KronQ quantized models and companion output-curvature calibration caches.
-
donghyunli/Llama-2-13b-KronQ-W2A16
Text Generation • 4B • Updated -
donghyunli/Llama-2-13b-KronQ-W2A16-g128
Text Generation • 4B • Updated • 28 -
donghyunli/Llama-2-13b-KronQ-W3A16-fake
Text Generation • 13B • Updated • 18 -
donghyunli/Llama-2-13b-KronQ-W3A16-g128-fake
Text Generation • 13B • Updated • 19