SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2402.08277 · ACL · 2024

Towards Faithful and Robust LLM Specialists for Evidence-Based Question-Answering

Markus Leippold, Elliott Ash, Jingwei Ni, Mathias Kraus, T. Schimanski

arXiv · PDF · Open in the Atlas

Code that ran

2 of 3 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 0 have no contract verdict; 2 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/2402.08277.json · badge: /api/ran/2402.08277.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
EdisonNi-hku/Robust_Evidence_Based_QA marked official in Papers with Code
Papers with Code archive
2 ran · 1 no recorded run
FunctionStatusWhere it lives
createQuestions Ran
ran; our contract draft was wrong, not the code
EdisonNi-hku/Robust_Evidence_Based_QA/data_generation/CreateSyntheticData.py
code served (permissive licence) · get_code("4fe5cd007605b269")
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
import re

# Define a mock model that implements the processPrompt function
class MockLanguageModel(nn.Module):
    def __init__(self):
        super().__init__()
        self.linear = nn.Linear(4, 8)
    
    def forward(self, x):
        return self.linear(x)

# Define a mock processPrompt function that will be called by createQuestions
def processPrompt(prompt, model):
    # Return a string with questions separated by backslashes
    # This mimics what a language model might return
    return "What is the nature of quantum entanglement\\ How do particles interact at subatomic levels\\ What experimental methods verify quantum theory\\ Why does wave-particle duality exist\\"

# Patch the processPrompt into the module scope where createQuestions expects it
import sys
current_module = sys.modules[__name__]
current_module.processPrompt = processPrompt

# Create a small model instance
torch.manual_seed(0)
model = MockLanguageModel()

# Define the arguments for createQuestions
ARGS = {
    "topic": "quantum mechanics",
    "n": 4,
    "model": model
}
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
createSources Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
EdisonNi-hku/Robust_Evidence_Based_QA/data_generation/CreateSyntheticData.py
code served (permissive licence) · get_code("c90d225e793f6530")
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

# Create a mock model that implements the processPrompt function call
class MockLanguageModel(nn.Module):
    def __init__(self):
        super().__init__()
        self.linear = nn.Linear(4, 4)
    
    def forward(self, x):
        return self.linear(x)

# Create a mock processPrompt function that returns generated paragraphs
def processPrompt(prompt, model):
    """Mock function that simulates prompt processing and returns formatted paragraphs"""
    paragraphs = [
        "Machine learning represents a paradigm shift in how we approach problem-solving. The field has evolved significantly over the past decades, with deep neural networks becoming increasingly sophisticated. [Smith et al., 2018, p.42]ENDOFPARAGRAPH",
        "From a practical standpoint, understanding the fundamentals of neural architectures is crucial for practitioners. Researchers have demonstrated that architectural choices directly impact model performance and computational efficiency. [Johnson, 2020, p.156]ENDOFPARAGRAPH",
        "The intersection of theoretical computer science and applied machine learning continues to produce novel insights. Academic institutions worldwide are investigating how these principles can be scaled to real-world applications. [Williams Institute, 2019, p.87]ENDOFPARAGRAPH",
    ]
    return " ".join(paragraphs)

# Build the model
torch.manual_seed(0)
model = MockLanguageModel()

# Define arguments for createSources
ARGS = {
    "topic": "machine learning and neural networks",
    "m": 3,
    "question": "How do neural networks learn and optimize their parameters?",
    "model": model,
}
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
createTopicData Not yet run
unverified (harvested, no recorded run)
EdisonNi-hku/Robust_Evidence_Based_QA/data_generation/CreateSyntheticData.py
code served (permissive licence) · get_code("52d8e56fc22190ac")
How it ran

no invocation recorded.

Also present in other papers' code

Of 3 samples lifted from this paper's linked repositories, 3 carry a function identity and 0 do not. Of the 3 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 3 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

Advances towards more faithful and traceable answers of Large Language Models (LLMs) are crucial for various research and practical endeavors. One avenue in reaching this goal is basing the answers on reliable sources. However, this Evidence-Based QA has proven to work insufficiently with LLMs in terms of citing the correct sources (source quality) and truthfully representing the information within sources (answer attributability). In this work, we systematically investigate how to robustly fine-tune LLMs for better source quality and answer attributability. Specifically, we introduce a data generation pipeline with automated data quality filters, which can synthesize diversified high-quality training and testing data at scale. We further introduce four test sets to benchmark the robustness of fine-tuned specialist models. Extensive evaluation shows that fine-tuning on synthetic data improves performance on both in-and out-of-distribution. Furthermore, we show that data quality, which can be drastically improved by proposed quality filters, matters more than quantity in improving Evidence-Based QA. 1 * Equal Contributions. 1 All our codes, LLM generations, and human annotations are accessible through https://github.com/ EdisonNi-hku/Robust_Evidence_Based_QA.

For agents

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

get_harvested_code_for_paper("2402.08277")
get_code_for_paper("2402.08277")
have("2402.08277")

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

curl https://syntology.ai/api/ran/2402.08277.json

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

[![Syntology run record](https://syntology.ai/api/ran/2402.08277.svg)](https://syntology.ai/paper/2402.08277)

Connect an agent — have() is free.