SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2009.11423 · 2020

Task-Oriented Dialogue as Dataflow Synthesis

arXiv · PDF · Open in the Atlas

Code that ran

11 of 12 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 11 have no contract verdict; 0 failures of our instrument; 1 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/2009.11423.json · badge: /api/ran/2009.11423.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
telepathylabsai/OpenDF community (classifier label)
classifier basis not recorded on the link
11 ran · 1 no recorded run
FunctionStatusWhere it lives
contradict Ran
ran on a synthesized input
telepathylabsai/OpenDF/opendf/applications/core/partial_time.py
code served (permissive licence) · get_code("ad6b942af25baf47")
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 bool, 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)

ARGS = {
    "i": 5,
    "j": 10
}
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 bool, 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
conv_obj Ran
ran on a synthesized input
telepathylabsai/OpenDF/opendf/compare_exec.py
code served (permissive licence) · get_code("c880c79f5c5c924f")
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 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 MockObject:
    def show(self):
        return "mock_output"

class AnotherMockObject:
    def show(self):
        return "another_mock_output"

objs = [
    "string_object",
    MockObject(),
    "another_string",
    AnotherMockObject(),
]

ARGS = {
    "objs": objs
}
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 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
create_index Ran
ran on a synthesized input
telepathylabsai/OpenDF/opendf/main_multiwoz_2_2.py
code served (permissive licence) · get_code("2a9eefd10c472eca")
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 os
import json
import tempfile
import shutil

# Create a temporary directory structure with test data
temp_dir = tempfile.mkdtemp()

# Create folder structure
folder1 = os.path.join(temp_dir, 'folder1')
folder2 = os.path.join(temp_dir, 'folder2')
os.makedirs(folder1)
os.makedirs(folder2)

# Create test JSON files with dialogues
dialogues1 = [
    {'dialogue_id': 'dial_001', 'content': 'hello'},
    {'dialogue_id': 'dial_002', 'content': 'world'}
]

dialogues2 = [
    {'dialogue_id': 'dial_003', 'content': 'foo'},
    {'dialogue_id': 'dial_004', 'content': 'bar'}
]

dialogues3 = [
    {'dialogue_id': 'dial_005', 'content': 'baz'}
]

# Write test JSON files
with open(os.path.join(folder1, 'test1.json'), 'w') as f:
    json.dump(dialogues1, f)

with open(os.path.join(folder1, 'test2.json'), 'w') as f:
    json.dump(dialogues2, f)

with open(os.path.join(folder2, 'test3.json'), 'w') as f:
    json.dump(dialogues3, f)

# Also add a non-JSON file that should be ignored
with open(os.path.join(folder1, 'readme.txt'), 'w') as f:
    f.write('This should be ignored')

ARGS = {
    'data_directory': temp_dir,
    'folders': ['folder1', 'folder2']
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
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 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
dow_to_name Ran
ran on a synthesized input
telepathylabsai/OpenDF/opendf/applications/core/partial_time.py
code served (permissive licence) · get_code("98e09f8a0e13439d")
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

torch.manual_seed(0)

def dow_to_name(i):
    """
    Gets the day of the week based on the index `i`. Such as 1: Monday, ..., 7: Sunday.

    :param i: the index of the day
    :type i: int
    :return: the day of the week
    :rtype: str
    """
    days_of_week = ['Monday', 'Tuesday', 'Wednesday', 'Thursday', 'Friday', 'Saturday', 'Sunday']
    return days_of_week[i - 1]

ARGS = {
    'i': 1
}
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 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
extract_p_expressions Ran
ran on a synthesized input
telepathylabsai/OpenDF/opendf/main_jsonl.py
code served (permissive licence) · get_code("6d622e80f2fc463c")
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 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

# Construct a minimal dialog dict that satisfies the function's requirements
# The function iterates over dialog['turns'] and accesses turn['lispress']

dialog = {
    'turns': [
        {'lispress': '(SELECT * FROM table1)'},
        {'lispress': '(SELECT * FROM table2)'},
    ]
}

ARGS = {
    'dialog': dialog
}
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 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_dialogue_index Ran
ran on a synthesized input
telepathylabsai/OpenDF/opendf/main_multiwoz_2_2.py
code served (permissive licence) · get_code("8ffa0adedd2aa088")
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 tempfile
import json
import os
from pathlib import Path

# Create a temporary directory structure with test data
temp_dir = tempfile.mkdtemp()

# Create subdirectories for each data folder
for folder in ['train', 'dev', 'test', 'gold']:
    os.makedirs(os.path.join(temp_dir, folder), exist_ok=True)

# Create sample JSON files with dialogue data
sample_dialogue_1 = [
    {'dialogue_id': 'dial_001', 'content': 'test dialogue 1'},
    {'dialogue_id': 'dial_002', 'content': 'test dialogue 2'}
]

sample_dialogue_2 = [
    {'dialogue_id': 'dial_003', 'content': 'test dialogue 3'}
]

# Write test data files
with open(os.path.join(temp_dir, 'train', 'data1.json'), 'w') as f:
    json.dump(sample_dialogue_1, f)

with open(os.path.join(temp_dir, 'dev', 'data2.json'), 'w') as f:
    json.dump(sample_dialogue_2, f)

# Define arguments for get_dialogue_index
ARGS = {
    'dialogue_id': 'dial_001',
    'data_directory': temp_dir
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
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 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
get_single_dialogue Ran
ran on a synthesized input
telepathylabsai/OpenDF/opendf/main_multiwoz_2_2.py
code served (permissive licence) · get_code("977aee9caff9d870")
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 json
import logging
import os
import tempfile
import shutil

# Set up logging
logging.basicConfig(level=logging.WARNING)

# Create a temporary directory structure with test data
temp_dir = tempfile.mkdtemp()

# Create the required folder structure
for folder in ['train', 'dev', 'test', 'gold']:
    os.makedirs(os.path.join(temp_dir, folder), exist_ok=True)

# Create sample dialogue data
sample_dialogues = [
    {
        'dialogue_id': 'dialogue_001.json',
        'turns': [
            {'speaker': 'user', 'text': 'hello'},
            {'speaker': 'system', 'text': 'hi there'}
        ]
    },
    {
        'dialogue_id': 'dialogue_002.json',
        'turns': [
            {'speaker': 'user', 'text': 'what time is it'},
            {'speaker': 'system', 'text': 'it is 3pm'}
        ]
    }
]

# Write sample dialogue files
with open(os.path.join(temp_dir, 'train', 'dialogues_001.json'), 'w') as f:
    json.dump(sample_dialogues, f)

with open(os.path.join(temp_dir, 'dev', 'dialogues_002.json'), 'w') as f:
    json.dump(sample_dialogues, f)

# Set up arguments for get_single_dialogue
ARGS = {
    'dialogue_id': 'dialogue_001.json',
    'data_directory': temp_dir
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
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 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
incomparable Ran
ran on a synthesized input
telepathylabsai/OpenDF/opendf/applications/core/partial_time.py
code served (permissive licence) · get_code("29c72b942887d53a")
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 bool, 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
import numpy as np

torch.manual_seed(0)

# The function incomparable(i, j) checks if exactly one of i or j is None
# It returns True if one is None and the other is not, False otherwise
# We need to provide values for both parameters that satisfy the logic

# Case 1: i is None, j is not None (should return True)
i_val = None
j_val = torch.tensor([1, 2, 3, 4])

ARGS = {
    "i": i_val,
    "j": j_val
}
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 bool, 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
is_pos Ran
ran on a synthesized input
telepathylabsai/OpenDF/opendf/defs.py
code served (permissive licence) · get_code("96e3e1ffc655a772")
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 bool, 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

def is_pos(s):
    return s.startswith('pos') and s[len('pos'):].isnumeric()

ARGS = {
    's': 'pos123'
}
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 bool, 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
posname Ran
ran on a synthesized input
telepathylabsai/OpenDF/opendf/defs.py
code served (permissive licence) · get_code("0469449473490a0b")
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

torch.manual_seed(0)

POS = 'pos'

def posname(i):
    return POS + '%d' % i

ARGS = {
    'i': 0
}
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 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
posname_idx Ran
ran on a synthesized input
telepathylabsai/OpenDF/opendf/defs.py
code served (permissive licence) · get_code("719190dcccc5b749")
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 int, ndim 0, dtype family int, 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
import numpy as np

POS = 'pos'

def is_pos(s):
    return s.startswith(POS) and s[len(POS):].isnumeric()

def posname_idx(s):
    return int(s[len(POS):]) if is_pos(s) else 0

# The function posname_idx takes a single parameter 's' which is a string
# It checks if the string starts with 'pos' and has numeric characters after it
# If so, it extracts and returns the numeric part as an integer
# Otherwise it returns 0

ARGS = {
    's': 'pos42'
}
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 int, ndim 0, dtype family int, 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
compare_graph_exec Not yet run
unverified (harvested, no recorded run)
telepathylabsai/OpenDF/opendf/compare_exec.py
code served (permissive licence) · get_code("b34f432ee096c189")
How it ran

no invocation recorded.

Also present in other papers' code

Of 12 samples lifted from this paper's linked repositories, 12 carry a function identity and 0 do not. Of the 12 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 12 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

We describe an approach to task-oriented dialogue in which dialogue state is represented as a dataflow graph. A dialogue agent maps each user utterance to a program that extends this graph. Programs include metacomputation operators for reference and revision that reuse dataflow fragments from previous turns. Our graph-based state enables the expression and manipulation of complex user intents, and explicit metacomputation makes these intents easier for learned models to predict. We introduce a new dataset, SMCalFlow, featuring complex dialogues about events, weather, places, and people. Experiments show that dataflow graphs and metacomputation substantially improve representability and predictability in these natural dialogues. Additional experiments on the MultiWOZ dataset show that our dataflow representation enables an otherwise off-the-shelf sequence-to-sequence model to match the best existing task-specific state tracking model. The SMCalFlow dataset and code for replicating experiments are available at https://www.microsoft.com/en-us/research/project/dataflow-based-dialogue-semantic-machines.

For agents

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

get_harvested_code_for_paper("2009.11423")
get_code_for_paper("2009.11423")
have("2009.11423")

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

curl https://syntology.ai/api/ran/2009.11423.json

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

[![Syntology run record](https://syntology.ai/api/ran/2009.11423.svg)](https://syntology.ai/paper/2009.11423)

Connect an agent — have() is free.