Papers › Differentially Private Adaptive Optimization with Delayed Preconditioners

Differentially Private Adaptive Optimization with Delayed Preconditioners

1 Dec 2022arXiv:2212.00309archive 2025-07-28

Tian Li, Manzil Zaheer, Ken Ziyu Liu, Sashank J. Reddi, H. Brendan McMahan, Virginia Smith

Privacy noise may negate the benefits of using adaptive optimizers in differentially private model training. Prior works typically address this issue by using auxiliary information (e.g., public data) to boost the effectiveness of adaptive optimization. In this work, we explore techniques to estimate and efficiently adapt to gradient geometry in private adaptive optimization without auxiliary data. Motivated by the observation that adaptive methods can tolerate stale preconditioners, we propose differentially private adaptive training with delayed preconditioners (DP^2), a simple method that constructs delayed but less noisy preconditioners to better realize the benefits of adaptivity. Theoretically, we provide convergence guarantees for our method for both convex and non-convex problems, and analyze trade-offs between delay and privacy noise reduction. Empirically, we explore DP^2 across several real-world datasets, demonstrating that it can improve convergence speed by as much as 4x relative to non-adaptive baselines and match the performance of state-of-the-art optimization methods that require auxiliary data.

PaperPDFCodeCode Syntology ranMethod as code

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.

Syntology holds a reference implementation of DP 2, generated from the paper text (not the authors' code), at level 2 on Syntology's ladder: V2 “two implementations GENERATED from the paper's signature and spec agree numerically (1e-6) on fixture inputs” “neither of the two ever read the authors' own code, so V2 measures how unambiguously the paper specifies the method, not fidelity to the authors' implementation (measured 2026-09-11, commit 75eaa77: cross_checked_by is a second LLM on 2,628 of 2,831 served samples)” (quoted from Syntology's server card). No rung compares this code against the authors' implementation. For agents: get_reference_implementation(name='DP 2') (how to connect); the free check have('2212.00309') needs no token.

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

Code

Syntology 31 code samples harvested from 1 repository linked to this paper: 23 ran (of which 0 constructed an object rather than computing a result; 4 with no instrument failure: 4 honoured, 0 violated, 0 with no contract checked; 19 where Syntology's instrument failed) · 8 unverified. Where Syntology's instrument failed: 17 where our contract draft was wrong; 2 where the contract check could not run; failures of Syntology's instrument, not of the code.

By repository: official repository: 31 samples from 1 repository, 23 ran (4 with no instrument failure; 19 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 2026-09-22 (UTC): the dates Syntology's graph wrote each sample's verification level, not when the code executed.

kenziyuliu/dp2 officialmentioned in papermentioned on GitHubjaxApache-2.0Syntology: 31 samples, 23 ran (19 instrument-failed) · 8 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

31 samples harvested: 23 ran (of which 0 constructed an object rather than computing a result; 4 with no instrument failure: 4 honoured, 0 violated, 0 with no contract checked; 19 where Syntology's instrument failed) · 8 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.

4ran · honoured contract
17ran · our draft was wrong
2ran; the contract check could not run
8unverified

Licence: 0 of the 31 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 kenziyuliu/dp2. “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. A census tag appears when Syntology's census ran implementations with that name on one shared input per bucket: it gives how many were compared, from how many papers, and how many distinct outputs they produced, and names this sample's bucket when this sample was one of them.

kenziyuliu/dp2 official repository 31 samples, 23 ran (19 instrument-failed) · 8 unverified
_config_tf_loader trainers/dp2_adagrad.py ran · our draft was wrong Apache-2.0 (permissive) · 57022752f9995f59 · report fixture: signature loader:_TensorSliceDataset/0/?, batch_size:int/0/int, shuffle_size:int/0/int, seed:int/0/int, drop_remainder:bool/0/bool · returned _PrefetchDataset · level recorded 2026-09-22
_create_tf_loader trainers/dp2_adagrad.py ran; the contract check could not run (RAISES) Apache-2.0 (permissive) · 5d8c2870379d3384 · report fixture: signature x_data:ndarray/2/float, y_data:ndarray/1/int, batch_size:int/0/int, shuffle_size:int/0/int, seed:int/0/int, drop_remainder:bool/0/bool, cache:bool/0/bool · returned _PrefetchDataset · level recorded 2026-09-22
_load_movielens_100k trainers/dp2_adagrad.py ran · our draft was wrong Apache-2.0 (permissive) · 09495c898e5d8e94 · report fixture: signature data_dir:str/0/str · returned tuple · level recorded 2026-09-22
bce_loss in 22 papers trainers/dp2_adagrad.py ran · honoured contract Apache-2.0 (permissive) · b3407cf1906789ea · report fixture: signature model:SimpleModel/0/?, params:dict/0/?, batch:tuple/1/ArrayImpl · returned ArrayImpl · level recorded 2026-09-22
global_l2_clip trainers/dp2_adagrad.py ran · our draft was wrong Apache-2.0 (permissive) · 735e56344d7303e1 · report fixture: signature tensor_struct:dict/0/?, clip:float/0/float · returned dict · level recorded 2026-09-22
global_l2_norm_sq trainers/dp2_adagrad.py ran · honoured contract Apache-2.0 (permissive) · bbd8d1efcf5829de · report fixture: signature tensor_struct:dict/0/? · returned ArrayImpl · level recorded 2026-09-22
logreg_classify trainers/dp2_adagrad.py ran · our draft was wrong Apache-2.0 (permissive) · 26f3490ca9bbe47e · report fixture: signature model:SimpleLogisticRegression/0/?, params:dict/0/?, batch_inputs:ArrayImpl/2/float, temperature:float/0/float · returned ArrayImpl · level recorded 2026-09-22
movielens_loader trainers/dp2_adagrad.py ran · our draft was wrong Apache-2.0 (permissive) · c3b75484cf237418 · report fixture: signature batch_size:int/0/int, matfac_dim_1:int/0/int, matfac_dim_2:int/0/int, density:float/0/float, seed:int/0/int, split:str/0/str, data_dir:str/0/str, is_eval:bool/0/bool, drop_remainder:bool/0/bool, public_split:float/0/float · returned tuple · level recorded 2026-09-22
mse_loss in 73 papers
census: 8 implementations from 12 papers, 6 distinct outputs across 4 buckets, one shared input per bucket
trainers/dp2_adagrad.py ran; the contract check could not run (RAISES) Apache-2.0 (permissive) · f320dea1d5541ada · report fixture: signature model:SimpleModel/0/?, params:ArrayImpl/2/float, batch:tuple/1/ArrayImpl, squeeze:bool/0/bool · returned ArrayImpl · level recorded 2026-09-22
num_params in 7 papers trainers/dp2_adagrad.py ran · honoured contract Apache-2.0 (permissive) · af4f5b33931d44c0 · report fixture: signature struct:dict/0/? · returned int64 · level recorded 2026-09-22
privatize_grad trainers/dp2_adagrad.py ran · our draft was wrong Apache-2.0 (permissive) · 591c9ea59777949b · report fixture: signature example_grads:dict/0/?, key:ArrayImpl/1/int, clip:float/0/float, noise_mult:float/0/float · returned dict · level recorded 2026-09-22
stackoverflow_tag_loader trainers/dp2_adagrad.py ran · our draft was wrong Apache-2.0 (permissive) · 19770acf0fa0aa44 · report fixture: signature batch_size:int/0/int, split:str/0/str, data_dir:str/0/str, is_eval:bool/0/bool, seed:int/0/int, drop_remainder:bool/0/bool, public_split:float/0/float · returned tuple · level recorded 2026-09-22
stackoverflow_tag_loader_cached trainers/dp2_adagrad.py ran · our draft was wrong Apache-2.0 (permissive) · 919aea51fdf26b6d · report fixture: signature batch_size:int/0/int, split:str/0/str, data_dir:str/0/str, is_eval:bool/0/bool, seed:int/0/int, drop_remainder:bool/0/bool, public_split:float/0/float · returned tuple · level recorded 2026-09-22
struct_add trainers/dp2_adagrad.py ran · our draft was wrong Apache-2.0 (permissive) · cca3cf0b80500ec6 · report fixture: signature struct_1:dict/0/?, struct_2:dict/0/? · returned dict · level recorded 2026-09-22
struct_add_scalar trainers/dp2_adagrad.py ran · our draft was wrong Apache-2.0 (permissive) · 34bf4ad4230cf55d · report fixture: signature struct:dict/0/?, value:float/0/float · returned dict · level recorded 2026-09-22
struct_div trainers/dp2_adagrad.py ran · our draft was wrong Apache-2.0 (permissive) · 7d340e00912389f7 · report fixture: signature struct_1:dict/0/?, struct_2:dict/0/? · returned dict · level recorded 2026-09-22
struct_div_scalar trainers/dp2_adagrad.py ran · our draft was wrong Apache-2.0 (permissive) · e488dedf8e381c96 · report fixture: signature struct:dict/0/?, factor:float/0/float · returned dict · level recorded 2026-09-22
struct_flatten trainers/dp2_adagrad.py ran · honoured contract Apache-2.0 (permissive) · 155ccee594f8679d · report fixture: signature struct:dict/0/? · returned ArrayImpl · level recorded 2026-09-22
struct_mul_scalar trainers/dp2_adagrad.py ran · our draft was wrong Apache-2.0 (permissive) · 6050b70841d335c1 · report fixture: signature struct:dict/0/?, factor:float/0/float · returned dict · level recorded 2026-09-22
struct_sq trainers/dp2_adagrad.py ran · our draft was wrong Apache-2.0 (permissive) · 145eda8014bfd82a · report fixture: signature struct:dict/0/? · returned dict · level recorded 2026-09-22
struct_sqrt trainers/dp2_adagrad.py ran · our draft was wrong Apache-2.0 (permissive) · ebb3869ec1266c1e · report fixture: signature struct:dict/0/? · returned dict · level recorded 2026-09-22
struct_sub trainers/dp2_adagrad.py ran · our draft was wrong Apache-2.0 (permissive) · 0bf286afd514640c · report fixture: signature struct_1:dict/0/?, struct_2:dict/0/? · returned dict · level recorded 2026-09-22
struct_zeros_like trainers/dp2_adagrad.py ran · our draft was wrong Apache-2.0 (permissive) · 06cd2aedfcb73bb6 · report fixture: signature struct:dict/0/? · returned dict · level recorded 2026-09-22
BaseTrainer in 16 papers trainers/dp2_adagrad.py unverified Apache-2.0 (permissive) · dca6afaa2e23a3be · report
Trainer in 24 papers trainers/dp2_adagrad.py unverified Apache-2.0 (permissive) · 458f345611469a88 · report
imdb_loader trainers/dp2_adagrad.py unverified Apache-2.0 (permissive) · f4a878bc9479f4bc · report
linear_model_fn trainers/dp2_adagrad.py unverified Apache-2.0 (permissive) · 29b863abc869ce7a · report
matfac_model_fn trainers/dp2_adagrad.py unverified Apache-2.0 (permissive) · 32b7bb707740780e · report
multiclass_classify trainers/dp2_adagrad.py unverified Apache-2.0 (permissive) · aeca8b544abd9191 · report
print_log trainers/dp2_adagrad.py unverified Apache-2.0 (permissive) · defe6270f77abbbd · report
sce_loss in 15 papers trainers/dp2_adagrad.py unverified Apache-2.0 (permissive) · 1d3f9fc33b06bb7f · report

Method as code Syntology

What Syntology's graph records as this paper's claimed method, lifted into code: each assembly names its bricks (functions and classes), each with its own role and its own verification level. An assembly is not a check that the code implements the method correctly or reproduces the paper's results. Levels are each brick's own verification level in Syntology's graph (level 0: no recorded run), listed brick by brick and never combined into one level; “ran” means the brick executed on a synthesized input, not that its output is correct, and the instrument figure counts failures of Syntology's instrument, not of the code. For a generated assembly, an independent peer is a second implementation generated separately from the same paper text; neither implementation read the authors' code. At level 2 the two agreed numerically on the fixture's inputs; level 3 is recorded when property tests derived from the paper also passed on both, and each level 3 card prints how many its report lists. Where a model read the code against the paper's claim, each assembly's label verdict is one model's judgement, kept in this page's JSON twin; it is not calibrated, so no card prints it. A generated assembly's label was set when its code was generated for the method, and no model judged it; its card prints it as "by construction" or "not judged by a model". Where an assembly's phrase has no constructed figure, a brick that ran is not among the harvested samples this build reads, and whether a brick's run only constructed an object is read from that brick's harvested sample, so the figure was not counted; it does not say there were none.

2 assemblies on this paper: 1 harvested, 1 generated.

generated Assembly for DP 2 (graph method: DP 2), generated from the paper text, not the repository's code. 1 brick; individually on a fixture: 1 ran (1 with no instrument failure: 0 honoured, 0 violated, 1 with no contract checked; 0 where Syntology's instrument failed) · 0 unverified. Ran end-to-end against an independent peer (level 2).

label matches by construction: the code was generated for this method (not judged by a model)

  1. dp2_clip_and_privatize_gradient entry · level 2 · ran · 7c381e951a8009bf01853c4cc5c49ed3d4a519b16bd49bc330e1335a8ca7fb68

harvested Assembly for DP 2 (graph method: DP 2). 31 bricks; individually on a fixture: 23 ran (of which 0 constructed an object rather than computing a result; 4 with no instrument failure: 4 honoured, 0 violated, 0 with no contract checked; 19 where Syntology's instrument failed) · 8 unverified. The assembly has not been run end-to-end.

  1. Trainer entry · level 0 · unverified · 458f345611469a88
  2. BaseTrainer helper · level 0 · unverified · dca6afaa2e23a3be
  3. global_l2_norm_sq helper · level 1 · ran · honoured contract · bbd8d1efcf5829de
  4. num_params helper · level 1 · ran · honoured contract · af4f5b33931d44c0
  5. privatize_grad helper · level 1 · ran · our draft was wrong · 591c9ea59777949b
  6. struct_add helper · level 1 · ran · our draft was wrong · cca3cf0b80500ec6
  7. struct_add_scalar helper · level 1 · ran · our draft was wrong · 34bf4ad4230cf55d
  8. struct_div helper · level 1 · ran · our draft was wrong · 7d340e00912389f7
  9. struct_div_scalar helper · level 1 · ran · our draft was wrong · e488dedf8e381c96
  10. struct_mul_scalar helper · level 1 · ran · our draft was wrong · 6050b70841d335c1
  11. struct_sq helper · level 1 · ran · our draft was wrong · 145eda8014bfd82a
  12. struct_sqrt helper · level 1 · ran · our draft was wrong · ebb3869ec1266c1e
  13. struct_sub helper · level 1 · ran · our draft was wrong · 0bf286afd514640c
  14. struct_zeros_like helper · level 1 · ran · our draft was wrong · 06cd2aedfcb73bb6
  15. bce_loss helper · level 1 · ran · honoured contract · b3407cf1906789ea
  16. imdb_loader helper · level 0 · unverified · f4a878bc9479f4bc
  17. linear_model_fn helper · level 0 · unverified · 29b863abc869ce7a
  18. logreg_classify helper · level 1 · ran · our draft was wrong · 26f3490ca9bbe47e
  19. matfac_model_fn helper · level 0 · unverified · 32b7bb707740780e
  20. movielens_loader helper · level 1 · ran · our draft was wrong · c3b75484cf237418
  21. mse_loss helper · level 1 · ran; the contract check could not run (RAISES) · f320dea1d5541ada
  22. multiclass_classify helper · level 0 · unverified · aeca8b544abd9191
  23. print_log helper · level 0 · unverified · defe6270f77abbbd
  24. sce_loss helper · level 0 · unverified · 1d3f9fc33b06bb7f
  25. stackoverflow_tag_loader helper · level 1 · ran · our draft was wrong · 19770acf0fa0aa44
  26. stackoverflow_tag_loader_cached helper · level 1 · ran · our draft was wrong · 919aea51fdf26b6d
  27. struct_flatten helper · level 1 · ran · honoured contract · 155ccee594f8679d
  28. global_l2_clip helper · level 1 · ran · our draft was wrong · 735e56344d7303e1
  29. _create_tf_loader helper · level 1 · ran; the contract check could not run (RAISES) · 5d8c2870379d3384
  30. _load_movielens_100k helper · level 1 · ran · our draft was wrong · 09495c898e5d8e94
  31. _config_tf_loader helper · level 1 · ran · our draft was wrong · 57022752f9995f59

Each brick ends with its code_sha256, Syntology's identity for that exact code; click one to copy the get_code call that returns it. For agents, on Syntology's MCP service: assembly(arxiv_id="2212.00309") returns these assemblies with each brick's role and wiring (how to connect).

Results from the paper archive 2025-07-28

No leaderboard rows for this paper in the archive.

Methods

SPEED

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