SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2210.03992 · EMNLP · 2022

Generative Language Models for Paragraph-Level Question Generation

Jose Camacho-Collados, Fernando Alva-Manchego, Asahi Ushio

arXiv · PDF · Open in the Atlas

Code that ran

1 of 1 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; 0 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/2210.03992.json · badge: /api/ran/2210.03992.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
asahi417/lm-question-generation canonical (classifier label)
classifier basis: classifier_2025
1 ran · 0 no recorded run
FunctionStatusWhere it lives
score Ran
ran, honoured the contract we drafted
asahi417/lm-question-generation/lmqg/automatic_evaluation_tool/bert_score/score.py
code served (permissive licence) · get_code("a57915bf6208b453")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Sonnet 4.5, prompt python_fixture_v1) and returned tuple, ndim 2, dtype family float; no output shape is stated. 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 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
from collections import defaultdict

torch.manual_seed(0)

# Stand-in tokenizer class
class MockTokenizer:
    def __init__(self):
        self.sep_token_id = 102
        self.cls_token_id = 101
        self.pad_token_id = 0
        
    def encode(self, text, add_special_tokens=True):
        # Simple mock encoding
        return [self.cls_token_id] + list(range(10, 10 + len(text.split()))) + [self.sep_token_id]
    
    def __call__(self, texts, padding=True, truncation=True, max_length=512, return_tensors=None):
        # Mock batch encoding
        encoded = []
        for text in texts:
            ids = self.encode(text)
            encoded.append(ids)
        
        # Pad to same length
        max_len = max(len(e) for e in encoded)
        attention_mask = []
        for i in range(len(encoded)):
            mask = [1] * len(encoded[i])
            while len(encoded[i]) < max_len:
                encoded[i].append(self.pad_token_id)
                mask.append(0)
            attention_mask.append(mask)
        
        if return_tensors == 'pt':
            return {
                'input_ids': torch.tensor(encoded),
                'attention_mask': torch.tensor(attention_mask)
            }
        return {'input_ids': encoded, 'attention_mask': attention_mask}

# Stand-in model class
class MockBertModel(nn.Module):
    def __init__(self, hidden_size=768, num_layers=12):
        super().__init__()
        self.hidden_size = hidden_size
        self.num_layers = num_layers
        self.embeddings = nn.Embedding(30522, hidden_size)
        self.encoder = nn.ModuleList([nn.Linear(hidden_size, hidden_size) for _ in range(num_layers)])
        
    def forward(self, input_ids, attention_mask=None, output_hidden_states=False):
        embeddings = self.embeddings(input_ids)
        hidden_states = [embeddings]
        
        x = embeddings
        for layer in self.encoder:
            x = layer(x)
            hidden_states.append(x)
        
        class Output:
            pass
        
        output = Output()
        if output_hidden_states:
            output.hidden_states = tuple(hidden_states)
        output.last_hidden_state = x
        return output

# Mock global functions that the score function expects
def get_tokenizer(model_type, use_fast):
    return MockTokenizer()

def get_model(model_type, num_layers, all_layers):
    return MockBertModel(hidden_size=768, num_layers=num_layers)

def get_idf_dict(refs, tokenizer, nthreads=4):
    idf_dict = defaultdict(lambda: 1.0)
    idf_dict[tokenizer.sep_token_id] = 0
    idf_dict[tokenizer.cls_token_id] = 0
    return idf_dict

def bert_cos_score_idf(model, refs, cands, tokenizer, idf_dict, verbose=False, device='cpu', batch_size=64, all_layers=False):
    # Mock implementation that returns scores
    num_pairs = len(cands)
    if all_layers:
        # Return shape (num_pairs, num_layers, 3) for P, R, F
        return torch.randn(num_pairs, model.num_layers, 3)
    else:
        # Return shape (num_pairs, 3) for P, R, F
        return torch.randn(num_pairs, 3)

# Mock lookup dictionaries
lang2model = {
    'en': 'bert-base-uncased',
    'zh': 'bert-base-chinese',
    'es': 'bert-base-multilingual-cased',
}

model2layers = {
    'bert-base-uncased': 9,
    'bert-base-chinese': 9,
    'bert-base-multilingual-cased': 9,
    'roberta-base': 9,
}

# Inject mocks into the appropriate namespace
import sys
current_module = sys.modules[__name__]
current_module.get_tokenizer = get_tokenizer
current_module.get_model = get_model
current_module.get_idf_dict = get_idf_dict
current_module.bert_cos_score_idf = bert_cos_score_idf
current_module.lang2model = lang2model
current_module.model2layers = model2layers

# Create arguments
ARGS = {
    'cands': ['This is a test sentence.', 'Another candidate sentence.'],
    'refs': ['This is a reference sentence.', 'Another reference sentence.'],
    'model_type': 'bert-base-uncased',
    'num_layers': 9,
    'verbose': False,
    'idf': False,
    'device': 'cpu',
    'batch_size': 2,
    'nthreads': 2,
    'all_layers': False,
    'lang': 'en',
    'return_hash': False,
    'rescale_with_baseline': False,
    'baseline_path': None,
    'use_fast_tokenizer': False,
}
Setup block drafted by
a language model, Claude Sonnet 4.5, prompt python_fixture_v1
Imports
pandas, 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 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 1 samples lifted from this paper's linked repositories, 1 carry a function identity and 0 do not. Of the 1 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 1 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

Powerful generative models have led to recent progress in question generation (QG). However, it is difficult to measure advances in QG research since there are no standardized resources that allow a uniform comparison among approaches. In this paper, we introduce QG-Bench, a multilingual and multidomain benchmark for QG that unifies existing question answering datasets by converting them to a standard QG setting. It includes generalpurpose datasets such as SQuAD (Rajpurkar et al., 2016) for English, datasets from ten domains and two styles, as well as datasets in eight different languages. Using QG-Bench as a reference, we perform an extensive analysis of the capabilities of language models for the task. First, we propose robust QG baselines based on fine-tuning generative language models. Then, we complement automatic evaluation based on standard metrics with an extensive manual evaluation, which in turn sheds light on the difficulty of evaluating QG models. Finally, we analyse both the domain adaptability of these models as well as the effectiveness of multilingual models in languages other than English. QG-Bench is released along with the fine-tuned models presented in the paper, 1 which are also available as a demo. 2

For agents

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

get_harvested_code_for_paper("2210.03992")
get_code_for_paper("2210.03992")
have("2210.03992")

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

curl https://syntology.ai/api/ran/2210.03992.json

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

[![Syntology run record](https://syntology.ai/api/ran/2210.03992.svg)](https://syntology.ai/paper/2210.03992)

Connect an agent — have() is free.