Papers › Data Feedback Loops: Model-driven Amplification of Dataset Biases

Data Feedback Loops: Model-driven Amplification of Dataset Biases

8 Sep 2022arXiv:2209.03942archive 2025-07-28

Rohan Taori, Tatsunori B. Hashimoto

Datasets scraped from the internet have been critical to the successes of large-scale machine learning. Yet, this very success puts the utility of future internet-derived datasets at potential risk, as model outputs begin to replace human annotations as a source of supervision. In this work, we first formalize a system where interactions with one model are recorded as history and scraped as training data in the future. We then analyze its stability over time by tracking changes to a test-time bias statistic (e.g. gender bias of model predictions). We find that the degree of bias amplification is closely linked to whether the model's outputs behave like samples from the training distribution, a behavior which we characterize and define as consistent calibration. Experiments in three conditional prediction scenarios - image classification, visual role-labeling, and language generation - demonstrate that models that exhibit a sampling-like behavior are more calibrated and thus more stable. Based on this insight, we propose an intervention to help calibrate and stabilize unstable feedback systems. Code is available at https://github.com/rtaori/data_feedback.

PaperPDFCodeCode Syntology ran

In Syntology View this paper on Syntology, its page in Syntology's graph. That page adds which of its functions are also present in other papers' code Syntology holds, with dates, and its dated run record as JSON and a README badge. It also lists the repositories linked to the paper, the abstract and the calls for agents.

Open this paper in Syntology's Atlas, the map of the papers in Syntology's graph and their citations.

For agents, on Syntology's MCP service (how to connect):

Code

Syntology 17 code samples harvested from 1 repository linked to this paper: 15 ran (of which 0 constructed an object rather than computing a result; 15 with no instrument failure: 0 honoured, 0 violated, 15 with no contract checked; 0 where Syntology's instrument failed) · 2 unverified. “Where Syntology's instrument failed” counts failures of Syntology's instrument, not of the code; here there are none.

By repository: official repository: 17 samples from 1 repository, 15 ran (15 with no instrument failure; 0 where Syntology's instrument failed). The run record, sample by sample. “Ran” means executed on a synthesized input, not that the code is correct or reproduces the paper.

Levels written on 2026-09-25 (UTC): the dates Syntology's graph wrote each sample's verification level, not when the code executed.

rtaori/data_feedback officialmentioned in papermentioned on GitHubpytorchApache-2.0Syntology: 17 samples, 15 ran (0 instrument-failed) · 2 unverified report

Repository list and official/mentioned flags are the archive's, frozen 2025-07-28. Reachability, where shown, is from one Syntology probe window (2026-09-16 to 2026-09-18); repositories not probed show nothing. GitHub stars are not tracked.

Code Syntology ran Syntology

17 samples harvested: 15 ran (of which 0 constructed an object rather than computing a result; 15 with no instrument failure: 0 honoured, 0 violated, 15 with no contract checked; 0 where Syntology's instrument failed) · 2 unverified. “Where Syntology's instrument failed” counts failures of Syntology's instrument, not of the code. Syntology's record for this page has not changed since , the first build that kept a record date for it; when this build read Syntology's graph is in the build record. That date is not when the samples ran.

15ran
2unverified

Licence: 0 of the 17 samples are pointer only, meaning Syntology does not serve that copy's text. This page shows no code text for any sample; each one links to its file in the repository.

Harvested from rtaori/data_feedback. “Ran” means the sample executed on a synthesized input. It does not mean the output is correct, and nothing here reproduces the paper's results. A line beginning “fixture:” under a sample that ran gives the arguments Syntology's fixture called the sample with, each as name:type/rank/element kind (? where none was recorded), and the type of what the fixture's run returned, exactly as the harness recorded them; the arguments and the returned type were recorded by two separate runs of the same fixture. No output value is shown, and a returned type says nothing about whether the output was right. “Level recorded” gives the UTC date when Syntology's graph wrote the sample's verification level, not when the code executed; where the level was written before Syntology's graph kept that date, the row says “level recorded, date not recorded”. “Honoured” and “violated” refer to a contract Syntology drafted from the code itself. “Our draft was wrong” and “the contract check could not run” are failures of Syntology's instrument, not of the code: the sample ran on Syntology's synthesized input, and then the drafted contract did not describe it, or the check itself raised an error (RAISES), lacked a dependency (DEP_MISSING) or ran out of time (TIMEOUT); hover the label for which. A metamorphic tier, where shown, is evidence of the output's shape, not that the sample implements the method.

Each sample ends with its code_sha256. A sample's code_sha256 is Syntology's identity for that exact code. It is printed in full (16 hexadecimal characters for every sample in this build), and get_code(code_sha256="…") takes it exactly as printed. An agent calls it on Syntology's MCP service (how to connect); click an identity to copy that call.

Repository labels, per sample. official repository: The archive marks this repository official for the paper. named in the paper: The archive records that the paper mentions this repository; it is not marked official. community (archive-listed): In the archive's code links for this paper, not marked official and not recorded as mentioned in the paper. found in paper text by Syntology: Syntology found this repository in the paper's own text; whether it is the authors' implementation is not asserted. community: Not in the archive's code links for this paper; a community repository Syntology harvested. Samples are grouped by repository: a repository marked official first, then by how directly the repository is tied to the paper, then in the archive's order; never by how many ran. Licence labels name the repository's licence as recorded at harvest. “Pointer only” means Syntology does not serve that copy's text, for one of six reasons: no licence file was found; the licence was not identified; the licence is recorded as permissive but that copy's record is not marked cleared; the licence is outside the permissive list Syntology serves text under (MIT, Apache-2.0, BSD and similar); this paper's copy was not located, so its licence is not recorded; or this paper's copy is cleared, but the stored copy of the same code that Syntology serves is not. Some licences outside that list permit redistribution, such as WTFPL, and GPL-3.0 under its conditions; they are simply not on the list. Hover a licence label for the reason. File links open the file on GitHub at the default branch, which may have changed since the harvest.

A function or class name links to its page when the name occurs in the harvested code of at least 5 papers; the name is a string, not an identity.

rtaori/data_feedback official repository 17 samples, 15 ran (0 instrument-failed) · 2 unverified
batch_norm in 46 papers image_classification/baidunet.py ran Apache-2.0 (permissive) · 98c5a9335415160f · report fixture: signature num_channels:int/0/int, bn_bias_init:float/0/float, bn_bias_freeze:bool/0/bool, bn_weight_init:float/0/float, bn_weight_freeze:bool/0/bool · returned BatchNorm2d · level recorded 2026-09-25
collapse_annotations visual_role_labeling/data.py ran Apache-2.0 (permissive) · faa2eeb8672f4711 · report fixture: signature dataset:dict/0/?, use_majority:bool/0/bool · returned dict · level recorded 2026-09-25
conv_bn in 88 papers image_classification/baidunet.py ran Apache-2.0 (permissive) · ac25750813c3e6f9 · report fixture: signature c_in:int/0/int, c_out:int/0/int, bn_weight_init:float/0/float, bn_bias_init:NoneType/0/?, bn_bias_freeze:bool/0/bool, bn_weight_freeze:bool/0/bool · returned dict · level recorded 2026-09-25
format_dict visual_role_labeling/baseline_crf.py ran Apache-2.0 (permissive) · 81d473f07388d56a · report fixture: signature d:dict/0/?, s:str/0/str, p:str/0/str · returned str · level recorded 2026-09-25
get_bias_std_lines plotting/plotting_utils.py ran Apache-2.0 (permissive) · 6d38eeb275e78417 · report fixture: signature runs_stats:list/1/dict, preds_key:str/0/str, filter_fxn:callable/0/?, targs_key:str/0/str · returned tuple · level recorded 2026-09-25
load_cifar5m image_classification/data.py ran Apache-2.0 (permissive) · 819ba6bdeceaaf30 · report fixture: signature root:str/0/str, num_parts:int/0/int · returned TensorDataset · level recorded 2026-09-25
load_cinic10 image_classification/data.py ran Apache-2.0 (permissive) · 804d4cfbb5a899a0 · report fixture: signature root:str/0/str · returned TensorDataset · level recorded 2026-09-25
load_data in 700 papers image_classification/data.py ran Apache-2.0 (permissive) · dfae76f2b889026e · report fixture: signature dataset:str/0/str, data_dir:str/0/str · returned TensorDataset · level recorded 2026-09-25
rand_split_dataset visual_role_labeling/data.py ran Apache-2.0 (permissive) · 71cfaeb6fde82196 · report fixture: signature dataset:dict/0/?, partition_size:int/0/int · returned tuple · level recorded 2026-09-25
rand_split_test_set visual_role_labeling/data.py ran Apache-2.0 (permissive) · cc062ce9df817b3a · report fixture: signature dataset:dict/0/?, test_set_imgs_per_class:int/0/int · returned tuple · level recorded 2026-09-25
shuffle_tensor image_classification/utils.py ran fingerprinted Apache-2.0 (permissive) · 9a0a2bca8191c1b6 · report fixture: signature tensor:Tensor/2/float · returned Tensor · level recorded 2026-09-25
theorem1_bound plotting/feedback_bound.py ran Apache-2.0 (permissive) · 116d1355faa1c187 · report fixture: signature t:int/0/int, n0:int/0/int, m:int/0/int, k:int/0/int, delta_n0:float/0/float · returned float64 · level recorded 2026-09-25
theorem1_continual_bound plotting/feedback_bound.py ran Apache-2.0 (permissive) · 7e564a313204ad07 · report fixture: signature T:int/0/int, n0:int/0/int, m:int/0/int, k:int/0/int, delta_n0:float/0/float · returned ndarray · level recorded 2026-09-25
tokenize_generation language_generation/data.py ran fingerprinted Apache-2.0 (permissive) · 997469cb5a9c5f62 · report fixture: signature prompt_ids:list/1/int · returned dict · level recorded 2026-09-25
tokenize_training language_generation/data.py ran fingerprinted Apache-2.0 (permissive) · a5dc158924c783cc · report fixture: signature prompt_ids:list/1/int, continuation_ids:list/1/int · returned dict · level recorded 2026-09-25
load_rtp_dataset language_generation/data.py unverified Apache-2.0 (permissive) · 935670c8c277b699 · report
residual in 9 papers image_classification/baidunet.py unverified Apache-2.0 (permissive) · dc51887b30539963 · report

Tasks

Image ClassificationText Generationimage-classification

Results from the paper archive 2025-07-28

No leaderboard rows for this paper in the archive.

Report a problem or propose a change · a person checks every report against the paper or source before anything changes; decisions are listed on /corrections