SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2310.04625 · 2023

Copy Suppression: Comprehensively Understanding an Attention Head

arXiv · PDF · Open in the Atlas

Code that ran

3 of 5 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; 2 honoured the contract we drafted, 0 violated it and 0 have no contract verdict; 1 failure of our instrument; 2 no recorded run), measured 2026-11-21; 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/2310.04625.json · badge: /api/ran/2310.04625.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
callummcdougall/seri-mats-2023-streamlit-pages canonical (classifier label)
classifier basis: census
3 ran · 2 no recorded run
FunctionStatusWhere it lives
get_duplicate_token_head_detection_pattern Ran
ran; our contract draft was wrong, not the code
callummcdougall/seri-mats-2023-streamlit-pages/transformer_lens/head_detector.py
pointer only (licence: MIT); the stored copy get_code serves, from dalab/transformer-hessian, is pointer only (licence: NONE) · get_code("ba931d66d841af5d")
How it ran

The body of this sample is withheld: this paper's link allows the body inline (MIT), but get_code serves the stored copy, first harvested from dalab/transformer-hessian, and GitHub detected no licence in that repository 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.

get_previous_token_head_detection_pattern Ran
ran, honoured the contract we drafted
callummcdougall/seri-mats-2023-streamlit-pages/transformer_lens/head_detector.py
pointer only (licence: MIT); the stored copy get_code serves, from dalab/transformer-hessian, is pointer only (licence: NONE) · get_code("79a2d1104ca128a0")
How it ran

The body of this sample is withheld: this paper's link allows the body inline (MIT), but get_code serves the stored copy, first harvested from dalab/transformer-hessian, and GitHub detected no licence in that repository 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.

sanity_check Ran
ran, honoured the contract we drafted
callummcdougall/seri-mats-2023-streamlit-pages/transformer_lens/evals.py
pointer only (licence: MIT); the stored copy get_code serves, from dalab/transformer-hessian, is pointer only (licence: NONE) · get_code("49d7660823635dab")
How it ran

The body of this sample is withheld: this paper's link allows the body inline (MIT), but get_code serves the stored copy, first harvested from dalab/transformer-hessian, and GitHub detected no licence in that repository 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.

convert_hf_model_config Not yet run
unverified (harvested, no recorded run)
callummcdougall/seri-mats-2023-streamlit-pages/transformer_lens/loading_from_pretrained.py
code served (permissive licence) · get_code("a40a57f1e417e33c")
How it ran

no invocation recorded.

get_official_model_name Not yet run
unverified (harvested, no recorded run)
callummcdougall/seri-mats-2023-streamlit-pages/transformer_lens/loading_from_pretrained.py
code served (permissive licence) · get_code("e5d367becb0bf75f")
How it ran

no invocation recorded.

Also present in other papers' code

Of 5 samples lifted from this paper's linked repositories, 5 carry a function identity and 0 do not. Of the 5 distinct functions with an identity, 3 are also present in other papers whose code we hold (0 with at least one paper dated earlier, 0 the same month, 3 later) and 2 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.

FunctionAlso present inEarliest-datedOthers named
get_duplicate_token_head_detection_pattern 1 other paper
1 later
2410.10986 (2024-10)
What Does It Mean to Be a Transformer? Insights from a Theoretical Hessian Analysis
get_previous_token_head_detection_pattern 1 other paper
1 later
2410.10986 (2024-10)
What Does It Mean to Be a Transformer? Insights from a Theoretical Hessian Analysis
sanity_check 1 other paper
1 later
2410.10986 (2024-10)
What Does It Mean to Be a Transformer? Insights from a Theoretical Hessian Analysis

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 2026-11-21.

Repositories linked to this paper

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

Abstract

We present a single attention head in GPT-2 Small that has one main role across the entire training distribution. If components in earlier layers predict a certain token, and this token appears earlier in the context, the head suppresses it: we call this copy suppression. Attention Head 10.7 (L10H7) suppresses naive copying behavior which improves overall model calibration. This explains why multiple prior works studying certain narrow tasks found negative heads that systematically favored the wrong answer. We uncover the mechanism that the Negative Heads use for copy suppression with weights-based evidence and are able to explain 76.9% of the impact of L10H7 in GPT-2 Small. To the best of our knowledge, this is the most comprehensive description of the complete role of a component in a language model to date. One major effect of copy suppression is its role in self-repair. Self-repair refers to how ablating crucial model components results in downstream neural network parts compensating for this ablation. Copy suppression leads to self-repair: if an initial overconfident copier is ablated, then there is nothing to suppress. We show that self-repair is implemented by several mechanisms, one of which is copy suppression, which explains 39% of the behavior in a narrow task. Interactive visualisations of the copy suppression phenomena may be seen at our web app https://copy-suppression.streamlit.app/

For agents

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

get_harvested_code_for_paper("2310.04625")
get_code_for_paper("2310.04625")
have("2310.04625")

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

curl https://syntology.ai/api/ran/2310.04625.json

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

[![Syntology run record](https://syntology.ai/api/ran/2310.04625.svg)](https://syntology.ai/paper/2310.04625)

Connect an agent — have() is free.