SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2010.09240 · EMNLP Findings · 2020

Multi-hop Question Generation with Graph Convolutional Network

Pascale Fung, Ziwei Ji, Yan Xu, Dan Su, Wenliang Dai, Tiezheng Yu

arXiv · PDF · Open in the Atlas

Code that ran

11 of 15 samples (functions and classes) lifted from this paper's linked repositories executed on a synthesized input in Syntology's jail (4 of them constructed an object rather than computing a result; 0 honoured the contract we drafted, 1 violated it and 6 have no contract verdict; 4 failures of our instrument; 4 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/2010.09240.json · badge: /api/ran/2010.09240.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
HLTCHKUST/MulQG community
Papers with Code archive
11 ran · 4 no recorded run
FunctionStatusWhere it lives
AttentionLayer Ran
ran on a synthesized input · constructed an object; no result was computed
HLTCHKUST/MulQG/GPG/encoders/graphfusion_encoder.py
code served (permissive licence) · get_code("50bbca6431037b3b")
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 an instance of its own class, AttentionLayer: the run constructed the object and did not call it, so no result was computed. 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
from torch import nn
import torch.nn.functional as F

torch.manual_seed(0)

class Config:
    def __init__(self):
        self.gnn_drop = 0.1
        self.q_attn = True
        self.n_type = 2
        self.hidden_dim = 8

def get_weights(size, gain=1.414):
    weights = nn.Parameter(torch.zeros(size=size))
    nn.init.xavier_uniform_(weights, gain=gain)
    return weights

def get_act(act):
    if act.startswith('lrelu'):
        return nn.LeakyReLU(float(act.split(':')[1]))
    elif act == 'relu':
        return nn.ReLU()
    else:
        raise NotImplementedError

# Create config
config = Config()

# Parameters for AttentionLayer.__init__
in_dim = 4
hid_dim = 8
n_head = 2
layer_id = 0

ARGS = {
    'in_dim': in_dim,
    'hid_dim': hid_dim,
    'n_head': n_head,
    'config': config,
    'layer_id': layer_id
}
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 AttentionLayer — the run constructed an instance of its own class and did not call it; no shape is stated — the instrument coerces any output to an array; for a AttentionLayer 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
BiAttention Ran
ran on a synthesized input · constructed an object; no result was computed
HLTCHKUST/MulQG/GPG/encoders/graphfusion_encoder.py
code served (permissive licence) · get_code("f12a92994940cda5")
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 an instance of its own class, BiAttention: the run constructed the object and did not call it, so no result was computed. 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
from torch import nn
import numpy as np

torch.manual_seed(0)

# Define the arguments dict for BiAttention
ARGS = {
    'input_dim': 4,
    'memory_dim': 4,
    'hid_dim': 8,
    'dropout': 0.1
}
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 BiAttention — the run constructed an instance of its own class and did not call it; no shape is stated — the instrument coerces any output to an array; for a BiAttention 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
GATSelfAttention Ran
ran on a synthesized input · constructed an object; no result was computed
HLTCHKUST/MulQG/GPG/encoders/graphfusion_encoder.py
code served (permissive licence) · get_code("94dafe1d4506cc7e")
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 an instance of its own class, GATSelfAttention: the run constructed the object and did not call it, so no result was computed. 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
from torch import nn
import torch.nn.functional as F

torch.manual_seed(0)

class Config:
    def __init__(self):
        self.gnn_drop = 0.1
        self.q_attn = True
        self.n_type = 2
        self.hidden_dim = 3

config = Config()

# Parameters for GATSelfAttention.__init__
in_dim = 4
out_dim = 5
layer_id = 0
head_id = 0

# Create an instance to get the module
gat_module = None

# Now construct forward pass arguments
batch_size = 2
num_entities = 3

# input_state: (batch, entities, in_dim)
input_state = torch.randn(batch_size, num_entities, in_dim)

# adj: (batch, entities, entities) with values in range [0, n_type+1]
adj = torch.randint(0, config.n_type + 2, (batch_size, num_entities, num_entities))

# entity_mask: (batch, entities, entities) - used as mask
entity_mask = torch.ones(batch_size, num_entities, num_entities)

# adj_mask: optional mask
adj_mask = torch.ones(batch_size, num_entities, in_dim)

# query_vec: (batch, hidden_dim * 2) - used when q_attn is True
query_vec = torch.randn(batch_size, config.hidden_dim * 2)

ARGS = {
    'in_dim': in_dim,
    'out_dim': out_dim,
    'config': config,
    'layer_id': layer_id,
    'head_id': head_id,
}
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 GATSelfAttention — the run constructed an instance of its own class and did not call it; no shape is stated — the instrument coerces any output to an array; for a GATSelfAttention 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
MeanMaxPooling Ran
ran on a synthesized input
HLTCHKUST/MulQG/GPG/encoders/graphfusion_encoder.py
code served (permissive licence) · get_code("d6cddce9da54ec40")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, class construct-then-call prompt, model not recorded) and returned Tensor, ndim 3, 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

# MeanMaxPooling has no parameters in __init__
INIT_ARGS = {}

# Create input tensors for the forward method
# doc_state: N x L x d
# entity_mapping: N x E x L
# entity_lens: N x E

N = 2  # batch size
L = 5  # sequence length
d = 4  # hidden dimension
E = 3  # number of entities

doc_state = torch.randn(N, L, d)
entity_mapping = torch.zeros(N, E, L)

# Create some entity mappings (binary masks indicating which tokens belong to each entity)
# Entity 0 in batch 0: tokens 0-1
entity_mapping[0, 0, 0:2] = 1.0
# Entity 1 in batch 0: tokens 2-3
entity_mapping[0, 1, 2:4] = 1.0
# Entity 2 in batch 0: token 4
entity_mapping[0, 2, 4:5] = 1.0

# Entity 0 in batch 1: tokens 0-2
entity_mapping[1, 0, 0:3] = 1.0
# Entity 1 in batch 1: token 3
entity_mapping[1, 1, 3:4] = 1.0
# Entity 2 in batch 1: token 4
entity_mapping[1, 2, 4:5] = 1.0

# Entity lengths (number of tokens per entity)
entity_lens = torch.tensor([[2.0, 2.0, 1.0],
                            [3.0, 1.0, 1.0]])

ARGS = {
    'doc_state': doc_state,
    'entity_mapping': entity_mapping,
    'entity_lens': entity_lens
}
Setup block drafted by
a language model, class construct-then-call prompt, model not recorded
Imports
numpy, torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type Tensor, ndim 3, dtype family float — no output shape is stated: observe_signature.py observed this class with its constructor sizes canonicalised where it could, which the node does not record, so its shape is not stated as this setup block's
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: torch and numpy seeded to 0 by the harness immediately before the class was constructed from INIT_ARGS (fixture_check.py)
MeanPooling Ran
ran on a synthesized input
HLTCHKUST/MulQG/GPG/encoders/graphfusion_encoder.py
code served (permissive licence) · get_code("d16b854d26a8c986")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, class construct-then-call prompt, model not recorded) and returned Tensor, ndim 3, 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

# MeanPooling has no parameters in __init__, so INIT_ARGS is empty
INIT_ARGS = {}

# For the forward method, we need:
# - doc_state: document token representations
# - entity_mapping: binary mask indicating which tokens belong to which entities
# - entity_lens: number of tokens per entity

batch_size = 2
seq_len = 5
hidden_dim = 4
num_entities = 3

# doc_state: (batch_size, seq_len, hidden_dim)
doc_state = torch.randn(batch_size, seq_len, hidden_dim)

# entity_mapping: (batch_size, num_entities, seq_len)
# Binary mask where entity_mapping[b, e, t] = 1 if token t belongs to entity e in batch b
entity_mapping = torch.zeros(batch_size, num_entities, seq_len)
# Assign some tokens to entities
entity_mapping[0, 0, 0:2] = 1.0  # entity 0 spans tokens 0-1
entity_mapping[0, 1, 2:4] = 1.0  # entity 1 spans tokens 2-3
entity_mapping[0, 2, 4:5] = 1.0  # entity 2 spans token 4
entity_mapping[1, 0, 0:3] = 1.0  # entity 0 spans tokens 0-2
entity_mapping[1, 1, 3:4] = 1.0  # entity 1 spans token 3
entity_mapping[1, 2, 4:5] = 1.0  # entity 2 spans token 4

# entity_lens: (batch_size, num_entities)
# Number of tokens in each entity
entity_lens = torch.tensor([[2.0, 2.0, 1.0],
                            [3.0, 1.0, 1.0]])

ARGS = {
    'doc_state': doc_state,
    'entity_mapping': entity_mapping,
    'entity_lens': entity_lens
}
Setup block drafted by
a language model, class construct-then-call prompt, model not recorded
Imports
numpy, torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type Tensor, ndim 3, dtype family float — no output shape is stated: observe_signature.py observed this class with its constructor sizes canonicalised where it could, which the node does not record, so its shape is not stated as this setup block's
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: torch and numpy seeded to 0 by the harness immediately before the class was constructed from INIT_ARGS (fixture_check.py)
ReasonLayer Ran
ran on a synthesized input · constructed an object; no result was computed
HLTCHKUST/MulQG/GPG/encoders/graphfusion_encoder.py
code served (permissive licence) · get_code("9b817c7350545867")
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 an instance of its own class, ReasonLayer: the run constructed the object and did not call it, so no result was computed. 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
from torch import nn

torch.manual_seed(0)

class Config:
    def __init__(self, num_layers=1, dropout=0.0):
        self.num_layers = num_layers
        self.dropout = dropout

doc_dim = 4
config = Config(num_layers=1, dropout=0.0)

doc_encoder_outputs = torch.randn(2, 3, 4)
ans_encoder_outputs = torch.randn(2, 5, 4)

ARGS = {
    'doc_dim': doc_dim,
    'config': config
}
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 ReasonLayer — the run constructed an instance of its own class and did not call it; no shape is stated — the instrument coerces any output to an array; for a ReasonLayer 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
cal_position_id_2 Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
HLTCHKUST/MulQG/GPG/encoders/graphfusion_encoder.py
code served (permissive licence) · get_code("ef246f715f2a9bb4")
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 Tensor, ndim 2, dtype family int, shape (2, 4) 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)

# Batch size 2, sequence length 4
batch_size = 2
seq_length = 4

# input_ids: shape (batch_size, seq_length)
input_ids = torch.randint(0, 100, (batch_size, seq_length), dtype=torch.long)

# ans_start_ids: shape (batch_size,) with values in valid range [0, seq_length-1]
ans_start_ids = torch.tensor([1, 0], dtype=torch.long)

# ans_end_ids: shape (batch_size,) with values >= ans_start_ids and < seq_length
ans_end_ids = torch.tensor([2, 2], dtype=torch.long)

# context_mask: shape (batch_size, seq_length) with binary values
context_mask = torch.ones((batch_size, seq_length), dtype=torch.float32)

ARGS = {
    "input_ids": input_ids,
    "ans_start_ids": ans_start_ids,
    "ans_end_ids": ans_end_ids,
    "context_mask": context_mask,
}
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 Tensor, ndim 2, dtype family int, shape (2, 4) — shape (2, 4) 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
get_act Ran
ran; our contract draft was wrong, not the code
HLTCHKUST/MulQG/GPG/encoders/graphfusion_encoder.py
code served (permissive licence) · get_code("7716abb39c80c83f")
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 LeakyReLU. 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
from torch import nn

def get_act(act):
    if act.startswith('lrelu'):
        return nn.LeakyReLU(float(act.split(':')[1]))
    elif act == 'relu':
        return nn.ReLU()
    else:
        raise NotImplementedError

ARGS = {
    'act': 'lrelu:0.2'
}
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 LeakyReLU — a LeakyReLU output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a LeakyReLU 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_weights Ran
ran; our contract draft was wrong, not the code
HLTCHKUST/MulQG/GPG/encoders/graphfusion_encoder.py
code served (permissive licence) · get_code("890a496ec7f7b9fd")
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 Parameter, ndim 2, 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
from torch import nn

torch.manual_seed(0)

ARGS = {
    "size": (2, 4),
    "gain": 1.414
}
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 Parameter, ndim 2, dtype family float — no output shape is stated: observe_signature.py recorded Tensor ndim 2, which disagrees with the fixture run
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
mean_pooling Ran
ran, violated the contract we drafted
HLTCHKUST/MulQG/GPG/encoders/graphfusion_encoder.py
code served (permissive licence) · get_code("567eab7347ce0c60")
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 Tensor, ndim 3, dtype family float, shape (2, 2, 3) 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 torch.nn as nn

torch.manual_seed(0)

# Create input tensor: (batch_size=2, seq_length=4, hidden_dim=3)
input = torch.randn(2, 4, 3)

# Create mask tensor: (batch_size=2, seq_length=4)
# Mask has 1s for valid positions and 0s for padding
mask = torch.ones(2, 4)
mask[0, 3] = 0  # Mask out last token in first sample
mask[1, 2:] = 0  # Mask out last two tokens in second sample
mask = mask.unsqueeze(-1)  # Shape: (2, 4, 1) to broadcast properly in sum

ARGS = {
    "input": input,
    "mask": mask,
}
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 Tensor, ndim 3, dtype family float, shape (2, 2, 3) — shape (2, 2, 3) 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
tok_to_ent Ran
ran; our contract draft was wrong, not the code
HLTCHKUST/MulQG/GPG/encoders/graphfusion_encoder.py
code served (permissive licence) · get_code("d3bf305922e49e02")
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 type. 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
from torch import nn

torch.manual_seed(0)

# Test case 1: tok2ent with 'mean'
ARGS_mean = {
    'tok2ent': 'mean'
}

# Test case 2: tok2ent with 'mean_max'
ARGS_mean_max = {
    'tok2ent': 'mean_max'
}

# Use the mean_max variant as the primary test
ARGS = {
    'tok2ent': 'mean_max'
}
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 type — a type output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a type 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
AnsUpdateLayer Not yet run
unverified (harvested, no recorded run)
HLTCHKUST/MulQG/GPG/encoders/graphfusion_encoder.py
code served (permissive licence) · get_code("bdd31cfb2ec8b9bf")
How it ran

no invocation recorded.

GraphFusionEncoder Not yet run
unverified (harvested, no recorded run)
HLTCHKUST/MulQG/GPG/encoders/graphfusion_encoder.py
code served (permissive licence) · get_code("a6aee6914b365ae4")
How it ran

no invocation recorded.

Reason_Ans_InterBlocks Not yet run
unverified (harvested, no recorded run)
HLTCHKUST/MulQG/GPG/encoders/graphfusion_encoder.py
code served (permissive licence) · get_code("e95d07bbe169ee9d")
How it ran

no invocation recorded.

init_lstm_wt Not yet run
unverified (harvested, no recorded run)
HLTCHKUST/MulQG/GPG/encoders/graphfusion_encoder.py
code served (permissive licence) · get_code("ac38dbf4fe6b8b25")
How it ran

no invocation recorded.

Also present in other papers' code

Of 15 samples lifted from this paper's linked repositories, 15 carry a function identity and 0 do not. Of the 15 distinct functions with an identity, 2 are also present in other papers whose code we hold (2 with at least one paper dated earlier, 0 the same month, 0 later) and 13 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
mean_pooling
brick of this paper's assembly: helper
1 other paper
1 dated earlier
1911.03631 (2019-11)
Hierarchical Graph Network for Multi-hop Question Answering
tok_to_ent
brick of this paper's assembly: helper
1 other paper
1 dated earlier
1911.03631 (2019-11)
Hierarchical Graph Network for Multi-hop Question Answering

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

Multi-hop Question Generation (QG) aims to generate answer-related questions by aggregating and reasoning over multiple scattered evidence from different paragraphs. It is a more challenging yet under-explored task compared to conventional single-hop QG, where the questions are generated from the sentence containing the answer or nearby sentences in the same paragraph without complex reasoning. To address the additional challenges in multi-hop QG, we propose Multi-Hop Encoding Fusion Network for Question Generation (MulQG), which does context encoding in multiple hops with Graph Convolutional Network and encoding fusion via an Encoder Reasoning Gate. To the best of our knowledge, we are the first to tackle the challenge of multi-hop reasoning over paragraphs without any sentence-level information. Empirical results on HotpotQA dataset demonstrate the effectiveness of our method, in comparison with baselines on automatic evaluation metrics. Moreover, from the human evaluation, our proposed model is able to generate fluent questions with high completeness and outperforms the strongest baseline by 20.8% in the multihop evaluation. The code is publicly available at https://github.com/HLTCHKUST/MulQG.

For agents

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

get_harvested_code_for_paper("2010.09240")
get_code_for_paper("2010.09240")
have("2010.09240")

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

curl https://syntology.ai/api/ran/2010.09240.json

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

[![Syntology run record](https://syntology.ai/api/ran/2010.09240.svg)](https://syntology.ai/paper/2010.09240)

Connect an agent — have() is free.