Papers › Loss-to-Loss Prediction: Scaling Laws for All Datasets

Loss-to-Loss Prediction: Scaling Laws for All Datasets

19 Nov 2024arXiv:2411.12925archive 2025-07-28

David Brandfonbrener, Nikhil Anand, Nikhil Vyas, Eran Malach, Sham Kakade

While scaling laws provide a reliable methodology for predicting train loss across compute scales for a single data distribution, less is known about how these predictions should change as we change the distribution. In this paper, we derive a strategy for predicting one loss from another and apply it to predict across different pre-training datasets and from pre-training data to downstream task data. Our predictions extrapolate well even at 20x the largest FLOP budget used to fit the curves. More precisely, we find that there are simple shifted power law relationships between (1) the train losses of two models trained on two separate datasets when the models are paired by training compute (train-to-train), (2) the train loss and the test loss on any downstream distribution for a single model (train-to-test), and (3) the test losses of two models trained on two separate train datasets (test-to-test). The results hold up for pre-training datasets that differ substantially (some are entirely code and others have no code at all) and across a variety of downstream tasks. Finally, we find that in some settings these shifted power law relationships can yield more accurate predictions than extrapolating single-dataset scaling laws.

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 12 code samples harvested from 2 repositories linked to this paper: 11 ran (of which 0 constructed an object rather than computing a result; 9 with no instrument failure: 2 honoured, 0 violated, 7 with no contract checked; 2 where Syntology's instrument failed) · 1 unverified. Where Syntology's instrument failed: 2 where our contract draft was wrong; failures of Syntology's instrument, not of the code.

By repository: official repository: 12 samples from 2 repositories, 11 ran (9 with no instrument failure; 2 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 between 2026-09-24 and 2026-09-25 (UTC): the dates Syntology's graph wrote each sample's verification level, not when the code executed.

kempnerinstitute/loss-to-loss-olmo officialmentioned in papermentioned on GitHubpytorchApache-2.0Syntology: 3 samples, 2 ran (0 instrument-failed) · 1 unverified report
KempnerInstitute/loss-to-loss-notebooks officialmentioned on GitHubjaxSyntology: 9 samples, 9 ran (2 instrument-failed) · 0 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

12 samples harvested: 11 ran (of which 0 constructed an object rather than computing a result; 9 with no instrument failure: 2 honoured, 0 violated, 7 with no contract checked; 2 where Syntology's instrument failed) · 1 unverified. “Where Syntology's instrument failed” counts failures of Syntology's instrument, not of the code. Syntology's record for this page last changed ; when this build read Syntology's graph is in the build record. That date is not when the samples ran.

2ran · honoured contract
2ran · our draft was wrong
7ran
1unverified

Licence: 10 of the 12 samples are pointer only, meaning Syntology does not serve that copy's text. For 1 of them, this paper's copy is under a permissive licence and cleared, but the stored copy of the same code that Syntology serves is not; the licence label's tooltip gives Syntology's recorded reason. This page shows no code text for any sample; each one links to its file in the repository.

Harvested from 2 repositories linked to this paper, official or community; the list below is grouped by repository, each group headed by its name and what it is to the paper. “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.

kempnerinstitute/loss-to-loss-olmo official repository 3 samples, 2 ran (0 instrument-failed) · 1 unverified
causal_attention_bias in 17 papers olmo/model.py ran · honoured contract Apache-2.0 (permissive) here; the stored copy is pointer only · f9abf6d74699cc1c · report fixture: signature seq_len:int/0/int, device:device/0/? · returned Tensor · level recorded 2026-09-24
load_score in 6 papers olmo/registry.py ran · honoured contract Apache-2.0 (permissive) · 64997ebee7737a54 · report fixture: signature key:str/0/str · returned list · level recorded 2026-09-25
fix_optim_state_dict in 6 papers olmo/optim.py unverified Apache-2.0 (permissive) · f751a49c75709d42 · report
KempnerInstitute/loss-to-loss-notebooks official repository 9 samples, 9 ran (2 instrument-failed) · 0 unverified
chinchilla_curve utils_scaling.py ran no licence file found · pointer only · 7c794f11c960d3b5 · report fixture: signature params:ArrayImpl/1/float, N:ArrayImpl/1/float, D:ArrayImpl/1/float, masks:ArrayImpl/1/float, vals:ArrayImpl/1/float · returned ArrayImpl · level recorded 2026-09-25
fit in 23 papers utils_scaling.py ran no licence file found · pointer only · f916802c9c8b374e · report fixture: signature N:ArrayImpl/1/float, D:ArrayImpl/1/float, L:ArrayImpl/1/float, alpha:NoneType/0/?, beta:NoneType/0/?, E:NoneType/0/?, A:NoneType/0/?, B:NoneType/0/?, kaplan:bool/0/bool · returned tuple · level recorded 2026-09-25
generate_dataset in 41 papers theory/theory_sweep_gpu.py ran no licence file found · pointer only · e346f8db1d5b5881 · report fixture: signature D:int/0/int, M:int/0/int, beta:float/0/float, sigma:float/0/float, key:ArrayImpl/1/int · returned tuple · level recorded 2026-09-25
generate_target theory/theory_sweep_gpu.py ran no licence file found · pointer only · e62f060d13677db3 · report fixture: signature N:int/0/int, M:int/0/int, sigma_w:int/0/int, sigma_v:int/0/int, key:ArrayImpl/1/int · returned tuple · level recorded 2026-09-25
get_params in 53 papers curve_fitting.py ran · our draft was wrong no licence file found · pointer only · f9b4804cb87312af · report fixture: signature params:dict/0/? · returned tuple · level recorded 2026-09-25
get_simple_power_law utils_plotting.py ran · our draft was wrong no licence file found · pointer only · 326b5329149b1741 · report fixture: signature c:float/0/float · returned function · level recorded 2026-09-25
kaplan_curve utils_scaling.py ran no licence file found · pointer only · 5f5df690732bb25c · report fixture: signature params:ArrayImpl/1/float, N:ArrayImpl/0/float, D:ArrayImpl/1/float, masks:ArrayImpl/1/bool, vals:ArrayImpl/1/float · returned ArrayImpl · level recorded 2026-09-25
plot_with_line utils_plotting.py ran no licence file found · pointer only · 1692167af263d3d8 · report fixture: signature ax:Axes/0/?, x:ndarray/1/float, y:ndarray/1/float, color:str/0/str, label:str/0/str, logx:bool/0/bool, logy:bool/0/bool, extrapolate:bool/0/bool, add_x:NoneType/0/? · returned DataFrame · level recorded 2026-09-25
power_law_with_constant utils_plotting.py ran fingerprinted no licence file found · pointer only · 85ebee271bb746fe · report fixture: signature x:ndarray/1/float, a:float/0/float, b:float/0/float, c:float/0/float, d:float/0/float · returned ndarray · level recorded 2026-09-25

Tasks

All

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