SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2005.01831 · ACL · 2020

Evaluating Explainable AI: Which Algorithmic Explanations Help Users Predict Model Behavior?

Peter Hase, Mohit Bansal

arXiv · PDF · Open in the Atlas

Code that ran

4 of 9 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; 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/2005.01831.json · badge: /api/ran/2005.01831.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
peterbhase/InterpretableNLP-ACL2020 canonical (classifier label)
classifier basis: census
4 ran · 5 no recorded run
FunctionStatusWhere it lives
make_graph Ran
ran on a synthesized input
peterbhase/InterpretableNLP-ACL2020/tabular/anchor/make_graphs_and_table.py
code served (permissive licence) · get_code("60da6efa17ee6dc2")
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 Figure. It ran in syntology-harvest-sandbox:v5-deps (sha256:6a292aa1aed1) at 2026-09-27T00:55:27Z. 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 matplotlib.pyplot as plt

# Create minimal pickles structure that matches what make_graph expects
pickles = {
    'dataset1': {
        'model1': {
            'lime_pred_submodular': [
                None,  # index 0
                None,  # index 1
                [0.5, 0.7, 0.85, 0.9]  # index 2 - lime_sub values
            ],
            'anchor_submodular': [
                None,  # index 0
                None,  # index 1
                [0.6, 0.75, 0.88, 0.92]  # index 2 - anchor_sub values
            ],
            'anchor_random': [
                None,
                [0.55, 0.72, 0.86, 0.91],  # means
                None,
                [0.05, 0.04, 0.03, 0.02]  # std errors
            ],
            'lime_pred_random': [
                None,
                [0.48, 0.68, 0.83, 0.89],  # means
                None,
                [0.06, 0.05, 0.04, 0.03]  # std errors
            ]
        }
    }
}

ARGS = {
    'pickles': pickles,
    'dataset': 'dataset1',
    'model': 'model1',
    'save': None
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
matplotlib, numpy, seaborn — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type Figure — a Figure output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a Figure output the recorded ndim 0 and dtype family float describe that coercion, not the output, and are not stated
Environment
image syntology-harvest-sandbox:v5-deps (sha256:6a292aa1aed1); 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
make_table Ran
ran on a synthesized input
peterbhase/InterpretableNLP-ACL2020/tabular/anchor/make_graphs_and_table.py
code served (permissive licence) · get_code("1db3f658ae6ec3e0")
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
import numpy as np

torch.manual_seed(0)

# Build the pickles dictionary structure
# pickles[dataset][model][explanation_key] = (mean, coverage, std, coverage_std)
pickles = {
    'dataset_1': {
        'model_1': {
            'anchor_1': (0.85, 0.72, 0.05, 0.08),
            'lime_naive_1': (0.78, 0.68, 0.06, 0.09),
            'lime_pred_1': (0.80, 0.70, 0.04, 0.07),
        },
        'model_2': {
            'anchor_1': (0.88, 0.75, 0.04, 0.06),
            'lime_naive_1': (0.82, 0.71, 0.05, 0.08),
            'lime_pred_1': (0.84, 0.73, 0.03, 0.06),
        },
    },
    'dataset_2': {
        'model_1': {
            'anchor_1': (0.81, 0.68, 0.06, 0.09),
            'lime_naive_1': (0.76, 0.65, 0.07, 0.10),
            'lime_pred_1': (0.79, 0.67, 0.05, 0.08),
        },
        'model_2': {
            'anchor_1': (0.86, 0.74, 0.04, 0.07),
            'lime_naive_1': (0.80, 0.70, 0.06, 0.09),
            'lime_pred_1': (0.83, 0.72, 0.04, 0.07),
        },
    },
}

datasets = ['dataset_1', 'dataset_2']
models = ['model_1', 'model_2']

ARGS = {
    'pickles': pickles,
    'datasets': datasets,
    'models': models,
}
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 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
matrix_subset Ran
ran on a synthesized input
peterbhase/InterpretableNLP-ACL2020/tabular/anchor/anchor_base.py
code served (permissive licence) · get_code("e1b1e0837af55402")
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; no output shape is stated. It ran in syntology-harvest-sandbox:v5-deps (sha256:6a292aa1aed1) at 2026-09-27T00:55:27Z. 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 a sample matrix with shape (10, 4)
# The function uses matrix.shape[0] to get number of rows
# and matrix[...] to index rows
matrix = np.random.randn(10, 4)

# Create n_samples parameter - smaller than matrix rows
n_samples = 5

ARGS = {
    "matrix": matrix,
    "n_samples": n_samples
}
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 — no output shape is stated: no later run recorded an output shape for this sample (observed_output and metamorphic_properties carry none; shape_contract and output_contract record ndim only)
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 syntology-harvest-sandbox:v5-deps (sha256:6a292aa1aed1); 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
random_until_k Ran
ran on a synthesized input
peterbhase/InterpretableNLP-ACL2020/tabular/anchor/process_results.py
code served (permissive licence) · get_code("3d17a26510e578cb")
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. It ran in syntology-harvest-sandbox:v5-deps (sha256:6a292aa1aed1) at 2026-09-27T00:55:27Z. 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 numpy as np

torch.manual_seed(0)

def mock_random_fn(i):
    """
    Mock random function that returns (prec, rec, prec_std, rec_std) for iteration i.
    """
    prec = np.random.rand()
    rec = np.random.rand()
    prec_std = np.random.rand() * 0.1
    rec_std = np.random.rand() * 0.1
    return (prec, rec, prec_std, rec_std)

ARGS = {
    "random_fn": mock_random_fn,
    "k": 3
}
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 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 syntology-harvest-sandbox:v5-deps (sha256:6a292aa1aed1); 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_balanced_batches Not yet run
unverified (harvested, no recorded run)
peterbhase/InterpretableNLP-ACL2020/tabular/gather-experiment-data.py
code served (permissive licence) · get_code("3fb4c9bcfe98d3f6")
How it ran

no invocation recorded.

get_balanced_batches Not yet run
unverified (harvested, no recorded run)
peterbhase/InterpretableNLP-ACL2020/text/gather-experiment-data.py
code served (permissive licence) · get_code("ae7b52b915114f8d")
How it ran

no invocation recorded.

load_pickles Not yet run
unverified (harvested, no recorded run)
peterbhase/InterpretableNLP-ACL2020/tabular/anchor/make_graphs_and_table.py
code served (permissive licence) · get_code("e8dfabee12cd3fcf")
How it ran

no invocation recorded.

pick_perturbation Not yet run
unverified (harvested, no recorded run)
peterbhase/InterpretableNLP-ACL2020/tabular/gather-experiment-data.py
code served (permissive licence) · get_code("583b76ee9dc07214")
How it ran

no invocation recorded.

pick_perturbation Not yet run
unverified (harvested, no recorded run)
peterbhase/InterpretableNLP-ACL2020/text/gather-experiment-data.py
code served (permissive licence) · get_code("65318aa5dc0ea291")
How it ran

no invocation recorded.

Also present in other papers' code

Of 9 samples lifted from this paper's linked repositories, 9 carry a function identity and 0 do not. Of the 9 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 9 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

Algorithmic approaches to interpreting machine learning models have proliferated in recent years. We carry out human subject tests that are the first of their kind to isolate the effect of algorithmic explanations on a key aspect of model interpretability, simulatability, while avoiding important confounding experimental factors. A model is simulatable when a person can predict its behavior on new inputs. Through two kinds of simulation tests involving text and tabular data, we evaluate five explanations methods: (1) LIME, (2) Anchor, (3) Decision Boundary, (4) a Prototype model, and (5) a Composite approach that combines explanations from each method. Clear evidence of method effectiveness is found in very few cases: LIME improves simulatability in tabular classification, and our Prototype method is effective in counterfactual simulation tests. We also collect subjective ratings of explanations, but we do not find that ratings are predictive of how helpful explanations are. Our results provide the first reliable and comprehensive estimates of how explanations influence simulatability across a variety of explanation methods and data domains. We show that (1) we need to be careful about the metrics we use to evaluate explanation methods, and (2) there is significant room for improvement in current methods. 1

For agents

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

get_harvested_code_for_paper("2005.01831")
get_code_for_paper("2005.01831")
have("2005.01831")

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

curl https://syntology.ai/api/ran/2005.01831.json

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

[![Syntology run record](https://syntology.ai/api/ran/2005.01831.svg)](https://syntology.ai/paper/2005.01831)

Connect an agent — have() is free.