0.0%	0.0%	→
Html	8.3%	0.0%	↓ 8.3%
Trivia	25.0%	0.0%	↓ 25.0%
Math	16.7%	0.0% ⭐	↓ 16.7%
Reasoning	8.3%	0.0%	↓ 8.3%
Medical	0.0%	0.0%	→
Linux	0.0%	0.0%	→
Writing	0.0%	0.0%	→

Average: 7.3% → 0.0% (-7.3%)

Quick Start

from transformers import AutoModelForCausalLM, AutoTokenizer

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

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-Base
Specialization	Math
Prune Mode	Safe
Weight Reduction	1% weights pruned

License

This model inherits the license from the base model.