Papers › How do Decisions Emerge across Layers in Neural Models? Interpretation with...

How do Decisions Emerge across Layers in Neural Models? Interpretation with Differentiable Masking

30 Apr 2020EMNLP 2020 11arXiv:2004.14992archive 2025-07-28

Nicola De Cao, Michael Schlichtkrull, Wilker Aziz, Ivan Titov

Attribution methods assess the contribution of inputs to the model prediction. One way to do so is erasure: a subset of inputs is considered irrelevant if it can be removed without affecting the prediction. Though conceptually simple, erasure's objective is intractable and approximate search remains expensive with modern deep NLP models. Erasure is also susceptible to the hindsight bias: the fact that an input can be dropped does not mean that the model `knows' it can be dropped. The resulting pruning is over-aggressive and does not reflect how the model arrives at the prediction. To deal with these challenges, we introduce Differentiable Masking. DiffMask learns to mask-out subsets of the input while maintaining differentiability. The decision to include or disregard an input token is made with a simple model based on intermediate hidden layers of the analyzed model. First, this makes the approach efficient because we predict rather than search. Second, as with probing classifiers, this reveals what the network `knows' at the corresponding layers. This lets us not only plot attribution heatmaps but also analyze how decisions are formed across network layers. We use DiffMask to study BERT models on sentiment classification and question answering.

PaperPDFConference PDFCodeCode Syntology ranMethod as code

In Syntology View this paper on Syntology, its page in Syntology's graph. That page adds which of its functions are also present in other papers' code Syntology holds, with dates, and its dated run record as JSON and a README badge. It also lists the repositories linked to the paper, the abstract and the calls for agents.

Open this paper in Syntology's Atlas, the map of the papers in Syntology's graph and their citations.

For agents, on Syntology's MCP service (how to connect):

Code

Syntology 10 code samples harvested from 2 repositories linked to this paper: 6 ran (of which 4 constructed an object rather than computing a result; 6 with no instrument failure: 0 honoured, 0 violated, 6 with no contract checked; 0 where Syntology's instrument failed) · 4 unverified. “Where Syntology's instrument failed” counts failures of Syntology's instrument, not of the code; here there are none.

By repository: official repository: 5 samples from 1 repository, 4 ran (4 with no instrument failure; 0 where Syntology's instrument failed); community (archive-listed): 5 samples from 1 repository, 2 ran (2 with no instrument failure; 0 where Syntology's instrument failed). The run record, sample by sample. “Ran” means executed on a synthesized input, not that the code is correct or reproduces the paper.

Levels written on 2026-09-22 (UTC): the dates Syntology's graph wrote each sample's verification level, not when the code executed.

nicola-decao/diffmask officialmentioned in papermentioned on GitHubpytorchSyntology: 5 samples, 4 ran (0 instrument-failed) · 1 unverified report
angelosnal/vision-diffmask mentioned on GitHubpytorchSyntology: 5 samples, 2 ran (0 instrument-failed) · 3 unverified report

Repository list and official/mentioned flags are the archive's, frozen 2025-07-28. Reachability, where shown, is from one Syntology probe window (2026-09-16 to 2026-09-18); repositories not probed show nothing. GitHub stars are not tracked.

Code Syntology ran Syntology

10 samples harvested: 6 ran (of which 4 constructed an object rather than computing a result; 6 with no instrument failure: 0 honoured, 0 violated, 6 with no contract checked; 0 where Syntology's instrument failed) · 4 unverified. “Where Syntology's instrument failed” counts failures of Syntology's instrument, not of the code. Syntology's record for this page has not changed since , the first build that kept a record date for it; when this build read Syntology's graph is in the build record. That date is not when the samples ran.

6ran
4unverified

Licence: 0 of the 10 samples are pointer only, meaning Syntology does not serve that copy's text. This page shows no code text for any sample; each one links to its file in the repository.

Harvested from 2 repositories linked to this paper, official or community; the list below is grouped by repository, each group headed by its name and what it is to the paper. “Ran” means the sample executed on a synthesized input. It does not mean the output is correct, and nothing here reproduces the paper's results. A line beginning “fixture:” under a sample that ran gives the arguments Syntology's fixture called the sample with, each as name:type/rank/element kind (? where none was recorded), and the type of what the fixture's run returned, exactly as the harness recorded them; the arguments and the returned type were recorded by two separate runs of the same fixture. No output value is shown, and a returned type says nothing about whether the output was right. “Level recorded” gives the UTC date when Syntology's graph wrote the sample's verification level, not when the code executed; where the level was written before Syntology's graph kept that date, the row says “level recorded, date not recorded”. “Honoured” and “violated” refer to a contract Syntology drafted from the code itself. “Our draft was wrong” and “the contract check could not run” are failures of Syntology's instrument, not of the code: the sample ran on Syntology's synthesized input, and then the drafted contract did not describe it, or the check itself raised an error (RAISES), lacked a dependency (DEP_MISSING) or ran out of time (TIMEOUT); hover the label for which. A metamorphic tier, where shown, is evidence of the output's shape, not that the sample implements the method.

Each sample ends with its code_sha256. A sample's code_sha256 is Syntology's identity for that exact code. It is printed in full (16 hexadecimal characters for every sample in this build), and get_code(code_sha256="…") takes it exactly as printed. An agent calls it on Syntology's MCP service (how to connect); click an identity to copy that call.

Repository labels, per sample. official repository: The archive marks this repository official for the paper. named in the paper: The archive records that the paper mentions this repository; it is not marked official. community (archive-listed): In the archive's code links for this paper, not marked official and not recorded as mentioned in the paper. found in paper text by Syntology: Syntology found this repository in the paper's own text; whether it is the authors' implementation is not asserted. community: Not in the archive's code links for this paper; a community repository Syntology harvested. Samples are grouped by repository: a repository marked official first, then by how directly the repository is tied to the paper, then in the archive's order; never by how many ran. Licence labels name the repository's licence as recorded at harvest. “Pointer only” means Syntology does not serve that copy's text, for one of six reasons: no licence file was found; the licence was not identified; the licence is recorded as permissive but that copy's record is not marked cleared; the licence is outside the permissive list Syntology serves text under (MIT, Apache-2.0, BSD and similar); this paper's copy was not located, so its licence is not recorded; or this paper's copy is cleared, but the stored copy of the same code that Syntology serves is not. Some licences outside that list permit redistribution, such as WTFPL, and GPL-3.0 under its conditions; they are simply not on the list. Hover a licence label for the reason. File links open the file on GitHub at the default branch, which may have changed since the harvest.

nicola-decao/diffmask official repository 5 samples, 4 ran (0 instrument-failed) · 1 unverified
BinaryConcrete diffmask/models/gates.py ran · metamorphic tier: deterministic MIT (permissive) · 9d22095a17af9672 · report fixture: returned BinaryConcrete · level recorded 2026-09-22
DiffMaskGateInput diffmask/models/gates.py ran · metamorphic tier: deterministic MIT (permissive) · 42f456f21c831412 · report fixture: returned DiffMaskGateInput · level recorded 2026-09-22
MLPMaxGate diffmask/models/gates.py ran · metamorphic tier: deterministic MIT (permissive) · 6c897bca72237121 · report fixture: returned MLPMaxGate · level recorded 2026-09-22
Streched diffmask/models/gates.py ran · metamorphic tier: deterministic MIT (permissive) · 121b859bf1d3c812 · report fixture: returned Streched · level recorded 2026-09-22
RectifiedStreched diffmask/models/gates.py unverified MIT (permissive) · f8f0147cc5dfb672 · report
angelosnal/vision-diffmask community (archive-listed) 5 samples, 2 ran (0 instrument-failed) · 3 unverified
BinaryConcrete code/models/gates.py ran · metamorphic tier: invariant fingerprinted MIT (permissive) · 6a5a019aa6eb2889 · report fixture: signature value:Tensor/2/float · returned Tensor · level recorded 2026-09-22
RectifiedStreched code/models/gates.py ran · metamorphic tier: well formed MIT (permissive) · 1812ba7431095707 · report fixture: signature sample_shape:Size/1/int · returned Tensor · level recorded 2026-09-22
DiffMaskGateInput code/models/gates.py unverified MIT (permissive) · ff21479d516498d0 · report
MLPMaxGate code/models/gates.py unverified MIT (permissive) · 746bf702c11c94f1 · report
Streched code/models/gates.py unverified MIT (permissive) · 71fdc59594de30ac · report

Method as code Syntology

What Syntology's graph records as this paper's claimed method, lifted into code: each assembly names its bricks (functions and classes), each with its own role and its own verification level. An assembly is not a check that the code implements the method correctly or reproduces the paper's results. Levels are each brick's own verification level in Syntology's graph (level 0: no recorded run), listed brick by brick and never combined into one level; “ran” means the brick executed on a synthesized input, not that its output is correct, and the instrument figure counts failures of Syntology's instrument, not of the code. Where a model read the code against the paper's claim, each assembly's label verdict is one model's judgement, kept in this page's JSON twin; it is not calibrated, so no card prints it. Siblings are assemblies lifted from this paper in the same batch for the same method: alternative readings kept side by side; none replaces or supersedes another.

2 assemblies on this paper: 2 harvested, 0 generated, listed in 1 groups (siblings lifted in one batch form one group).

2 sibling assemblies for Differentiable Masking (harvested), lifted in the same batch on 2026-09-09: alternative readings kept side by side; none replaces or supersedes another.

harvested Assembly for Differentiable Masking (graph method: Differentiable Masking). 5 bricks; individually on a fixture: 4 ran (of which 4 constructed an object rather than computing a result; 4 with no instrument failure: 0 honoured, 0 violated, 4 with no contract checked; 0 where Syntology's instrument failed) · 1 unverified; every one of the 4 samples that ran constructed an object rather than computing a result. The assembly has not been run end-to-end.

one of 2 siblings

  1. DiffMaskGateInput entry · level 1 · ran · 42f456f21c831412
  2. BinaryConcrete sampler · level 1 · ran · 9d22095a17af9672
  3. MLPMaxGate helper · level 1 · ran · 6c897bca72237121
  4. RectifiedStreched helper · level 0 · unverified · f8f0147cc5dfb672
  5. Streched helper · level 1 · ran · 121b859bf1d3c812

harvested Assembly for Differentiable Masking (graph method: Differentiable Masking). 5 bricks; individually on a fixture: 2 ran (of which 0 constructed an object rather than computing a result; 2 with no instrument failure: 0 honoured, 0 violated, 2 with no contract checked; 0 where Syntology's instrument failed) · 3 unverified. The assembly has not been run end-to-end.

one of 2 siblings

  1. DiffMaskGateInput entry · level 0 · unverified · ff21479d516498d0
  2. BinaryConcrete helper · level 1 · ran · 6a5a019aa6eb2889
  3. MLPMaxGate helper · level 0 · unverified · 746bf702c11c94f1
  4. RectifiedStreched helper · level 1 · ran · 1812ba7431095707
  5. Streched helper · level 0 · unverified · 71fdc59594de30ac

Each brick ends with its code_sha256, Syntology's identity for that exact code; click one to copy the get_code call that returns it. For agents, on Syntology's MCP service: assembly(arxiv_id="2004.14992") returns these assemblies with each brick's role and wiring (how to connect).

Tasks

PredictionQuestion AnsweringSentiment AnalysisSentiment Classification

Results from the paper archive 2025-07-28

No leaderboard rows for this paper in the archive.

Methods

AdamAttentionAttention DropoutBERTDense ConnectionsDropoutLayer NormalizationLinear LayerLinear Warmup With Linear DecayMulti-Head AttentionPruningResidual ConnectionSoftmaxWeight DecayWordPiece

Report a problem or propose a change · a person checks every report against the paper or source before anything changes; decisions are listed on /corrections