SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2602.01246 · 2026

Parse: An Open-Domain Reasoning Question Answering Benchmark for Persian

Adam Jatowt, Jamshid Mozafari, Seyed Mousavinasab, English Persian

arXiv · PDF · Open in the Atlas

Code that ran

9 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; 1 honoured the contract we drafted, 0 violated it and 1 has no contract verdict; 7 failures of our instrument; 0 no recorded run), measured 2026-11-22; 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/2602.01246.json · badge: /api/ran/2602.01246.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
DataScienceUIBK/Parse canonical (classifier label)
classifier basis: census
9 ran · 0 no recorded run
FunctionStatusWhere it lives
boolean Ran
ran; our contract draft was wrong, not the code
DataScienceUIBK/Parse/finetune/to_together_ai.py
code served (permissive licence) · get_code("ff4751a0d146efe1")
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

# Analyze the function to determine required arguments
# The function takes:
# - question: a string used in prompt_template.replace('<Question>', question)
# - answer: a boolean value used in 'Yes' if answer else 'No'
# - prompt_template: a string that contains '<Question>' placeholder

ARGS = {
    'question': 'Is the sky blue?',
    'answer': True,
    'prompt_template': 'The user asks: <Question> Please answer yes or no.'
}
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
boolean_multihop_comparative Ran
ran; our contract draft was wrong, not the code
DataScienceUIBK/Parse/evaluation/human_difficulty_validation/evaluation_results.py
code served (permissive licence) · get_code("448e9a368a4099c6")
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

torch.manual_seed(0)

ground_truth = {
    'difficulty': 2,
    'answer': True
}

prediction = {
    'human_answer': True
}

ARGS = {
    'ground_truth': ground_truth,
    'prediction': prediction
}
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 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
boolean_multihop_negation Ran
ran; our contract draft was wrong, not the code
DataScienceUIBK/Parse/evaluation/human_difficulty_validation/evaluation_results.py
code served (permissive licence) · get_code("22782fc5e1893dcf")
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

torch.manual_seed(0)

ground_truth = {
    'difficulty': 2,
    'answer': True
}

prediction = {
    'human_answer': True
}

ARGS = {
    'ground_truth': ground_truth,
    'prediction': prediction
}
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 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
boolean_multihop_simple Ran
ran; our contract draft was wrong, not the code
DataScienceUIBK/Parse/evaluation/human_difficulty_validation/evaluation_results.py
code served (permissive licence) · get_code("8381eb414ffd3b36")
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 numpy as np

torch.manual_seed(0)

# The function expects:
# - ground_truth: dict with keys 'difficulty' and 'answer'
# - prediction: dict with key 'human_answer'
# Both 'answer' values should be comparable (booleans or same type)

ground_truth = {
    'difficulty': 3,
    'answer': True
}

prediction = {
    'human_answer': True
}

ARGS = {
    'ground_truth': ground_truth,
    'prediction': prediction
}
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
evaluate_boolean Ran
ran, honoured the contract we drafted
DataScienceUIBK/Parse/evaluation/chain_of_thought/evaluate_finetuned_results.py
code served (permissive licence) · get_code("ff98a74e5b67cf9e")
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

torch.manual_seed(0)

# The function evaluate_boolean simply compares answer == predicted_answer
# and returns 1 if they match, 0 otherwise.
# Both answer and predicted_answer appear to be values that support equality comparison.
# sub_sub_method is not used in the function body.

# Set up simple comparable values
answer = "yes"
predicted_answer = "yes"
sub_sub_method = "some_method"

ARGS = {
    "answer": answer,
    "predicted_answer": predicted_answer,
    "sub_sub_method": sub_sub_method
}
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
evaluate_multichoice Ran
ran on a synthesized input
DataScienceUIBK/Parse/evaluation/chain_of_thought/evaluate_finetuned_results.py
code served (permissive licence) · get_code("e8073d271652a4e3")
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

torch.manual_seed(0)

# Analyze the function to understand what's needed:
# - answer: str or list of strings (answer text)
# - predicted_answer: list of integers (answer IDs)
# - options: list of strings (available options)
# - sub_sub_method: str (one of 'multi', 'non_answerable', 'single')

# The function:
# 1. Converts answer to list if it's a string
# 2. Finds indices of answers in options (adds 1 to index)
# 3. Checks if predicted_answer is empty
# 4. Based on sub_sub_method, computes overlap/equality metrics

# Set up valid arguments
options = ["apple", "banana", "cherry", "date"]
answer = "apple"  # Can be str or list
predicted_answer = [1]  # List of integers (1-indexed answer IDs)
sub_sub_method = "single"  # One of: 'multi', 'non_answerable', 'single'

ARGS = {
    "answer": answer,
    "predicted_answer": predicted_answer,
    "options": options,
    "sub_sub_method": sub_sub_method
}
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
factoid Ran
ran; our contract draft was wrong, not the code
DataScienceUIBK/Parse/finetune/to_together_ai.py
code served (permissive licence) · get_code("17e925f97dc019db")
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

torch_imported = False
try:
    import torch
    torch_imported = True
    torch.manual_seed(0)
except ImportError:
    pass

# Define arguments for the factoid function
question = "What is the capital of France?"
answer = "Paris"
prompt_template = "Question: <Question>\nProvide the answer:"

ARGS = {
    'question': question,
    'answer': answer,
    'prompt_template': prompt_template
}
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
multichoice Ran
ran; our contract draft was wrong, not the code
DataScienceUIBK/Parse/finetune/to_together_ai.py
code served (permissive licence) · get_code("9ad5a0ab7b9aadb9")
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

ARGS = {
    'question': 'What is the capital of France?',
    'answer': 'Paris',
    'options': ['London', 'Paris', 'Berlin', 'Madrid'],
    'prompt_template': 'Question: <Question>\nOptions:\n1. <Option_1>\n2. <Option_2>\n3. <Option_3>\n4. <Option_4>\nAnswer:'
}
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
prepare_answers_as_dict Ran
ran; our contract draft was wrong, not the code
DataScienceUIBK/Parse/evaluation/chain_of_thought/evaluate_finetuned_results.py
code served (permissive licence) · get_code("1bd5d858552e40c2")
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

torch.manual_seed(0)

# Construct the answers list with the expected structure
answers = [
    {'id': 'q1', 'predicted_answer': 'answer_one'},
    {'id': 'q2', 'predicted_answer': 'answer_two'},
    {'id': 'q3', 'predicted_answer': 'answer_three'},
]

ARGS = {
    'answers': answers
}
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

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

Reasoning-focused Question Answering (QA) has advanced rapidly with Large Language Models (LLMs), yet high-quality benchmarks for low-resource languages remain scarce. Persian, spoken by roughly 130 million people, lacks a comprehensive open-domain resource for evaluating reasoning-capable QA systems. We introduce Parse, the first open-domain Persian reasoning QA benchmark, containing 10,800 questions across Boolean, multiple-choice, and factoid formats, with diverse reasoning types, difficulty levels, and answer structures. The benchmark is built via a controlled LLM-based generation pipeline and validated through human evaluation. We also ensure linguistic and factual quality through multi-stage filtering, annotation, and consistency checks. We benchmark multilingual and Persian LLMs under multiple prompting strategies and show that Persian prompts and structured prompting (CoT for Boolean/multiple-choice; few-shot for factoid) improve performance. Fine-tuning further boosts results, especially for Persianspecialized models. These findings highlight how Parse supports both fair comparison and practical model adaptation. Parse fills a critical gap in Persian QA research and provides a strong foundation for developing and evaluating reasoning-capable LLMs in low-resource settings. • Information systems → Question answering; Test collections; Retrieval models and ranking; Retrieval tasks and goals; Evaluation of retrieval results.

For agents

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

get_harvested_code_for_paper("2602.01246")
get_code_for_paper("2602.01246")
have("2602.01246")

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

curl https://syntology.ai/api/ran/2602.01246.json

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

[![Syntology run record](https://syntology.ai/api/ran/2602.01246.svg)](https://syntology.ai/paper/2602.01246)

Connect an agent — have() is free.