1,432 3 months ago

3B model that shouldn't be this good - crushes benchmarks through deep chain-of-thought reasoning