mx check: add a per-element bound like check_quant's
Nobody has claimed this yet.
Assessment
- Difficulty
- 3/5
- Estimated time
- 1-2 days
- Newbie friendliness
- 58/100
- Issue type
- Feature
- Clarity
- Mostly clear
- Activity status
- Active
- Tech stack
- python
- Domain
- machine-learning
Research direction
Find the implementation of check_mx and the existing check_quant per-element bound, including how viol_frac handles affine quantization. Start by tracing how check_mx obtains each group’s E8M0 scale and applies its current error-ratio and rank-1 checks. Add a count for elements beyond half a step, using one step at the clipped edge, and check the affected cases against the reported clean and edited matrices.
Written by the indexing model from the issue text.
Description
From the review of https://github.com/STRML/quantdiff/pull/1 (P2). check_mx judges an mxfp8/mxfp4 tensor on its error ratio and one-direction share only. Against a bf16 upstream, a sparse edit can stay under 1.5x the error norm, and its rank-1 share fades on large matrices: on a 64x256 matrix, 1 and 8 edited elements tripped only the rank-1 test (30% and 13% against 3% clean). The mx step is known from each group's E8M0 scale, so an element past half a step (one step at the clipped edge) can be counted the way `viol_frac" does for affine quants. FP8-to-mxfp8 and FP4-to-mxfp4 are unaffected: any difference already flags.
- Dominant language
- Python
- Stars
- 0
- Forks
- 1
- Avg merge
- 5m
- Merged PRs (30d)
- 3
Getting set up
This project ships no dev container, Dockerfile or contributing guide, so setting up is up to you: start from its README, and see our first-contribution guide for the general steps.
First steps
- Read the whole issue, then the project's contributing guide.
- Comment on the issue to say you are picking it up — it saves two people doing the same work.
- Fork the repository and make your change on a branch.
- Open a pull request that references the issue number.
More from STRML/quantdiff
-
Diffing an FP8 pack against its own FP8 upstream exits 2 with an opaque errorPossibly taken @Vamsi-2581 claimed this 17 days ago. Open
Difficulty 2/5 1-3 hours Newbie friendliness 76/100
-
Difficulty 2/5 1-3 hours Newbie friendliness 75/100
-
Difficulty 3/5 1-2 days Newbie friendliness 54/100
-
Difficulty 4/5 3-5 days Newbie friendliness 44/100
-
Stacked experts: fetch only the sampled expert slices, count decoded bytes in the prefetch budgetOpen
Difficulty 4/5 1-2 days Newbie friendliness 48/100
Similar issues
-
Difficulty 2/5 1-3 hours Newbie friendliness 75/100
alltheplaces/alltheplaces#20570 · 2 comments ·
Maintainers usually reply within 1 day
-
Difficulty 2/5 1-3 hours Newbie friendliness 74/100
Maintainers usually reply within 1 day
-
area:ci bug
Difficulty 2/5 1-3 hours Newbie friendliness 85/100
catincloud-labs/dagster-dex#206 · 1 comment ·
Maintainers usually reply within 1 day
-
number_to_words raises ValueError with threshold= for ordinals ("12th") and comma-grouped strings ("1,000")Possibly taken @hemtilva claimed this today. Open
Difficulty 2/5 1-3 hours Newbie friendliness 74/100
Maintainers usually reply within 1 day
-
providers security
Difficulty 2/5 1-3 hours Newbie friendliness 72/100
agentbreeder/agentbreeder#1031 ·
Maintainers usually reply within 1 day