SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2401.09587 · ICLR · 2024

Bilevel Optimization under Unbounded Smoothness: A New Algorithm and Convergence Analysis

Jie Hao, Mingrui Liu, Xiaochuan Gong

arXiv · PDF · Open in the Atlas

Code that ran

3 of 6 samples (functions and classes) lifted from this paper's linked repositories executed on a synthesized input in Syntology's jail (3 of them constructed an object rather than computing a result; 0 honoured the contract we drafted, 0 violated it and 3 have no contract verdict; 0 failures of our instrument; 3 no recorded run), measured 2026-11-22; repository links from the Papers with Code archive attributed under CC BY-SA, no affiliation; 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/2401.09587.json · badge: /api/ran/2401.09587.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
mingruiliu-ml-lab/bilevel-optimization-under-unbounded-smoothness canonical (classifier label)
classifier basis: classifier_2025
3 ran · 3 no recorded run
FunctionStatusWhere it lives
NLIRNN Ran
ran on a synthesized input · constructed an object; no result was computed
mingruiliu-ml-lab/bilevel-optimization-under-unbounded-smoothness/meta_learning/bo_rep.py
code served (permissive licence) · get_code("2d6a5a813b57b029")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned an instance of its own class, NLIRNN: the run constructed the object and did not call it, so no result was computed. The image it ran in was not recorded. The code mentions cuda, and the harness aliases cuda to the CPU for every sample; whether this run needed that is not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
import torch.nn as nn
import numpy as np

torch.manual_seed(0)

# Create minimal test data
batch_size = 2
seq_len = 5
word_embed_dim = 4
encoder_dim = 3
n_classes = 3

# Create sentence tuples (sent, sent_len)
# sent shape: (seq_len, batch_size, word_embed_dim)
sent_1 = torch.randn(seq_len, batch_size, word_embed_dim)
sent_len_1 = np.array([seq_len, seq_len - 1])
s_1 = (sent_1, sent_len_1)

sent_2 = torch.randn(seq_len, batch_size, word_embed_dim)
sent_len_2 = np.array([seq_len, seq_len - 1])
s_2 = (sent_2, sent_len_2)

ARGS = {
    'word_embed_dim': 4,
    'encoder_dim': 3,
    'n_enc_layers': 1,
    'dpout_model': 0.0,
    'dpout_fc': 0.0,
    'fc_dim': 8,
    'n_classes': 3,
    'pool_type': 'mean',
    'linear_fc': False,
    'bidirectional': False,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type NLIRNN — the run constructed an instance of its own class and did not call it; no shape is stated — the instrument coerces any output to an array; for a NLIRNN output the recorded ndim 0 and dtype family float describe that coercion, not the output, and are not stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
RNN Ran
ran on a synthesized input · constructed an object; no result was computed
mingruiliu-ml-lab/bilevel-optimization-under-unbounded-smoothness/meta_learning/bo_rep.py
code served (permissive licence) · get_code("9e863452892d1052")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned an instance of its own class, RNN: the run constructed the object and did not call it, so no result was computed. The image it ran in was not recorded. The code mentions cuda, and the harness aliases cuda to the CPU for every sample; whether this run needed that is not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
import numpy as np
from torch import nn

torch.manual_seed(0)

ARGS = {
    'word_embed_dim': 8,
    'encoder_dim': 16,
    'n_enc_layers': 2,
    'dpout_model': 0.1,
    'dpout_fc': 0.2,
    'fc_dim': 32,
    'n_classes': 5,
    'pool_type': 'mean',
    'linear_fc': True,
    'bidirectional': False,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type RNN — the run constructed an instance of its own class and did not call it; no shape is stated — the instrument coerces any output to an array; for a RNN output the recorded ndim 0 and dtype family float describe that coercion, not the output, and are not stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
RecurrentEncoder Ran
ran on a synthesized input · constructed an object; no result was computed
mingruiliu-ml-lab/bilevel-optimization-under-unbounded-smoothness/meta_learning/bo_rep.py
code served (permissive licence) · get_code("0ba6798dc81f9d8e")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned an instance of its own class, RecurrentEncoder: the run constructed the object and did not call it, so no result was computed. The image it ran in was not recorded. The code mentions cuda, and the harness aliases cuda to the CPU for every sample; whether this run needed that is not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
import numpy as np
from torch import nn

torch.manual_seed(0)

# Create a minimal RecurrentEncoder to test
ARGS = {
    "n_enc_layers": 2,
    "word_embed_dim": 4,
    "encoder_dim": 8,
    "pool_type": "mean",
    "dpout_model": 0.1,
    "bidirectional": True,
    "rnn": True
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type RecurrentEncoder — the run constructed an instance of its own class and did not call it; no shape is stated — the instrument coerces any output to an array; for a RecurrentEncoder output the recorded ndim 0 and dtype family float describe that coercion, not the output, and are not stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
Learner Not yet run
unverified (harvested, no recorded run)
mingruiliu-ml-lab/bilevel-optimization-under-unbounded-smoothness/meta_learning/bo_rep.py
code served (permissive licence) · get_code("4c731af150f1c43c")
How it ran

no invocation recorded.

hypergradient Not yet run
unverified (harvested, no recorded run)
mingruiliu-ml-lab/bilevel-optimization-under-unbounded-smoothness/meta_learning/bo_rep.py
code served (permissive licence) · get_code("381ebfbecfd506cd")
How it ran

no invocation recorded.

predict Not yet run
unverified (harvested, no recorded run)
mingruiliu-ml-lab/bilevel-optimization-under-unbounded-smoothness/meta_learning/bo_rep.py
code served (permissive licence) · get_code("8640c30e03b9b8b4")
How it ran

no invocation recorded.

Also present in other papers' code

Of 6 samples lifted from this paper's linked repositories, 6 carry a function identity and 0 do not. Of the 6 distinct functions with an identity, 2 are also present in other papers whose code we hold (0 with at least one paper dated earlier, 0 the same month, 2 later) and 4 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.

FunctionAlso present inEarliest-datedOthers named
RecurrentEncoder
brick of this paper's assembly: helper
1 other paper
1 later · 1 only as an assembly brick
2409.19212 (2024-09)
An Accelerated Algorithm for Stochastic Bilevel Optimization under Unbounded Smoothness
hypergradient
brick of this paper's assembly: update_rule
1 other paper
1 later
2412.20017 (2024-12)
A Nearly Optimal Single Loop Algorithm for Stochastic Bilevel Optimization under Unbounded Smoothness

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

Bilevel optimization is an important formulation for many machine learning problems, such as meta-learning and hyperparameter optimization. Current bilevel optimization algorithms assume that the gradient of the upper-level function is Lipschitz (i.e., the upper-level function has a bounded smoothness parameter). However, recent studies reveal that certain neural networks such as recurrent neural networks (RNNs) and long-short-term memory networks (LSTMs) exhibit potential unbounded smoothness, rendering conventional bilevel optimization algorithms unsuitable for these neural networks. In this paper, we design a new bilevel optimization algorithm, namely BO-REP, to address this challenge. This algorithm updates the upper-level variable using normalized momentum and incorporates two novel techniques for updating the lower-level variable: initialization refinement and periodic updates. Specifically, once the upper-level variable is initialized, a subroutine is invoked to obtain a refined estimate of the corresponding optimal lower-level variable, and the lower-level variable is updated only after every specific period instead of each iteration. When the upper-level problem is nonconvex and unbounded smooth, and the lower-level problem is strongly convex, we prove that our algorithm requires O(1/ϵ 4 ) 1 iterations to find an ϵstationary point in the stochastic setting, where each iteration involves calling a stochastic gradient or Hessian-vector product oracle. Notably, this result matches the state-of-the-art complexity results under the bounded smoothness setting and without mean-squared smoothness of the stochastic gradient, up to logarithmic factors. Our proof relies on novel technical lemmas for the periodically updated lower-level variable, which are of independent interest. Our experiments on hyperrepresentation learning, hyperparameter optimization, and data hyper-cleaning for text classification tasks demonstrate the effectiveness of our proposed algorithm.

For agents

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

get_harvested_code_for_paper("2401.09587")
get_code_for_paper("2401.09587")
have("2401.09587")

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

curl https://syntology.ai/api/ran/2401.09587.json

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

[![Syntology run record](https://syntology.ai/api/ran/2401.09587.svg)](https://syntology.ai/paper/2401.09587)

Connect an agent — have() is free.