SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2602.09987 · ICLR · 2026

Infusion: Shaping Model Behavior by Editing Training Data via Influence Functions

Jakob Foerster, Laura Ruis, Edward Grefenstette, Robert Kirk, J Rosser

arXiv · PDF · Open in the Atlas

Code that ran

9 of 13 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; 3 honoured the contract we drafted, 1 violated it and 1 has no contract verdict; 4 failures of our instrument; 4 no recorded run), measured 2026-11-22; 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/2602.09987.json · badge: /api/ran/2602.09987.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
jrosseruk/infusion canonical (classifier label)
classifier basis: census
9 ran · 4 no recorded run
FunctionStatusWhere it lives
checkpoints_exist Ran
ran, violated the contract we drafted
jrosseruk/infusion/caesar_prime/train_model.py
pointer only (licence: NONE) · get_code("72727ffcecbe5ef8")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, jrosseruk/infusion 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.

compute_baseline_contrastive Ran
ran; our contract draft was wrong, not the code
jrosseruk/infusion/caesar_prime/run_infusion_experiment.py
pointer only (licence: NONE) · get_code("430e1afa951dc987")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, jrosseruk/infusion 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.

estimate_loss Ran
ran, honoured the contract we drafted
jrosseruk/infusion/gpt_neo/run_animal_infusion.py
pointer only (licence: NONE) · get_code("4edd5e3ab7157b26")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, jrosseruk/infusion 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_checkpoint_dir Ran
ran; our contract draft was wrong, not the code
jrosseruk/infusion/caesar_prime/train_model.py
pointer only (licence: NONE) · get_code("305ee0e1a72d572e")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, jrosseruk/infusion 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_dataloader Ran
ran on a synthesized input
jrosseruk/infusion/infusion/dataloader.py
pointer only (licence: NONE) · get_code("c501a194e511cddf")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, jrosseruk/infusion 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.

load_dataset Ran
ran, honoured the contract we drafted
jrosseruk/infusion/caesar/dataset.py
pointer only (licence: NONE) · get_code("fafa0f0eaa20cda2")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, jrosseruk/infusion 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.

make_pad_collate_fn Ran
ran; our contract draft was wrong, not the code
jrosseruk/infusion/caesar_prime/run_infusion_experiment.py
pointer only (licence: NONE) · get_code("b2de257b6a4f5031")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, jrosseruk/infusion 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.

observable_gradient Ran
ran; our contract draft was wrong, not the code
jrosseruk/infusion/infusion/observable.py
pointer only (licence: NONE) · get_code("315b2dce5e4ee0bf")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, jrosseruk/infusion 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.

pad_sequences Ran
ran, honoured the contract we drafted
jrosseruk/infusion/caesar_prime/run_infusion_experiment.py
pointer only (licence: NONE) · get_code("b0b3a15939c84309")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, jrosseruk/infusion 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_bnb_config Not yet run
unverified (harvested, no recorded run)
jrosseruk/infusion/owl/model.py
pointer only (licence: NONE) · get_code("24da333b411655ff")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, jrosseruk/infusion 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. No invocation recorded.

load_checkpoint_data Not yet run
unverified (harvested, no recorded run)
jrosseruk/infusion/gpt_neo/run_animal_infusion.py
pointer only (licence: NONE) · get_code("d3ff33faab8bb2b3")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, jrosseruk/infusion 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. No invocation recorded.

load_llama2_base Not yet run
unverified (harvested, no recorded run)
jrosseruk/infusion/owl/model.py
pointer only (licence: NONE) · get_code("fe0e46d1f36bb204")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, jrosseruk/infusion 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. No invocation recorded.

load_model_for_inference Not yet run
unverified (harvested, no recorded run)
jrosseruk/infusion/gpt_neo/run_animal_infusion.py
pointer only (licence: NONE) · get_code("f6fff85fd1b06ee4")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, jrosseruk/infusion 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. No invocation recorded.

Also present in other papers' code

Of 13 samples lifted from this paper's linked repositories, 13 carry a function identity and 0 do not. Of the 13 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 13 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

Influence functions are commonly used to attribute model behavior to training documents. We explore the reverse: crafting training data that induces model behavior. Our framework, INFUSION, uses scalable influence-function approximations to compute small perturbations to training documents that induce targeted changes in model behavior through parameter shifts. We evaluate INFUSION on data poisoning tasks across vision and language domains. On CIFAR-10, we show that making subtle edits via INFUSION to just 0.2% (100/45,000) of the training documents can be competitive with the baseline of inserting a small number of explicit behavior examples. We also find that INFUSION transfers across architectures (ResNet ↔ CNN), suggesting a single poisoned corpus can affect multiple independently trained models. In preliminary language experiments, we characterize when our approach increases the probability of target behaviors and when it fails, finding it most effective at amplifying behaviors the model has already learned. Taken together, these results show that small, subtle edits to training data can systematically shape model behavior, underscoring the importance of training data interpretability for adversaries and defenders alike. We provide the code here: https://github.com/jrosseruk/infusion.

For agents

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

get_harvested_code_for_paper("2602.09987")
get_code_for_paper("2602.09987")
have("2602.09987")

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

curl https://syntology.ai/api/ran/2602.09987.json

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

[![Syntology run record](https://syntology.ai/api/ran/2602.09987.svg)](https://syntology.ai/paper/2602.09987)

Connect an agent — have() is free.