mlcommons / mlcommons/tiny

Measure performance and accuracy on customized dataset

Open
#109 0 comments 0 reactions 0 assignees View on GitHub

Nobody has claimed this yet.

Dominant language
C
Stars
479
Forks
116
PR merge metrics
No merged PRs in 30d

Description

Hi,
Is there any way to measure performance and accuracy on customized dataset (other than ad, vww, kws, ic) ?
I have tried setting EE_MODEL_VERSION to my own dataset and put the dataset under /ulp-mlperf/datasets/ directory, but I got the following error when running medium performance:

Need at least 5 inputs to run benchmark mode

Contributor guide

Open the contributing guide

First steps

  1. Read the whole issue, then the project's contributing guide.
  2. Comment on the issue to say you are picking it up — it saves two people doing the same work.
  3. Fork the repository and make your change on a branch.
  4. Open a pull request that references the issue number.

Research direction

Start by reproducing the medium-performance benchmark with EE_MODEL_VERSION set to the custom dataset and the dataset under /ulp-mlperf/datasets/. Trace why benchmark mode reports “Need at least 5 inputs” and determine the requirements for measuring performance and accuracy on datasets beyond ad, vww, kws, and ic. Done means custom-dataset evaluation works or its supported limitations are documented.

Written by the indexing model from the issue text.

Assessment

Tech stack
c
Domain
machine-learning, performance
Issue type
Feature
Difficulty
5/5
Estimated time
Over a week
Activity status
Stale
Clarity
Needs clarification
Newbie friendliness
25/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.