tiiuae

/

falcon-7b-instruct

like 924

Follow

Technology Innovation Institute 954

Hi,

I would like to draw everyone's attention to AWQ quantized version of falcon-7b-instruct model at https://huggingface.co/abhinavkulkarni/falcon-7b-instruct-w4-g64-awq.

For more on AWQ, click here.

The quantized model size on disk is 4.76GB vs 14.43GB for the original model. Similar gains could be observed for VRAM usage. The perplexity is only worse by 5%.

Please take a look and give it a try.

Thanks!