SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2402.14735 · ICML · 2024

How Transformers Learn Causal Structure with Gradient Descent

Jason Lee, Alex Damian, Eshaan Nichani

arXiv · PDF · Open in the Atlas

Code that ran

4 of 4 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, 1 violated it and 0 have no contract verdict; 3 failures of our instrument; 0 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/2402.14735.json · badge: /api/ran/2402.14735.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
eshnich/transformers-learn-causal-structure canonical (classifier label)
classifier basis: census
4 ran · 0 no recorded run
FunctionStatusWhere it lives
get_stationary Ran
ran, violated the contract we drafted
eshnich/transformers-learn-causal-structure/problems.py
pointer only (licence: NONE) · get_code("cf90fc4b68f425dc")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, eshnich/transformers-learn-causal-structure when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

plot_A Ran
ran; our contract draft was wrong, not the code
eshnich/transformers-learn-causal-structure/plots.py
pointer only (licence: NONE) · get_code("72cb679283a62f39")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, eshnich/transformers-learn-causal-structure when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

plot_A1 Ran
ran; our contract draft was wrong, not the code
eshnich/transformers-learn-causal-structure/plots.py
pointer only (licence: NONE) · get_code("ac819defc5e7d02b")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, eshnich/transformers-learn-causal-structure when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

plot_losses Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
eshnich/transformers-learn-causal-structure/plots.py
pointer only (licence: NONE) · get_code("aa51a3b782a251f4")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, eshnich/transformers-learn-causal-structure when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

Also present in other papers' code

Of 4 samples lifted from this paper's linked repositories, 4 carry a function identity and 0 do not. Of the 4 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 4 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

The incredible success of transformers on sequence modeling tasks can be largely attributed to the self-attention mechanism, which allows information to be transferred between different parts of a sequence. Self-attention allows transformers to encode causal structure which makes them particularly suitable for sequence modeling. However, the process by which transformers learn such causal structure via gradient-based training algorithms remains poorly understood. To better understand this process, we introduce an in-context learning task that requires learning latent causal structure. We prove that gradient descent on a simplified two-layer transformer learns to solve this task by encoding the latent causal graph in the first attention layer. The key insight of our proof is that the gradient of the attention matrix encodes the mutual information between tokens. As a consequence of the data processing inequality, the largest entries of this gradient correspond to edges in the latent causal graph. As a special case, when the sequences are generated from in-context Markov chains, we prove that transformers learn an induction head [27]. We confirm our theoretical findings by showing that transformers trained on our in-context learning task are able to recover a wide variety of causal structures.

For agents

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

get_harvested_code_for_paper("2402.14735")
get_code_for_paper("2402.14735")
have("2402.14735")

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

curl https://syntology.ai/api/ran/2402.14735.json

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

[![Syntology run record](https://syntology.ai/api/ran/2402.14735.svg)](https://syntology.ai/paper/2402.14735)

Connect an agent — have() is free.