AI Paper Atlas

Research map / Architectures and efficiency / Quantization

UDC: Unified DNAS for Compressible TinyML Models for Neural Processing Units

Igor Fedorov, Ramon Matas, Hokchhay Tann, Chuteng Zhou, Matthew Mattina, Paul Whatmough

NeurIPS 2022 · Architectures and efficiency › Quantization · approach: Post-training weight quantization · 0 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
Increase Gumbel-softmax temperature tau from 0.66 to 10 (ablation)Vanilla DNAS · MBNetV2 search spaceL_E^z-hat: 0.04 -> 0.53Hurt
Add projection onto exploration set (xi=0.5) to DNAS (ablation)Vanilla DNAS · MBNetV2, tau=0.66L_E^z-hat: 0.04 -> 0.33Hurt
Add rejection sampling (theta=0.99) on top of projection (ablation)UDC · MBNetV2, tau=0.66L_E^z-hat: 0.33 -> 0.18Improved
Add rejection sampling (theta=0.5) on top of projection (ablation)UDC · MBNetV2, tau=0.66L_E^z-hat: 0.33 -> 0.27Improved
Use rejection sampling at high temperature tau=10 (theta=0.99) (ablation)UDC · MBNetV2, tau=10L_E^z-hat: 0.61 -> 0.59No clear effect
Use 32 MC samples instead of 1 with fewer steps (ablation)UDC · DNAS search speedspeed-up: 5.3x speed-upImproved
Disable UDC components (partial UDC variant) (ablation)UDC · CIFAR100 0.5MBaccuracy (%): 79.24 -> 76.48Hurt
Disable a UDC component requiring lambda tuning (partial UDC) (ablation)UDC · CIFAR100 0.5MBGPUD: 1.17 -> 6.21Hurt
Use vanilla DNAS instead of UDC (ablation)Vanilla DNAS · CIFAR100/ImageNetHW constraint satisfactionHurt
Replace uniform quantization with proposed number format that quantizes beyond pruning boundary (ablation)UDC · ImageNetweight norm growth: 4.6x -> 2.2xImproved
Use proposed number format instead of uniform quantization (ablation)UDC · ImageNetaccuracyImproved
Deploy with b*=8 uniform quantization of trained modelUDC · ImageNetaccuracyNo clear effect

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy