SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2110.04744 · ICLR · 2022

Long Expressive Memory for Sequence Modeling

U Berkeley, Michael Mahoney, N Erichson, T Rusch, Siddhartha Mishra

arXiv · PDF · Open in the Atlas

Code that ran

5 of 10 samples (functions and classes) lifted from this paper's linked repositories executed on a synthesized input in Syntology's jail (1 of them constructed an object rather than computing a result; 0 honoured the contract we drafted, 0 violated it and 5 have no contract verdict; 0 failures of our instrument; 5 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/2110.04744.json · badge: /api/ran/2110.04744.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
deepmind/lamb canonical (classifier label)
classifier basis not recorded on the link
4 ran · 5 no recorded run
tk-rusch/lem community
Papers with Code archive
1 ran · 0 no recorded run
FunctionStatusWhere it lives
LEMCell Ran
ran on a synthesized input · constructed an object; no result was computed
tk-rusch/lem/src/Google12/network.py
pointer only (licence: NONE) · get_code("9cfb904b319e5586")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, tk-rusch/lem 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.

handle_config_defaults Ran
ran on a synthesized input
deepmind/lamb/lamb/lamb_flags.py
code served (permissive licence) · get_code("db14e2261c3a6e84")
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 Config. The image it ran in was 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 math
from copy import copy

class Config:
    def __init__(self):
        self.num_params = 10000
        self.hidden_size = [-1]
        self.num_layers = 3
        self.hidden_size_multiplier = 0.8
        self.input_embedding_size = -1
        self.output_embedding_size = -1
        self.input_embedding_ratio = 0.5
        self.output_embedding_ratio = 0.5
        self.sparsity_ratio = 0.1
        self.learning_rate_decay = 0.5
        self.feature_mask_rank = -1
        self.inter_layer_dropout = -1.0
        self.input_dropout = 0.1
        self.downprojected_output_dropout = -1.0
        self.output_dropout = 0.1
        self.feature_mask = False
        self.feature_mask_rounds = 0
        self.share_input_and_output_embeddings = False

def num_params_fn(config):
    """Mock function that estimates number of parameters based on config."""
    if config.hidden_size is None or len(config.hidden_size) == 0:
        return None
    
    total = 0
    hidden_size = config.hidden_size[0]
    
    # Input embedding parameters
    if hasattr(config, 'input_embedding_size') and config.input_embedding_size > 0:
        total += config.input_embedding_size * 1000
    
    # Hidden layer parameters
    for i in range(config.num_layers):
        if i < len(config.hidden_size):
            h = config.hidden_size[i]
        else:
            h = hidden_size
        total += h * h * 4
    
    # Output embedding parameters
    if hasattr(config, 'output_embedding_size') and config.output_embedding_size > 0:
        total += config.output_embedding_size * 1000
    
    return total

config = Config()

ARGS = {
    'config': config,
    'num_params_fn': num_params_fn,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
absl, six — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type Config — a Config output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a Config 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
maybe_truncate Ran
ran on a synthesized input
deepmind/lamb/lamb/corpus.py
code served (permissive licence) · get_code("41b09314d9718a2f")
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 list, ndim 1, dtype family int; no output shape is stated. The image it ran in was not recorded. If your copy returns the same type and ndim 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

torch.manual_seed(0)

# Test case 1: sequence that exceeds truncation limit
seq1 = [1, 2, 3, 4, 5]
n1 = 3

# Test case 2: sequence that doesn't exceed truncation limit
seq2 = [10, 20, 30]
n2 = 5

# Test case 3: n is None (no truncation)
seq3 = [100, 200, 300, 400]
n3 = None

# Use the first test case for ARGS
ARGS = {
    "seq": seq1,
    "n": n1
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type list, ndim 1, dtype family int — a list output has no array shape of its own; none is stated
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
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
option_visibility Ran
ran on a synthesized input
deepmind/lamb/lamb/lamb_flags.py
code served (permissive licence) · get_code("e9fa8fcbfffec562")
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 str. The image it ran in was 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

torch.manual_seed(0)

# The function checks if option has length 4, and if so returns option[3]
# Otherwise returns None. So option needs to be a sequence-like object.

# Test case 1: option with 4 elements
option = [1, 2, 3, "visibility_value"]

ARGS = {
    "option": option
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type str — a str output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a str 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
round_to_int Ran
ran on a synthesized input
deepmind/lamb/lamb/lamb_flags.py
code served (permissive licence) · get_code("fe22ebdbafcd2a30")
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 int, ndim 0, dtype family int, shape () as a later run recorded it. The image it ran in was not recorded. If your copy returns the same type, ndim and shape 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

torch.manual_seed(0)

# The function round_to_int takes a single parameter x and applies round() then int()
# x can be a number (int, float) or anything that round() accepts

ARGS = {
    "x": 3.7
}
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 int, ndim 0, dtype family int, shape () — shape () as a later run recorded it
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
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
evaluate_deterministic Not yet run
unverified (harvested, no recorded run)
deepmind/lamb/lamb/evaluation.py
code served (permissive licence) · get_code("2c67d2b0758d5fe0")
How it ran

no invocation recorded.

evaluate_mc_with_geometric_mean Not yet run
unverified (harvested, no recorded run)
deepmind/lamb/lamb/evaluation.py
code served (permissive licence) · get_code("c598ebd47e698d5c")
How it ran

no invocation recorded.

evaluate_mc_with_power_mean Not yet run
unverified (harvested, no recorded run)
deepmind/lamb/lamb/evaluation.py
code served (permissive licence) · get_code("184a036257da4448")
How it ran

no invocation recorded.

read_character_based_corpus Not yet run
unverified (harvested, no recorded run)
deepmind/lamb/lamb/corpus.py
code served (permissive licence) · get_code("e82bc4ce41514342")
How it ran

no invocation recorded.

read_word_based_corpus Not yet run
unverified (harvested, no recorded run)
deepmind/lamb/lamb/corpus.py
code served (permissive licence) · get_code("d52c0a249edc34f9")
How it ran

no invocation recorded.

Also present in other papers' code

Of 10 samples lifted from this paper's linked repositories, 10 carry a function identity and 0 do not. Of the 10 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 10 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

We propose a novel method called Long Expressive Memory (LEM) for learning long-term sequential dependencies. LEM is gradient-based, it can efficiently process sequential tasks with very long-term dependencies, and it is sufficiently expressive to be able to learn complicated input-output maps. To derive LEM, we consider a system of multiscale ordinary differential equations, as well as a suitable time-discretization of this system. For LEM, we derive rigorous bounds to show the mitigation of the exploding and vanishing gradients problem, a wellknown challenge for gradient-based recurrent sequential learning methods. We also prove that LEM can approximate a large class of dynamical systems to high accuracy. Our empirical results, ranging from image and time-series classification through dynamical systems prediction to keyword spotting and language modeling, demonstrate that LEM outperforms state-of-the-art recurrent neural networks, gated recurrent units, and long short-term memory models.

For agents

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

get_harvested_code_for_paper("2110.04744")
get_code_for_paper("2110.04744")
have("2110.04744")

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

curl https://syntology.ai/api/ran/2110.04744.json

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

[![Syntology run record](https://syntology.ai/api/ran/2110.04744.svg)](https://syntology.ai/paper/2110.04744)

Connect an agent — have() is free.