X · @teortaxesTex
· X / Twitter
> the improvement over DFlash: > +20% acceptance length > +14% throughput > avg 127 tok/s vs 111 for DFlash and 81 for EAGLE-3 > for single GPU infere…
> the improvement over DFlash:> +20% acceptance length> +14% throughput> avg 127 tok/s vs 111 for DFlash and 81 for EAGLE-3> for single GPU inference, DSpark's lightweight + early stopping approach is clearly superior…DeepSeek did just casually ship a SoTA spec decodingmohit: Deepseek's DSpark compared with DFlash and