SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2006.06848 · ICLR · 2021

Getting a CLUE: A Method for Explaining Uncertainty Estimates

Javier Antorán, José Hernández-Lobato, Umang Bhatt, Tameem Adel, Adrian Weller

arXiv · PDF · Open in the Atlas

Code that ran

4 of 8 samples (functions and classes) lifted from this paper's linked repositories executed on a synthesized input in Syntology's jail (0 of them constructed an object rather than computing a result; 0 honoured the contract we drafted, 0 violated it and 4 have no contract verdict; 0 failures of our instrument; 4 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/2006.06848.json · badge: /api/ran/2006.06848.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
cambridge-mlg/CLUE canonical (classifier label)
classifier basis: census
4 ran · 4 no recorded run
FunctionStatusWhere it lives
X_dims_to_input_dim_vec Ran
ran on a synthesized input
cambridge-mlg/CLUE/src/compas_loader.py
code served (permissive licence) · get_code("c0a6c860eb0a4920")
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 ndarray, ndim 1, dtype family int, shape (2,) 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 numpy as np

# The function X_dims_to_input_dim_vec takes X_dims as input
# It iterates through X_dims, treating each element as both a dimension value
# and as an indicator of how many positions to skip forward
# It builds input_dim_vec by appending X_dims[i] and then jumping by that amount

# Example: if X_dims = [2, 1, 3, 1, 4]
# i=0: append X_dims[0]=2, i becomes 0+2=2
# i=2: append X_dims[2]=3, i becomes 2+3=5
# i=5: loop ends (5 >= len(X_dims))
# Result: [2, 3]

# Constructing a valid X_dims that will execute properly
# We need a list where values serve dual purpose: dimension size and skip amount
X_dims = [2, 1, 3, 1, 4]

ARGS = {
    "X_dims": X_dims
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type ndarray, ndim 1, dtype family int, shape (2,) — shape (2,) 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
input_dim_vec_to_X_dims Ran
ran on a synthesized input
cambridge-mlg/CLUE/src/LSAT_loader.py
code served (permissive licence) · get_code("50f57aef8a925df8")
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 ndarray, ndim 1, dtype family int, shape (10,) 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 numpy as np

def input_dim_vec_to_X_dims(input_dim_vec):
    """This is for our cat_Gauss VAE model"""
    X_dims = []
    for i in input_dim_vec:
        for ii in range(i):
            X_dims.append(i)
    return np.array(X_dims)

ARGS = {
    "input_dim_vec": np.array([2, 3, 1, 4])
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type ndarray, ndim 1, dtype family int, shape (10,) — shape (10,) 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
join_compas_targets Ran
ran on a synthesized input
cambridge-mlg/CLUE/src/compas_loader.py
code served (permissive licence) · get_code("9f964290543f2ad8")
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 tuple. 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 numpy as np
from sklearn import preprocessing

# Construct the arguments for join_compas_targets

# Create small sample data
# x_train: shape (2, 4) - 2 samples, 4 features
x_train = np.array([
    [0.1, 0.2, 0.3, 0.4],
    [0.5, 0.6, 0.7, 0.8]
], dtype=np.float32)

# x_test: shape (2, 4) - 2 samples, 4 features
x_test = np.array([
    [0.9, 1.0, 1.1, 1.2],
    [1.3, 1.4, 1.5, 1.6]
], dtype=np.float32)

# y_train: shape (2,) - 2 binary labels
y_train = np.array([0, 1], dtype=np.int32)

# y_test: shape (2,) - 2 binary labels
y_test = np.array([1, 0], dtype=np.int32)

# X_dims: describes the structure of features
# Format: [dim_size, dim_size, ...] where each element indicates dimension size
# For 4 features total, we can use X_dims = [2, 2] meaning 2 features + 2 features
X_dims = np.array([2, 2], dtype=np.int32)

ARGS = {
    'x_train': x_train,
    'x_test': x_test,
    'y_train': y_train,
    'y_test': y_test,
    'X_dims': X_dims
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, sklearn — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type tuple — a tuple output has no array shape of its own; none is 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
unnormalise_cat_vars Ran
ran on a synthesized input
cambridge-mlg/CLUE/src/UCI_loader.py
code served (permissive licence) · get_code("252202c54d21c615")
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 ndarray, ndim 2, dtype family float, shape (2, 4) 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 numpy as np

# Create small test data
np.random.seed(0)

# x: normalized data array, shape (batch_size, num_features)
# We'll use batch_size=2, num_features=4
x = np.random.randn(2, 4).astype(np.float32)

# x_means: mean values for each feature, shape (num_features,)
x_means = np.array([1.0, 2.0, 3.0, 4.0], dtype=np.float32)

# x_stds: standard deviation values for each feature, shape (num_features,)
x_stds = np.array([0.5, 1.0, 1.5, 2.0], dtype=np.float32)

# input_dim_vec: dimension vector indicating categorical variable sizes
# Values > 1 mean categorical with that many categories
# Value = 1 means continuous
input_dim_vec = [1, 3, 5, 2]

ARGS = {
    "x": x,
    "x_means": x_means,
    "x_stds": x_stds,
    "input_dim_vec": input_dim_vec
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type ndarray, ndim 2, dtype family float, shape (2, 4) — shape (2, 4) 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
check_data_file Not yet run
unverified (harvested, no recorded run)
cambridge-mlg/CLUE/src/LSAT_loader.py
code served (permissive licence) · get_code("d18f0dd4a69ae09c")
How it ran

no invocation recorded.

get_my_COMPAS Not yet run
unverified (harvested, no recorded run)
cambridge-mlg/CLUE/src/compas_loader.py
code served (permissive licence) · get_code("8872f711e7965d33")
How it ran

no invocation recorded.

get_my_LSAT Not yet run
unverified (harvested, no recorded run)
cambridge-mlg/CLUE/src/LSAT_loader.py
code served (permissive licence) · get_code("967e06b170f2f53b")
How it ran

no invocation recorded.

load_doodle_dset Not yet run
unverified (harvested, no recorded run)
cambridge-mlg/CLUE/src/doodle_loader.py
code served (permissive licence) · get_code("325087ca64c93dc9")
How it ran

no invocation recorded.

Also present in other papers' code

Of 8 samples lifted from this paper's linked repositories, 8 carry a function identity and 0 do not. Of the 8 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 8 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

Both uncertainty estimation and interpretability are important factors for trustworthy machine learning systems. However, there is little work at the intersection of these two areas. We address this gap by proposing a novel method for interpreting uncertainty estimates from differentiable probabilistic models, like Bayesian Neural Networks (BNNs). Our method, Counterfactual Latent Uncertainty Explanations (CLUE), indicates how to change an input, while keeping it on the data manifold, such that a BNN becomes more confident about the input's prediction. We validate CLUE through 1) a novel framework for evaluating counterfactual explanations of uncertainty, 2) a series of ablation experiments, and 3) a user study. Our experiments show that CLUE outperforms baselines and enables practitioners to better understand which input patterns are responsible for predictive uncertainty.

For agents

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

get_harvested_code_for_paper("2006.06848")
get_code_for_paper("2006.06848")
have("2006.06848")

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

curl https://syntology.ai/api/ran/2006.06848.json

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

[![Syntology run record](https://syntology.ai/api/ran/2006.06848.svg)](https://syntology.ai/paper/2006.06848)

Connect an agent — have() is free.