SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2602.05910 · 2026

Chunky Post-Training: Data Driven Failures of Generalization

Collin Burns, John Schulman, Sara Price, Allison Qi, Seoirse Murray, Timothy Qian

arXiv · PDF · Open in the Atlas

Code that ran

6 of 9 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; 1 honoured the contract we drafted, 0 violated it and 0 have no contract verdict; 5 failures of our instrument; 3 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/2602.05910.json · badge: /api/ran/2602.05910.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
seoirsem/SURF canonical (classifier label)
classifier basis: census
6 ran · 3 no recorded run
FunctionStatusWhere it lives
build_pseudo_sae Ran
ran, honoured the contract we drafted
seoirsem/SURF/surf/clustering/pseudo_sae.py
pointer only (licence: NONE) · get_code("def73f974f86fc6c")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, seoirsem/SURF 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_principle_from_rubric Ran
ran; our contract draft was wrong, not the code
seoirsem/SURF/surf/em_loop/judge.py
pointer only (licence: NONE) · get_code("2531f3751e78cc28")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, seoirsem/SURF 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_jsonl Ran
ran; our contract draft was wrong, not the code
seoirsem/SURF/surf/core/streaming.py
pointer only (licence: NONE) · get_code("0b4599e397fec376")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, seoirsem/SURF 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_rubric Ran
ran; our contract draft was wrong, not the code
seoirsem/SURF/surf/em_loop/judge.py
pointer only (licence: NONE) · get_code("bd18e4007c80e147")
How it ran

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

parse_model_string Ran
ran; our contract draft was wrong, not the code
seoirsem/SURF/surf/core/models.py
pointer only (licence: NONE) · get_code("26155e49501ab0db")
How it ran

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

render_jinja Ran
ran; our contract draft was wrong, not the code
seoirsem/SURF/surf/core/utils.py
pointer only (licence: NONE) · get_code("1892acdc179ff0b2")
How it ran

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

cluster_attributes Not yet run
unverified (harvested, no recorded run)
seoirsem/SURF/surf/clustering/cluster.py
pointer only (licence: NONE) · get_code("d08fffc56a789964")
How it ran

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

parse_xml_tags Not yet run
unverified (harvested, no recorded run)
seoirsem/SURF/surf/core/utils.py
pointer only (licence: NONE) · get_code("3d44325ea89abee7")
How it ran

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

parse_xml_tags_optional Not yet run
unverified (harvested, no recorded run)
seoirsem/SURF/surf/core/utils.py
pointer only (licence: NONE) · get_code("ce9e608d6b029852")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, seoirsem/SURF 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 9 samples lifted from this paper's linked repositories, 9 carry a function identity and 0 do not. Of the 9 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 9 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

LLM post-training involves many diverse datasets, each targeting a specific behavior. But these datasets encode incidental patterns alongside intended ones: correlations between formatting and content, narrow phrasings across diverse problems, and implicit associations arising from the discrete data curation process. These patterns are often invisible to developers yet salient to models, producing behaviors that surprise their creators, such as rejecting true facts presented in a particular question format. We call this chunky posttraining: the model learns spurious correlations as a result of distinct chunks of post-training data. We introduce SURF, a black-box pipeline which surfaces these unintended behaviors at run time, and TURF, a tool that traces these failures back to specific post-training data. Applying these tools to frontier models (Claude 4.5, GPT-5.1, Grok 4.1, Gemini 3) and open models (Tülu 3), we show that chunky post-training produces miscalibrated behaviors, which often result from imbalanced or underspecified chunks of post-training data.

For agents

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

get_harvested_code_for_paper("2602.05910")
get_code_for_paper("2602.05910")
have("2602.05910")

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

curl https://syntology.ai/api/ran/2602.05910.json

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

[![Syntology run record](https://syntology.ai/api/ran/2602.05910.svg)](https://syntology.ai/paper/2602.05910)

Connect an agent — have() is free.