Dropbox: low-bit inference enables efficient AI model serving for Dash
Running large AI models in production for Dropbox Dash requires substantial and growing memory, compute, and energy, and efficiently serving them within practical hardware, cost, and latency constraints is a central engineering challenge.
Dropbox already employs a range of quantization strategies to optimize model deployment and fully utilize modern accelerators for Dash's AI features, though FP4 adoption and framework support remain incomplete across the industry.
Frequently asked questions
What did this team achieve with this AI workflow?
Dropbox already employs a range of quantization strategies to optimize model deployment and fully utilize modern accelerators for Dash's AI features, though FP4 adoption and framework support remain incomplete across…
What tools did this team use?
Dropbox Dash, AWQ, HQQ, Flash Attention 3, Sage Attention, Triton.
What results were reported?
Inference throughput from halved precision: roughly double throughput; Model serving speed and cost: faster and cheaper to run; energy efficiency with FP4 (Blackwell vs H100): significant energy savings (source-reported, not independently verified).
How is this workflow AI workflow structured?
Scale-driven efficiency demand → Quantization applied to weights and activations → Calibration and grouping → GPU hardware matrix core execution → Production AI feature serving in Dash.