SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2106.00225 · NeurIPS · 2021

Locally Valid and Discriminative Prediction Intervals for Deep Learning Models

Jimeng Sun, Zhen Lin, Shubhendu Trivedi

arXiv · PDF · Open in the Atlas

Code that ran

12 of 16 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 9 have no contract verdict; 3 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/2106.00225.json · badge: /api/ran/2106.00225.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
zlin7/LVD canonical (classifier label)
classifier basis: census
8 ran · 1 no recorded run
zlin7/lvd canonical (classifier label)
classifier basis: classifier_2025
4 ran · 3 no recorded run
FunctionStatusWhere it lives
CQR_from_results Ran
ran on a synthesized input
zlin7/LVD/models/baselines/CQR.py
code served (permissive licence) · get_code("7e2b26ed7dad164c")
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, ndim 2, dtype family float; 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 numpy as np

# Create small numpy arrays for the function parameters
# The function expects 1D arrays of predictions and bounds

np.random.seed(0)

# Validation set: actual values and predicted bounds
val_y = np.array([1.0, 2.0, 3.0, 4.0, 5.0])
val_ylo = np.array([0.5, 1.5, 2.5, 3.5, 4.5])
val_yhi = np.array([1.5, 2.5, 3.5, 4.5, 5.5])

# Test set: predicted bounds
test_ylo = np.array([0.3, 1.2, 2.8, 3.9, 4.1])
test_yhi = np.array([1.7, 2.8, 3.2, 4.1, 5.9])

# Alpha parameter for confidence level
alpha = 0.1

ARGS = {
    "val_y": val_y,
    "val_ylo": val_ylo,
    "val_yhi": val_yhi,
    "test_ylo": test_ylo,
    "test_yhi": test_yhi,
    "alpha": alpha,
}
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, ndim 2, dtype family float — a tuple 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
MADSplit_from_results Ran
ran on a synthesized input
zlin7/LVD/models/baselines/MADSplit.py
code served (permissive licence) · get_code("d3eb64fc08ffefb5")
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, ndim 2, dtype family float; 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 numpy as np

# Create small numpy arrays for the function parameters
np.random.seed(0)

# valid_y: ground truth validation labels
valid_y = np.array([1.5, 2.3, 0.8, 1.2])

# valid_yhat: predicted validation labels
valid_yhat = np.array([1.6, 2.1, 0.9, 1.1])

# valid_rhat: prediction intervals (or residuals) on validation set
valid_rhat = np.array([0.3, 0.4, 0.2, 0.25])

# test_yhat: predicted test labels
test_yhat = np.array([2.5, 3.1, 1.5])

# test_rhat: prediction intervals (or residuals) on test set
test_rhat = np.array([0.35, 0.45, 0.28])

# alpha: significance level (default 0.1 for 90% confidence)
alpha = 0.1

# eps: small constant to avoid division by zero
eps = 1e-05

ARGS = {
    "valid_y": valid_y,
    "valid_yhat": valid_yhat,
    "valid_rhat": valid_rhat,
    "test_yhat": test_yhat,
    "test_rhat": test_rhat,
    "alpha": alpha,
    "eps": eps,
}
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, ndim 2, dtype family float — a tuple 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
NaiveKernel Ran
ran on a synthesized input
zlin7/lvd/models/conformal.py
code served (permissive licence) · get_code("d6e7f479334eeec3")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, class construct-then-call prompt, model not recorded) and returned tuple. 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 types as _types
import numpy as np

def gpuid_to_device(gpuid, mod=True):
    if isinstance(gpuid, str):
        if gpuid.startswith('cpu') or gpuid.startswith('cuda'):
            return gpuid
        raise ValueError
    if gpuid == -1:
        return 'cpu'
    if gpuid is None:
        return 'cuda'
    if mod:
        import torch
        gpuid = gpuid % torch.cuda.device_count()
    return 'cuda:%d' % gpuid

utils = _types.SimpleNamespace(**{'gpuid_to_device': gpuid_to_device})

class NaiveKernel:

    def __init__(self, type='Gaussian'):
        self._device = utils.gpuid_to_device(-1)
        self.type = type

    def to(self, device):
        self._device = device

    def K(self, x1, x2=None):
        if self.type == 'Gaussian':
            if x2 is None:
                return 1.0
            with torch.no_grad():
                diff = torch.tensor(x1 - x2, device=self._device, dtype=torch.float)
                kij = torch.exp(-torch.pow(diff, 2).sum(-1))
                return float(kij.detach())
        raise NotImplementedError()

    def Ki(self, xi, Xs, speedup_info=None):
        if self.type == 'Gaussian':
            with torch.no_grad():
                if speedup_info is None:
                    speedup_info = torch.tensor(Xs, device=self._device, dtype=torch.float, requires_grad=False)
                Xs = speedup_info
                if Xs.device != self._device:
                    Xs = Xs.to(self._device)
                xi = torch.tensor(xi, device=self._device, dtype=torch.float)
                diff_i = xi - Xs
                kij = torch.exp(-torch.pow(diff_i, 2).sum(-1))
                return (kij, Xs)


INIT_ARGS = {
    'type': 'Gaussian'
}

ARGS = {
    'xi': np.array([1.0, 2.0, 3.0, 4.0]),
    'Xs': np.array([[1.1, 2.1, 3.1, 4.1], [0.9, 1.9, 2.9, 3.9]]),
    'speedup_info': None
}

METHOD = 'Ki'

if __name__ == '__main__':
    kernel = NaiveKernel(**INIT_ARGS)
    result = getattr(kernel, METHOD)(**ARGS)
    print(f"Result: {result}")
Setup block drafted by
a language model, class construct-then-call prompt, model not recorded
Imports
numpy, torch — 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: torch and numpy seeded to 0 by the harness immediately before the class was constructed from INIT_ARGS (fixture_check.py)
build_architecture Ran
ran on a synthesized input
zlin7/LVD/models/baselines/DNN.py
code served (permissive licence) · get_code("9199fc21b3ba4e69")
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 Sequential. 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
import torch.nn as nn

torch.manual_seed(0)

class BaseModel:
    def __init__(self):
        self.dropout_active = True
        self.dropout_prob = 0.5
        self.n_dim = 4
        self.num_hidden = 8
        self.activation = 'ReLU'
        self.num_layers = 3
        self.output_size = 2

base_model = BaseModel()

ARGS = {
    'base_model': base_model
}
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 Sequential — a Sequential output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a Sequential output the recorded ndim 1 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
clean_df Ran
ran on a synthesized input
zlin7/LVD/utils/eval_utils.py
code served (permissive licence) · get_code("38a69e35d2a391c7")
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 DataFrame. 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
import pandas as pd
import matplotlib
matplotlib.use('Agg')  # Non-interactive backend
import matplotlib.pyplot as plt

# Create sample data for the dataframe
np.random.seed(0)
n_samples = 50

# Generate synthetic data with required columns
x_vals = np.linspace(0, 10, n_samples)
y_vals = np.sin(x_vals) + np.random.normal(0, 0.2, n_samples)
yhat_vals = np.sin(x_vals)
lo_vals = yhat_vals - 0.5
hi_vals = yhat_vals + 0.5

df = pd.DataFrame({
    'x': x_vals,
    'y': y_vals,
    'yhat': yhat_vals,
    'lo': lo_vals,
    'hi': hi_vals
})

ARGS = {
    'df': df,
    'title': 'Test Plot',
    'fillinf': np.inf,
    'alpha': 0.05,
    'plot': True,
    'smooth': None
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
matplotlib, numpy, pandas, scipy — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type DataFrame — a DataFrame output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a DataFrame output the recorded ndim 2 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
get_DNN_and_trainkwargs Ran
ran on a synthesized input
zlin7/LVD/models/baselines/DNN.py
code served (permissive licence) · get_code("64ce108b3e36bd41")
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, ndim 1, dtype family float; 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
import numpy as np

torch.manual_seed(0)
np.random.seed(0)

# Create small dummy data for testing
X_train = np.random.randn(2, 4).astype(np.float32)
y_train = np.random.randn(2, 1).astype(np.float32)

ARGS = {
    'setting_id': 0,
}
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 tuple, ndim 1, dtype family float — a tuple 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
get_default_fitkwargs Ran
ran on a synthesized input
zlin7/LVD/demos/experiments.py
code served (permissive licence) · get_code("f9bc950f0b053082")
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 dict. 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
import numpy as np

torch.manual_seed(0)

class SimpleDataset:
    def __init__(self):
        self.data = [(torch.randn(2, 4), torch.randn(2, 1)) for _ in range(5)]
    
    def __len__(self):
        return len(self.data)
    
    def __getitem__(self, idx):
        return self.data[idx]

dataset = SimpleDataset()

ARGS = {
    'dataset': dataset
}
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 dict — a dict output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a dict 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
gpuid_to_device Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
zlin7/lvd/models/conformal.py
code served (permissive licence) · get_code("93ae6a4d87ad6dc2")
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) and returned str. 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

torch.manual_seed(0)

ARGS = {
    'gpuid': 0,
    'mod': True,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
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
merge_dict_inline Ran
ran; our contract draft was wrong, not the code
zlin7/lvd/models/conformal.py
code served (permissive licence) · get_code("dad8377818739ce0")
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 dict. 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
import torch.nn as nn

torch.manual_seed(0)

# Construct arguments for merge_dict_inline(d1, d2)
# The function takes two dicts, copies d1, updates it with d2, and returns the result

d1 = {
    "key1": "value1",
    "key2": 42,
    "key3": [1, 2, 3],
}

d2 = {
    "key2": 100,
    "key4": "new_value",
    "key5": {"nested": "dict"},
}

ARGS = {
    "d1": d1,
    "d2": d2,
}
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 dict — a dict output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a dict 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
smoothing Ran
ran on a synthesized input
zlin7/LVD/utils/eval_utils.py
code served (permissive licence) · get_code("a08c0e709979c15a")
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, ndim 2, dtype family float; 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 numpy as np
from scipy.interpolate import UnivariateSpline

# Create simple x and y data for the smoothing function
x_data = [1.0, 2.0, 3.0, 4.0, 5.0]
y_data = [1.1, 2.1, 2.9, 4.2, 4.9]

ARGS = {
    'x': x_data,
    'y': y_data,
    'smooth': None
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, scipy — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type tuple, ndim 2, dtype family float — a tuple 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
summ_df Ran
ran on a synthesized input
zlin7/LVD/utils/eval_utils.py
code served (permissive licence) · get_code("e56f011f0044c5d0")
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 Series. 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
import pandas as pd
from sklearn.metrics import roc_auc_score, r2_score

# Create a sample DataFrame with the required columns
np.random.seed(0)
n_samples = 20

df = pd.DataFrame({
    'y': np.random.randn(n_samples),
    'yhat': np.random.randn(n_samples),
    'lo': np.random.randn(n_samples) - 1,
    'hi': np.random.randn(n_samples) + 1,
})

# Ensure lo < hi for valid intervals
df['lo'] = df[['lo', 'hi']].min(axis=1)
df['hi'] = df[['lo', 'hi']].max(axis=1) + 0.5

# Create non_inf_idx - indices where width is finite (all in this case)
non_inf_idx = df.index[df['hi'] - df['lo'] < np.inf]

# Define measures to test a subset
measures = ['AUROC', 'resid', 'cover', 'Corr(noninf)', 'cover(noninf)', 
            'cover(tail)', 'MSE', 'R2', 'mean_width(noninf)', 'cnt_width(noninf)']

ARGS = {
    'df': df,
    'alpha': 0.05,
    'resid_bound': None,
    'fill_inf_for_corr': False,
    'non_inf_idx': non_inf_idx,
    'measures': measures,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, pandas, sklearn — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type Series — a Series output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a Series output the recorded ndim 1 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
weighted_quantile_faster2 Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
zlin7/lvd/models/conformal.py
code served (permissive licence) · get_code("9c7355cbbe4b3419")
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) and returned float64, 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
import bisect

np.random.seed(0)

# Create a small array of values
arr = np.array([1.5, 2.3, 3.1, 4.7, 5.2])

# Create weights that sum to a reasonable value for quantile computation
weights = np.array([0.1, 0.2, 0.3, 0.25, 0.15])

# Create a cumulative sum target for bisect_left to search
# q should be a value that bisect_left can find in the cumsum range
qs_test = np.cumsum(weights)  # [0.1, 0.3, 0.6, 0.85, 1.0]
q = 0.5  # A value within the cumsum range

ARGS = {
    "arr": arr,
    "weights": weights,
    "q": q
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
Imports
numpy — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type float64, 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
LocalConditional Not yet run
unverified (harvested, no recorded run)
zlin7/lvd/models/conformal.py
code served (permissive licence) · get_code("a347cfbd81f921df")
How it ran

no invocation recorded.

PIConstructor Not yet run
unverified (harvested, no recorded run)
zlin7/lvd/models/conformal.py
code served (permissive licence) · get_code("077638700e3448f0")
How it ran

no invocation recorded.

my_split Not yet run
unverified (harvested, no recorded run)
zlin7/LVD/demos/experiments.py
code served (permissive licence) · get_code("d33ff76452c269e0")
How it ran

no invocation recorded.

weighted_quantile_faster Not yet run
unverified (harvested, no recorded run)
zlin7/lvd/models/conformal.py
code served (permissive licence) · get_code("471ab010d45f085c")
How it ran

no invocation recorded. this sample carries JSON test cases instead of a setup block; get_reference_implementation serves them with the level they back.

Also present in other papers' code

Of 16 samples lifted from this paper's linked repositories, 16 carry a function identity and 0 do not. Of the 16 distinct functions with an identity, 1 are also present in other papers whose code we hold (0 with at least one paper dated earlier, 0 the same month, 1 later) and 15 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
merge_dict_inline
brick of this paper's assembly: helper
1 other paper
1 later
2205.12940 (2022-05)
Conformal Prediction Intervals with Temporal Dependence

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

Crucial for building trust in deep learning models for critical real-world applications is efficient and theoretically sound uncertainty quantification, a task that continues to be challenging. Useful uncertainty information is expected to have two key properties: It should be valid (guaranteeing coverage) and discriminative (more uncertain when the expected risk is high). Moreover, when combined with deep learning (DL) methods, it should be scalable and affect the DL model performance minimally. Most existing Bayesian methods lack frequentist coverage guarantees and usually affect model performance. The few available frequentist methods are rarely discriminative and/or violate coverage guarantees due to unrealistic assumptions. Moreover, many methods are expensive or require substantial modifications to the base neural network. Building upon recent advances in conformal prediction [13,33] and leveraging the classical idea of kernel regression, we propose Locally Valid and Discriminative prediction intervals (LVD), a simple, efficient and lightweight method to construct discriminative prediction intervals (PIs) for almost any DL model. With no assumptions on the data distribution, such PIs also offer finite-sample local coverage guarantees (contrasted to the simpler marginal coverage). We empirically verify, using diverse datasets, that besides being the only locally valid method for DL, LVD also exceeds or matches the performance (including coverage rate and prediction accuracy) of existing uncertainty quantification methods, while offering additional benefits in scalability and flexibility.

For agents

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

get_harvested_code_for_paper("2106.00225")
get_code_for_paper("2106.00225")
have("2106.00225")

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

curl https://syntology.ai/api/ran/2106.00225.json

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

[![Syntology run record](https://syntology.ai/api/ran/2106.00225.svg)](https://syntology.ai/paper/2106.00225)

Connect an agent — have() is free.