Lint report · stored run
OCGQuant: Outlier-Companion Grouping for NVFP4 Quantization
0errors
0warnings
0info
17claims read
613numbers extracted
32references
How well we read this paper
4 observations
These are notes about our reading of the paper, not about the paper. They affect how much weight the findings below can carry.
Extraction warning 1
- ablation-labels-match-nothing None of the experiment ablation labels match any method component. Labels found: 'OCG w/o AP'. Either the components were extracted with different wording, or these labels are baselines rather than ablations — in which case RL004 has no evidence that anything was ablated.
Note 3
- numbers-are-identifier-fragments 14 extracted numbers are digits inside identifiers. e.g. '1' in 'Given a high-precision tensor X and its ', '2' in 'Given a high-precision tensor X and its ', '3' in 'Given a high-precision tensor X and its '. Rules that compare numbers across sections can report a contradiction between an identifier and a real value.
- compile-warning references: the reference list carries no entry numbers; citation keys are enumeration order [1]..[N], not labels printed in the paper, so a key may not match the [N] used in the body
- compile-warning semantic extraction: dropped 3 items whose quote was not found in the paper
Some checks could not run
- RL002 reported nothing, and it could not reach OpenAlex — so its silence here is not evidence that this paper is clean.
- RL003 reported nothing, and it could not reach OpenAlex — so its silence here is not evidence that this paper is clean.
Everything else on this page was computed from the paper itself and is unaffected. This is a limit of this run, not a finding about the paper.
Some checks did not apply to this paper
- RL005 did not apply to this paper — it links no repository we could read, so there was no released configuration to compare against. Its absence here says nothing about whether the paper and its code agree.
Compiler extraction log (2)
references: the reference list carries no entry numbers; citation keys are enumeration order [1]..[N], not labels printed in the paper, so a key may not match the [N] used in the body semantic extraction: dropped 3 items whose quote was not found in the paper
No findings on this paper.
That is a result, not a failure. The compiler read the paper without errors, and no rule found something it could prove with evidence — this product would rather say nothing than say something vague.
Read the extraction notes above before taking this as a clean bill of health: they say which checks were working with incomplete information.
Rule notes (28)
RL001 funnel arithmetic: sentences_with_from=2 endpoint_pairs=0 claims_compared=0 emitted=0 [dropped: not-a-measurement=0]
RL001 funnel delta-column: structured_tables=8 with_exactly_one_delta_column=0 rows_checked=0 emitted=0
RL001 funnel cross-location: quant=613 prose=102 prose_with_metric=8 prose_with_dataset=59 prose_with_metric_and_dataset=1 distinct_buckets=1 emitted=0 [dropped: not-a-measurement=0 not-in-quote=0]
RL001 funnel identifier-drift: settings_checked=2 sections_stating_one_value=1 emitted=0
RL002 citation entailment: attempted 18 of 23 ranked cited claims (45 in-text markers from 32 bibliography entries); 6 cited works could not be identified, 6 had no retrievable text, 0 model quotes failed verbatim verification, 0 were topically unrelated to the retrieved text.
RL002 candidate ranking (why the top 18 were chosen): [13] [introduction, names an artefact (GPTQ), title/sentence overlap near zero, no DOI/arXiv id]; [26] [introduction, names an artefact (SmoothQuant), title/sentence overlap near zero, no DOI/arXiv id]; [12] [introduction, names an artefact (MR-GPTQ), title/sentence overlap near zero, no DOI/arXiv id]; [30] [introduction, names an artefact (RPTQ), title/sentence overlap near zero]; [10] [introduction, names an artefact (FourOverSix), title/sentence overlap near zero]; [2] [introduction, names an artefact (QuaRot), no DOI/arXiv id]; [32] [introduction, title/sentence overlap near zero, no DOI/arXiv id]; [5] [introduction, names an artefact (RaZeR)]
RL002 label distribution over 6 classified citations: SUPPORTED=6
RL002 base-rate self-check: not tripped (0/6 = 0% NOT_SUPPORTED, threshold 40%).
RL002 attribution check (case B): 22 candidates name an artefact, 13 of them made the top-N and were checked against their cited source
RL002: 5 citations skipped because the sentence reports this paper's own results (the citation there is a comparison pointer, not evidence for the sentence)
RL002: 4 markers skipped because the claim is jointly attributed to several works
RL002: 1 enumerated-citation sentences skipped (they cite >=3 works, so the claim is about the set and cannot be refuted by any single cited work)
RL002: 6 duplicate candidates dropped (same reference and claim, or the same artefact attributed twice)
RL002: 6 cited works unresolvable (first 3: [{"key": "[13]", "title": "Gptq: Accurate post-training quantization for generative pre-trained transformers", "why": "no work record matched the bibliography title"}, {"key": "[26]", "title": "Smoothquant: Accurate and efficient post-training quantization for large language models", "why": "no work record matched the bibliography title"}, {"key": "[12]", "title": "Bridging the gap between promise)
RL003: 3 queries -> 151 unique works (arXiv search 76, category listings 0, scholarly graph 75 [crossref 75], 0 graph works matched to an arXiv id); source errors during retrieval: arXiv 0, graph 0
RL003: parsed the reference list of 3/12 comparable papers (9 unreadable) — that is the PeerUsage denominator
RL003: could read the reference list of only 3 comparable papers (need 4) — peer usage unavailable, no findings
RL004 funnel: components=9 contribution_backed=2 candidates=1 ablations_found=3 matched_by_ablation=1 unmatched=0 experiments=10
RL004 component M1 'Collateral Quantization Error': excluded:defined-quantity-not-a-mechanism
RL004 component M2 'Outlier-Companion Grouping (OCG)': covered (candidate)
RL004 component M3 'Companion Selection': excluded:not-a-claimed-contribution
RL004 component M4 'Adaptive Outlier-Companion Packing': covered (filtered earlier)
RL004 component M5 'channel permutation': excluded:not-a-claimed-contribution
RL004 component M6 'RMS-based channel calibration': covered (filtered earlier)
RL004 component M7 'GPTQ reconstruction': covered (filtered earlier)
RL004 component M8 'fused CUDA kernel': covered (filtered earlier)
RL004 component M9 'contiguous reorder-segment fast path': excluded:not-a-claimed-contribution
RL004 funnel: ranked=0 emitted=0 suppressed_by_cap=0 dropped_unverified_quote=0
Reading the report as data
This page is a rendering of the report the pipeline wrote. The JSON below
is the stored artefact, unedited — the same document
research-lint --format json produces.
loading…