SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 1902.06977 · 2019

Evaluating model calibration in classification

arXiv · PDF · Open in the Atlas

Code that ran

6 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 6 have no contract verdict; 0 failures of our instrument; 2 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/1902.06977.json · badge: /api/ran/1902.06977.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
uu-sml/calibration canonical (classifier label)
classifier basis: census
6 ran · 2 no recorded run
FunctionStatusWhere it lives
consistent_targets Ran
ran on a synthesized input
uu-sml/calibration/calibration/sample/consistent.py
code served (permissive licence) · get_code("6c403355b95b1151")
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

# Set seed for reproducibility
np.random.seed(0)

# Create a valid probability matrix
# Shape: (N, C) where N is batch size and C is number of classes
# N=2 (batch size), C=4 (number of classes)
probs = np.array([
    [0.1, 0.2, 0.3, 0.4],  # batch sample 1
    [0.25, 0.25, 0.25, 0.25]  # batch sample 2
])

# Ensure probabilities sum to 1 along each row
probs = probs / probs.sum(axis=1, keepdims=True)

ARGS = {
    'probs': probs
}
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
group_lens Ran
ran on a synthesized input
uu-sml/calibration/calibration/lenses/group.py
code served (permissive licence) · get_code("2971e1e1803b3bb6")
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

# Create sample data
np.random.seed(0)

# Batch size N=2, number of targets C=4
N = 2
C = 4

# Create probabilities: shape (N, C) = (2, 4)
probs = np.random.rand(N, C).astype(np.float32)

# Create targets as 1D array: shape (N,)
y = np.array([0, 2], dtype=np.int32)

# Create groups: list of arrays representing non-overlapping groups of target indices
# Groups should cover all targets 0, 1, 2, 3
groups = [
    np.array([0, 1]),  # group 0 contains targets 0 and 1
    np.array([2]),     # group 1 contains target 2
    np.array([3]),     # group 2 contains target 3
]

ARGS = {
    "probs": probs,
    "y": y,
    "groups": groups,
    "check_groups": True,
}
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 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
logarithmic_score Ran
ran on a synthesized input
uu-sml/calibration/calibration/scores/logarithmic.py
code served (permissive licence) · get_code("d9cccedb054bf4d1")
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 float32, ndim 0, dtype family float, 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 numpy as np

np.random.seed(0)

# Create valid arguments for logarithmic_score
# probs: shape (N, C) - batch_size=2, num_classes=4
probs = np.array([
    [0.1, 0.2, 0.5, 0.2],
    [0.3, 0.3, 0.2, 0.2]
], dtype=np.float32)

# y: shape (N,) - target class indices for each sample in batch
y = np.array([2, 0], dtype=np.int64)

ARGS = {
    'probs': probs,
    'y': y
}
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 float32, ndim 0, dtype family float, 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
quadratic_score Ran
ran on a synthesized input
uu-sml/calibration/calibration/scores/quadratic.py
code served (permissive licence) · get_code("4f7a1cc8b67a6813")
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 float32, ndim 0, dtype family float, 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 numpy as np

# Set random seed for reproducibility
np.random.seed(0)

# Create small test data
batch_size = 2
num_classes = 4

# probs: shape (N, C) where N=batch_size, C=num_classes
# Should be valid probabilities (sum to 1 along class dimension)
probs = np.array([
    [0.1, 0.2, 0.5, 0.2],
    [0.3, 0.1, 0.1, 0.5]
], dtype=np.float32)

# y: shape (N,) with class indices
y = np.array([2, 3], dtype=np.int32)

ARGS = {
    'probs': probs,
    'y': y
}
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 float32, ndim 0, dtype family float, 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
spherical_score Ran
ran on a synthesized input
uu-sml/calibration/calibration/scores/spherical.py
code served (permissive licence) · get_code("7774e83d22de764f")
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 float32, ndim 0, dtype family float, 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 numpy as np

# Set seed for reproducibility
np.random.seed(0)

# Create small test data
batch_size = 2
num_classes = 4

# probs: shape (N, C) - batch_size x num_classes
# These should be probability-like values (positive, typically summing to 1)
probs = np.array([
    [0.1, 0.2, 0.5, 0.2],
    [0.3, 0.3, 0.2, 0.2]
], dtype=np.float32)

# y: shape (N,) - target class indices for each sample in batch
# Values should be valid indices into range [0, num_classes)
y = np.array([2, 0], dtype=np.int64)

ARGS = {
    "probs": probs,
    "y": y
}
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 float32, ndim 0, dtype family float, 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
top2_lens Ran
ran on a synthesized input
uu-sml/calibration/calibration/lenses/top2.py
code served (permissive licence) · get_code("01608ff38bf20a2e")
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

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

# probs: shape (N, C) - probabilities for N samples and C classes
# Let N=2 (batch size), C=4 (number of classes)
N = 2
C = 4
probs = np.random.dirichlet(np.ones(C), size=N).astype(np.float32)

# y: can be either shape (N,) with class indices or shape (N, C) with one-hot encoding
# Test with shape (N,) - class indices
y = np.array([1, 2], dtype=np.int64)

ARGS = {
    "probs": probs,
    "y": y,
}
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 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
maximum_lens Not yet run
unverified (harvested, no recorded run)
uu-sml/calibration/calibration/lenses/max.py
code served (permissive licence) · get_code("f052a2430375201f")
How it ran

no invocation recorded.

resample_stats Not yet run
unverified (harvested, no recorded run)
uu-sml/calibration/calibration/sample/resample.py
code served (permissive licence) · get_code("fc42a6bf6e5ed961")
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

Probabilistic classifiers output a probability distribution on target classes rather than just a class prediction. Besides providing a clear separation of prediction and decision making, the main advantage of probabilistic models is their ability to represent uncertainty about predictions. In safety-critical applications, it is pivotal for a model to possess an adequate sense of uncertainty, which for probabilistic classifiers translates into outputting probability distributions that are consistent with the empirical frequencies observed from realized outcomes. A classifier with such a property is called calibrated. In this work, we develop a general theoretical calibration evaluation framework grounded in probability theory, and point out subtleties present in model calibration evaluation that lead to refined interpretations of existing evaluation techniques. Lastly, we propose new ways to quantify and visualize miscalibration in probabilistic classification, including novel multidimensional reliability diagrams.

For agents

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

get_harvested_code_for_paper("1902.06977")
get_code_for_paper("1902.06977")
have("1902.06977")

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

curl https://syntology.ai/api/ran/1902.06977.json

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

[![Syntology run record](https://syntology.ai/api/ran/1902.06977.svg)](https://syntology.ai/paper/1902.06977)

Connect an agent — have() is free.