SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2210.12786 · EMNLP · 2022

When Can Transformers Ground and Compose: Insights from Compositional Generalization Benchmarks

Navin Goyal, Ankur Sikarwar, Arkil Patel

arXiv · PDF · Open in the Atlas

Code that ran

1 of 8 samples (functions and classes) lifted from this paper's linked repositories executed on a synthesized input in Syntology's jail (0 of them constructed an object rather than computing a result; 0 honoured the contract we drafted, 0 violated it and 1 has no contract verdict; 0 failures of our instrument; 7 no recorded run), measured 2026-11-22; repository links from the Papers with Code archive attributed under CC BY-SA, no affiliation; bodies inline only under a permissive licence as GitHub detects it. Not a reproduction of the paper's results.

"Ran" means the function executed on a synthesized input and returned a value. It is not a reproduction of the paper's results. This record, dated, as JSON: /api/ran/2210.12786.json · badge: /api/ran/2210.12786.svg

Each repository's role is a label: a classifier's output recorded on the link, or, where no role is recorded, what the Papers with Code archive marked. The line under it names its basis. Neither says who wrote the code.

RepositoryRoleRan
ankursikarwar/grounded-compositional-generalization canonical (classifier label)
classifier basis not recorded on the link
1 ran · 7 no recorded run
FunctionStatusWhere it lives
calc_target_accuracy Ran
ran on a synthesized input
ankursikarwar/grounded-compositional-generalization/reascan/src/train_target.py
code served (permissive licence) · get_code("8630adb451905844")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned tuple. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch

torch.manual_seed(0)

# world_logits: a batch of logits, each sample should have argmax operation
# The function iterates through world_logits and calls torch.argmax(sample)
world_logits = torch.randn(2, 4)  # batch_size=2, logit_dim=4

# target_gridnum: target class indices for each sample in batch
target_gridnum = torch.tensor([2, 1])  # batch_size=2, matching world_logits

# batch_index: integer index of current batch
batch_index = 0

# config: object with batch_size attribute
class Config:
    def __init__(self):
        self.batch_size = 32

config = Config()

# errors: list to collect error indices, or None
errors = []

ARGS = {
    "world_logits": world_logits,
    "target_gridnum": target_gridnum,
    "batch_index": batch_index,
    "config": config,
    "errors": errors,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type tuple — a tuple output has no array shape of its own; none is stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
greedy_decode_batch_dual_mcnn Not yet run
unverified (harvested, no recorded run)
ankursikarwar/grounded-compositional-generalization/reascan/src/train_mcnn.py
code served (permissive licence) · get_code("03fb912963c88308")
How it ran

no invocation recorded.

train_one_epoch_mcnn Not yet run
unverified (harvested, no recorded run)
ankursikarwar/grounded-compositional-generalization/reascan/src/train_mcnn.py
code served (permissive licence) · get_code("d75fc675031d5c92")
How it ran

no invocation recorded.

train_one_epoch_target Not yet run
unverified (harvested, no recorded run)
ankursikarwar/grounded-compositional-generalization/reascan/src/train_target.py
code served (permissive licence) · get_code("2a1a8ca2202a8076")
How it ran

no invocation recorded.

train_one_epoch_target_simple Not yet run
unverified (harvested, no recorded run)
ankursikarwar/grounded-compositional-generalization/reascan/src/train_target_simple.py
code served (permissive licence) · get_code("f9e94beba0354b20")
How it ran

no invocation recorded.

validate_one_epoch_mcnn Not yet run
unverified (harvested, no recorded run)
ankursikarwar/grounded-compositional-generalization/reascan/src/train_mcnn.py
code served (permissive licence) · get_code("5094ea97fa470b20")
How it ran

no invocation recorded.

validate_one_epoch_target Not yet run
unverified (harvested, no recorded run)
ankursikarwar/grounded-compositional-generalization/reascan/src/train_target.py
code served (permissive licence) · get_code("1a805e8b4534d128")
How it ran

no invocation recorded.

validate_one_epoch_target_simple Not yet run
unverified (harvested, no recorded run)
ankursikarwar/grounded-compositional-generalization/reascan/src/train_target_simple.py
code served (permissive licence) · get_code("958c6f494632868c")
How it ran

no invocation recorded.

Also present in other papers' code

Of 8 samples lifted from this paper's linked repositories, 8 carry a function identity and 0 do not. Of the 8 distinct functions with an identity, 0 are also present in other papers whose code we hold (0 with at least one paper dated earlier, 0 the same month, 0 later) and 8 are only here among the 52,977 papers whose code Syntology holds (as of 2026-09-26).

Only here means: among the 52,977 papers whose code Syntology holds (51,887 with lifted samples, 1,090 through their assemblies only), as of 2026-09-26, no other paper's lifted samples or current assembly bricks contain an identical function body after normalisation.

Presence, not direction: an identical function body in two papers' repositories does not say who wrote it, which came from which, or that either paper's method rests on it. Matching is exact after normalisation, so a renamed or edited function is not matched. Dates are when papers appeared, not when code was written. Only .py files the harvester lifted are covered; notebooks are not read. A new-style arXiv id YYMM.NNNN dates a paper to the month 20YY-MM; otherwise Paper.year dates it to the year; otherwise it is undated. Dates are when the paper appeared, not when the code was written. Read 2026-11-22.

Repositories linked to this paper

Some links come from the archived Papers with Code dataset (CC BY-SA 4.0): attribution and licence.

Abstract

Humans can reason compositionally whilst grounding language utterances to the real world. Recent benchmarks like ReaSCAN (Wu et al., 2021) use navigation tasks grounded in a grid world to assess whether neural models exhibit similar capabilities. In this work, we present a simple transformer-based model that outperforms specialized architectures on ReaSCAN and a modified version (Qiu et al., 2021) of gSCAN (Ruis et al., 2020). On analyzing the task, we find that identifying the target location in the grid world is the main challenge for the models. Furthermore, we show that a particular split in ReaSCAN, which tests depth generalization, is unfair. On an amended version of this split, we show that transformers can generalize to deeper input structures. Finally, we design a simpler grounded compositional generalization task, RefEx, to investigate how transformers reason compositionally. We show that a single self-attention layer with a single head generalizes to novel combinations of object attributes. Moreover, we derive a precise mathematical construction of the transformer's computations from the learned network. Overall, we provide valuable insights about the grounded compositional generalization task and the behaviour of transformers on it, which would be useful for researchers working in this area.

For agents

The same record, over MCP at https://syntology.ai/mcp:

get_harvested_code_for_paper("2210.12786")
get_code_for_paper("2210.12786")
have("2210.12786")

The run record, dated, one paper per request, free:

curl https://syntology.ai/api/ran/2210.12786.json

A badge for a README (the split and the date, never a ratio):

[![Syntology run record](https://syntology.ai/api/ran/2210.12786.svg)](https://syntology.ai/paper/2210.12786)

Connect an agent — have() is free.