RAMP: Robust Adaptive Mixed-Precision Quantization for Edge CPU Vision Models

A self-hosted assistant is too expensive because its model uses most of the GPU memory.

What you would change

Try the paper's supported lower-precision weights on a copy of your serving setup.

Try the paper's supported lower-precision weights on a copy of your serving setup.

Compare against

The same model at its original precision, on the same GPU.

1. Locate the paper's implementation and check its license, model support, data needs, and hardware requirements. Stop here if you cannot obtain the required artifacts.
2. Save a small set of representative product tasks and expected outcomes. Keep evaluation tasks out of any training or tuning, and choose quality and cost limits before running.
3. Run the baseline and the proposed change on identical inputs. Keep other settings fixed and repeat paired runs to expose variation.
4. Save per-task outputs, failures, timings, and all usage costs. Review regressions before expanding the test.

What to measure

Peak memory, task pass rate, latency, and total serving cost at the same context and load.

When to keep it

Keep the change only if it meets your preselected quality and latency limits and improves the metric you care about after setup and operating costs. Otherwise retain the baseline. This is an exploration plan, not a validated recommendation.

recommendation

test

research area

quantization

technical change

JSD-based layer sensitivity with K-Means mixed-precision INT8 quantization

published

2026-09-23

The method is cheap to run without a GPU and directly addresses a real production failure mode of standard sensitivity metrics, but its 1.81x claim is specific to ARM64 edge CPU vision inference and must be validated on the reader's own models and hardware before adoption.

1. claim 01Paperraft inference
   Read the brief interpretation

   Replaces gradient-based and weight-statistics layer sensitivity metrics (and fixed-threshold policy selection) for mixed-precision INT8 quantization of vision models on edge CPUs, using Jensen-Shannon Divergence scoring plus K-Means clustering to choose per-layer precision. Costs are modest: a calibration pass with no GPU or gradient computation, plus integration effort to apply per-layer policies, but the benefit materializes only on integer-accelerated CPU deployments, not GPU or API-hosted inference. Failures after adoption include workload-specific accuracy degradation on architectures not covered by the four studied networks, and counterproductive latency regressions if insensitive-but-slow-to-quantize layers are excluded, since that fragments the compute graph and disables operator fusion.

   This is a Paperraft inference. It is not pinned to a verified PDF passage.

   Inspect the source
2. claim 02Paperraft inference

   Mean 1.81x speed-up over the full-precision model with near-lossless accuracy, validated on two ARM64 platforms; Jensen-Shannon Divergence had zero catastrophic failures across 8 model-hardware configurations versus 4 for gradient-based and 2 for weight-based metrics.

   This is a Paperraft inference. It is not pinned to a verified PDF passage.

   Inspect the source