SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2305.14740 · 2023

ECHo: A Visio-Linguistic Dataset for Event Causality Inference via Human-Centric Reasoning

arXiv · PDF · Open in the Atlas

Code that ran

6 of 6 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; 1 honoured the contract we drafted, 0 violated it and 0 have no contract verdict; 5 failures of our instrument; 0 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/2305.14740.json · badge: /api/ran/2305.14740.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
yuxixie/echo canonical (classifier label)
classifier basis: classifier_2025
6 ran · 0 no recorded run
FunctionStatusWhere it lives
get_data Ran
ran; our contract draft was wrong, not the code
yuxixie/echo/ToM_CoT_reasoning/prompt_generation.py
code served (permissive licence) · get_code("7eab34312111c82b")
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 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

torch.manual_seed(0)

raw_data = [
    {
        'ID': 'scene_001',
        'character': 'Detective Smith',
        'script': 'A detective enters a dark warehouse looking for clues about the missing artifact.',
        'role': 'detective'
    },
    {
        'ID': 'scene_001',
        'character': 'Detective Smith',
        'script': 'A detective enters a dark warehouse looking for clues about the missing artifact.',
        'role': 'detective'
    },
    {
        'ID': 'scene_002',
        'character': 'Suspect Jones',
        'script': 'The suspect nervously fidgets while being interrogated about their whereabouts.',
        'role': 'suspect'
    },
    {
        'ID': 'scene_003',
        'character': 'Officer Brown',
        'script': 'An officer arrives at the crime scene and begins documenting evidence.',
        'role': 'officer'
    }
]

ARGS = {
    'raw_data': raw_data
}
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 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_data Ran
ran; our contract draft was wrong, not the code
yuxixie/echo/ToM_CoT_reasoning/prompt_inference.py
code served (permissive licence) · get_code("f59d1ee55f418d0e")
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 list, 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 numpy as np
from tqdm import tqdm

torch.manual_seed(0)

# Construct raw_data: list of dicts with required structure
raw_data = [
    {
        'ID': 'id_001',
        'aid': 'aid_001',
        'character': 'Alice',
        'script': 'A detective investigates a crime scene.',
        'frames': [
            {'fid': 1, 'checked': True},
            {'fid': 2, 'checked': True},
            {'fid': 3, 'checked': False},
        ],
        'identity': {
            'role': 'detective',
            'age': '40s',
            'emotions': ['suspicious', 'determined'],
        },
        'event1': ['Alice', 'finds', 'evidence', 'at', 'the', 'scene'],
        'event2': ['Alice', 'arrests', 'the', 'suspect'],
    },
    {
        'ID': 'id_002',
        'aid': 'aid_002',
        'character': 'Bob',
        'script': 'A suspect is interrogated.',
        'frames': [
            {'fid': 4, 'checked': True},
            {'fid': 5, 'checked': True},
        ],
        'identity': {
            'role': 'suspect',
            'age': '30s',
            'emotions': ['nervous', 'defensive'],
        },
        'event1': ['Bob', 'denies', 'involvement'],
        'event2': ['Bob', 'confesses'],
    },
]

# Construct ToM_dict with nested structure
# Keys are tuples of (ID, aid, character, fid_string)
ToM_dict = {
    'role': {
        ('id_001', 'aid_001', 'Alice', '1-2'): 'detective investigating',
        ('id_002', 'aid_002', 'Bob', '4-5'): 'suspect under interrogation',
    },
    'emotion': {
        ('id_001', 'aid_001', 'Alice', '1-2'): 'focused and alert',
        ('id_002', 'aid_002', 'Bob', '4-5'): 'anxious and tense',
    },
}

# Construct txt_descriptions with nested structure
txt_descriptions = {
    'role': {
        ('id_001', 'aid_001', 'Alice', '1-2'): {
            'appearance': ['professional attire', 'badge visible'],
            'demeanor': ['authoritative', 'observant'],
        },
        ('id_002', 'aid_002', 'Bob', '4-5'): {
            'appearance': ['disheveled', 'tired'],
            'demeanor': ['evasive', 'uncomfortable'],
        },
    },
    'emotion': {
        ('id_001', 'aid_001', 'Alice', '1-2'): {
            'appearance': ['focused eyes', 'tense jaw'],
            'demeanor': ['concentrated', 'determined'],
        },
        ('id_002', 'aid_002', 'Bob', '4-5'): {
            'appearance': ['sweating', 'avoiding eye contact'],
            'demeanor': ['nervous', 'guilty'],
        },
    },
}

ARGS = {
    'raw_data': raw_data,
    'ToM_dict': ToM_dict,
    'txt_descriptions': txt_descriptions,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
Imports
numpy, torch, tqdm — 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 float — 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
get_internal Ran
ran, honoured the contract we drafted
yuxixie/echo/ToM_CoT_reasoning/prompt_generation.py
code served (permissive licence) · get_code("c635ec1357ea148d")
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 float, 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 torch

torch.manual_seed(0)

# Analyzing the function:
# - keys_used: a dict-like object that supports indexing with [k]
# - k: an integer or key to index into keys_used
# - cur_time: a numeric value (float or int) that supports subtraction
# The function checks if keys_used[k] is not None, and if so, computes cur_time - keys_used[k]

# Construct valid arguments
keys_used = {
    0: 100.5,
    1: None,
    2: 250.0
}

k = 0

cur_time = 700.5

ARGS = {
    "keys_used": keys_used,
    "k": k,
    "cur_time": cur_time
}
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 float, 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
instructed_generation Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
yuxixie/echo/ToM_CoT_reasoning/prompt_captioning.py
code served (permissive licence) · get_code("0ffac16b1589bc30")
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 list, 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

torch.manual_seed(0)

class MockVisualProcessor(nn.Module):
    def __init__(self):
        super().__init__()
        self.linear = nn.Linear(3, 4)
    
    def forward(self, x):
        return self.linear(x)

class MockGenerativeModel(nn.Module):
    def __init__(self):
        super().__init__()
        self.linear = nn.Linear(4, 8)
    
    def generate(self, inputs, **kwargs):
        captions = ["a cat sitting on a mat"] * kwargs.get('num_captions', 1)
        return captions

class MockVisEval(nn.Module):
    def __init__(self):
        super().__init__()
        self.linear = nn.Linear(3, 4)
    
    def forward(self, x):
        return self.linear(x)

vis_eval = MockVisEval()
generative_model = MockGenerativeModel()

model_dict = {
    'vis': {
        'eval': vis_eval
    },
    'model': generative_model
}

img = torch.randn(2, 3)

ARGS = {
    'img': img,
    'model_dict': model_dict,
    'prompt': 'a photo of',
    'device': 'cpu'
}
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 list, ndim 1, dtype family float — 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
select_key Ran
ran; our contract draft was wrong, not the code
yuxixie/echo/ToM_CoT_reasoning/prompt_generation.py
code served (permissive licence) · get_code("0d2e6a0e358a59ab")
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. 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 random
from time import time

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

# Construct arguments for select_key function
keys_used = {
    'key_a': time() - 700,  # Last used 700 seconds ago (> 600)
    'key_b': time() - 300,  # Last used 300 seconds ago (< 600)
    'key_c': None,          # Never used
}

keys = ['key_a', 'key_b']

all_keys = ['key_a', 'key_b', 'key_c', 'key_d', 'key_e', 'key_f']

ARGS = {
    'keys_used': keys_used,
    'keys': keys,
    'all_keys': all_keys,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
Imports
none outside the standard library — 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
select_key Ran
ran; our contract draft was wrong, not the code
yuxixie/echo/ToM_CoT_reasoning/prompt_inference.py
code served (permissive licence) · get_code("401d6a12ef4cfcb0")
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. 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 random
from time import time

# Set random seed for reproducibility
random.seed(42)

# Create keys_used dictionary - maps keys to timestamps (or None)
keys_used = {
    'key_1': None,
    'key_2': time() - 700.0,  # Last used 700 seconds ago (> 600)
    'key_3': time() - 300.0,  # Last used 300 seconds ago (< 600)
    'key_4': None,
    'key_5': time() - 800.0,  # Last used 800 seconds ago (> 600)
}

# Create keys list - subset of available keys
keys = ['key_1', 'key_3', 'key_4']

# Create all_keys list - all possible keys
all_keys = ['key_1', 'key_2', 'key_3', 'key_4', 'key_5', 'key_6', 'key_7']

ARGS = {
    'keys_used': keys_used,
    'keys': keys,
    'all_keys': all_keys,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
Imports
none outside the standard library — 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

Also present in other papers' code

Of 6 samples lifted from this paper's linked repositories, 6 carry a function identity and 0 do not. Of the 6 distinct functions with an identity, 6 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 0 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
get_data 1 other paper
1 same year, order not determinable
2023.findings-emnlp.268 (2023 (year only))
ECHO: A Visio-Linguistic Dataset for Event Causality Inference via Human-Centric ReasOning
get_data 1 other paper
1 same year, order not determinable
2023.findings-emnlp.268 (2023 (year only))
ECHO: A Visio-Linguistic Dataset for Event Causality Inference via Human-Centric ReasOning
get_internal 1 other paper
1 same year, order not determinable
2023.findings-emnlp.268 (2023 (year only))
ECHO: A Visio-Linguistic Dataset for Event Causality Inference via Human-Centric ReasOning
instructed_generation 1 other paper
1 same year, order not determinable
2023.findings-emnlp.268 (2023 (year only))
ECHO: A Visio-Linguistic Dataset for Event Causality Inference via Human-Centric ReasOning
select_key 1 other paper
1 same year, order not determinable
2023.findings-emnlp.268 (2023 (year only))
ECHO: A Visio-Linguistic Dataset for Event Causality Inference via Human-Centric ReasOning
select_key 1 other paper
1 same year, order not determinable
2023.findings-emnlp.268 (2023 (year only))
ECHO: A Visio-Linguistic Dataset for Event Causality Inference via Human-Centric ReasOning

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 introduce ECHo (Event Causality Inference via Human-Centric Reasoning), a diagnostic dataset of event causality inference grounded in visio-linguistic social scenarios. ECHo employs real-world human-centric deductive information building on a television crime drama. ECHo requires the Theory-of-Mind (ToM) ability to understand and reason about social interactions based on multimodal information. Using ECHo, we propose a unified Chain-of-Thought (CoT) framework to assess the reasoning capability of current AI systems. Our ToM-enhanced CoT pipeline accommodates various large foundation models in both zero-shot and few-shot visio-linguistic reasoning. We use this framework to scrutinize recent large foundation models such as InstructGPT and MiniGPT-4 on three diagnostic human-centric tasks. Further analysis demonstrates ECHo as a challenging dataset to expose imperfections and inconsistencies in reasoning. Our data and code are publicly available at https://github.com/YuxiXie/ECHo.

For agents

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

get_harvested_code_for_paper("2305.14740")
get_code_for_paper("2305.14740")
have("2305.14740")

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

curl https://syntology.ai/api/ran/2305.14740.json

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

[![Syntology run record](https://syntology.ai/api/ran/2305.14740.svg)](https://syntology.ai/paper/2305.14740)

Connect an agent — have() is free.