@RicursiveAI's Ripple-65B-NVFP4 runs on ONE AMD Strix Halo desktop on your desk 🖥️
Fitting across 128 GB unified memory, with DFlash2 drafter this massive DENSE model decodes with pristine BFCL accuracy at 6.3 tok/s ⚙️
@AIatAMD's Strix Halos can run some serious models 🔥
Fitting across 128 GB unified memory, with DFlash2 drafter this massive DENSE model decodes with pristine BFCL accuracy at 6.3 tok/s ⚙️
@AIatAMD's Strix Halos can run some serious models 🔥