Papers › When does compositional structure yield compositional generalization? A kernel theory

When does compositional structure yield compositional generalization? A kernel theory

26 May 2024arXiv:2405.16391archive 2025-07-28

Samuel Lippl, Kim Stachenfeld

Compositional generalization (the ability to respond correctly to novel combinations of familiar components) is thought to be a cornerstone of intelligent behavior. Compositionally structured (e.g. disentangled) representations support this ability; however, the conditions under which they are sufficient for the emergence of compositional generalization remain unclear. To address this gap, we present a theory of compositional generalization in kernel models with fixed, compositionally structured representations. This provides a tractable framework for characterizing the impact of training data statistics on generalization. We find that these models are limited to functions that assign values to each combination of components seen during training, and then sum up these values ("conjunction-wise additivity"). This imposes fundamental restrictions on the set of tasks compositionally structured kernel models can learn, in particular preventing them from transitively generalizing equivalence relations. Even for compositional tasks that they can learn in principle, we identify novel failure modes in compositional generalization (memorization leak and shortcut bias) that arise from biases in the training data. Finally, we empirically validate our theory, showing that it captures the behavior of deep neural networks (convolutional networks, residual networks, and Vision Transformers) trained on a set of compositional tasks with similarly structured data. Ultimately, this work examines how statistical structure in the training data can affect compositional generalization, with implications for how to identify and remedy failure modes in deep learning models.

PaperPDFCodeCode 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 12 code samples harvested from 1 repository linked to this paper: 8 ran (of which 0 constructed an object rather than computing a result; 5 with no instrument failure: 1 honoured, 3 violated, 1 with no contract checked; 3 where Syntology's instrument failed) · 4 unverified. Where Syntology's instrument failed: 3 where our contract draft was wrong; failures of Syntology's instrument, not of the code.

By repository: official repository: 12 samples from 1 repository, 8 ran (5 with no instrument failure; 3 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 between 2026-09-22 and 2026-09-24 (UTC): the dates Syntology's graph wrote each sample's verification level, not when the code executed.

sflippl/compositional-generalization officialmentioned in paperpytorchSyntology: 12 samples, 8 ran (3 instrument-failed) · 4 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

12 samples harvested: 8 ran (of which 0 constructed an object rather than computing a result; 5 with no instrument failure: 1 honoured, 3 violated, 1 with no contract checked; 3 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.

1ran · honoured contract
3ran · violated contract
3ran · our draft was wrong
1ran
4unverified

Licence: 12 of the 12 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 sflippl/compositional-generalization. “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.

A function or class name links to its page when the name occurs in the harvested code of at least 5 papers; the name is a string, not an identity. A census tag appears when Syntology's census ran implementations with that name on one shared input per bucket: it gives how many were compared, from how many papers, and how many distinct outputs they produced, and names this sample's bucket when this sample was one of them.

sflippl/compositional-generalization official repository 12 samples, 8 ran (3 instrument-failed) · 4 unverified
CSVDataFrame simple_compositional_tasks/models/kernel_machines.py ran no licence file found · pointer only · 8d0ab110639937d8 · report fixture: signature wd:str/0/str · returned NoneType · level recorded 2026-09-24
binary_accuracy in 13 papers simple_compositional_tasks/models/kernel_machines.py ran · honoured contract fingerprinted no licence file found · pointer only · a8f89e3acd77db3a · report fixture: signature preds:ndarray/1/float, targets:ndarray/1/float · returned float64 · level recorded 2026-09-22
complete_sims simple_compositional_tasks/models/kernel_machines.py ran · our draft was wrong no licence file found · pointer only · 9ae529d03358d2a1 · report fixture: signature sims:dict/0/?, comps:list/1/int · returned dict · level recorded 2026-09-24
is_pos_def in 9 papers simple_compositional_tasks/models/kernel_machines.py ran · violated contract no licence file found · pointer only · 329d4d83983cce32 · report fixture: signature x:ndarray/2/float, tol:float/0/float · returned bool · level recorded 2026-09-24
kernel in 22 papers simple_compositional_tasks/models/kernel_machines.py ran · our draft was wrong no licence file found · pointer only · 4dd4ca690ab53814 · report fixture: signature x1:ndarray/2/int, x2:ndarray/2/int, sims:dict/0/? · returned ndarray · level recorded 2026-09-24
margin simple_compositional_tasks/models/kernel_machines.py ran · violated contract fingerprinted no licence file found · pointer only · 5c2f9837b73be1db · report fixture: signature preds:ndarray/2/float, targets:ndarray/2/float · returned float32 · level recorded 2026-09-24
mse in 78 papers
census: 5 implementations from 6 papers, 4 distinct outputs across 3 buckets, one shared input per bucket
simple_compositional_tasks/models/kernel_machines.py ran · violated contract fingerprinted no licence file found · pointer only · 577f80aee94bb229 · report fixture: signature preds:ndarray/2/float, targets:ndarray/2/float · returned float64 · level recorded 2026-09-24
sims_pos_def simple_compositional_tasks/models/kernel_machines.py ran · our draft was wrong no licence file found · pointer only · 83c8b732856a57d7 · report fixture: signature sims:dict/0/?, comps:list/1/int · returned bool · level recorded 2026-09-22
KernelMachine simple_compositional_tasks/models/kernel_machines.py unverified no licence file found · pointer only · acb11a7bf4474e2f · report
KernelMachines simple_compositional_tasks/models/kernel_machines.py unverified no licence file found · pointer only · fff4dd3a3388381f · report
NonEquivariantKernelMachine simple_compositional_tasks/models/kernel_machines.py unverified no licence file found · pointer only · 7f5ae78b574db2df · report
kernel_index simple_compositional_tasks/models/kernel_machines.py unverified no licence file found · pointer only · 8a698dc7543122d9 · 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.

1 assembly on this paper: 1 harvested, 0 generated.

harvested Assembly for A kernel theory of compositional generalization (graph method: A kernel theory of compositional generalization). 12 bricks; individually on a fixture: 8 ran (of which 0 constructed an object rather than computing a result; 5 with no instrument failure: 1 honoured, 3 violated, 1 with no contract checked; 3 where Syntology's instrument failed) · 4 unverified. The assembly has not been run end-to-end.

  1. KernelMachines entry · level 0 · unverified · fff4dd3a3388381f
  2. KernelMachine helper · level 0 · unverified · acb11a7bf4474e2f
  3. NonEquivariantKernelMachine helper · level 0 · unverified · 7f5ae78b574db2df
  4. sims_pos_def helper · level 1 · ran · our draft was wrong · 83c8b732856a57d7
  5. CSVDataFrame helper · level 1 · ran · 8d0ab110639937d8
  6. complete_sims helper · level 1 · ran · our draft was wrong · 9ae529d03358d2a1
  7. kernel model · level 1 · ran · our draft was wrong · 4dd4ca690ab53814
  8. kernel_index helper · level 0 · unverified · 8a698dc7543122d9
  9. is_pos_def helper · level 1 · ran · violated contract · 329d4d83983cce32
  10. binary_accuracy helper · level 1 · ran · honoured contract · a8f89e3acd77db3a
  11. margin helper · level 1 · ran · violated contract · 5c2f9837b73be1db
  12. mse helper · level 1 · ran · violated contract · 577f80aee94bb229

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="2405.16391") returns these assemblies with each brick's role and wiring (how to connect).

Tasks

MemorizationRepresentation Learning

Results from the paper archive 2025-07-28

No leaderboard rows for this paper in the archive.

Methods

SET

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