SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2406.12221 · 2024

On-Policy Self-Alignment with Fine-grained Knowledge Feedback for Hallucination Mitigation

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; 0 honoured the contract we drafted, 0 violated it and 0 have no contract verdict; 1 failure 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/2406.12221.json · badge: /api/ran/2406.12221.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
AlignRM/RLFH canonical (classifier label)
classifier basis: classifier_2025
1 ran · 0 no recorded run
FunctionStatusWhere it lives
paragraph_rewards_from_labels Ran
ran; our contract draft was wrong, not the code
AlignRM/RLFH/source/feedback/reward.py
code served (permissive licence) · get_code("90a4763ce6469af0")
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) and returned list, 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
import torch.nn as nn
from math import log

torch.manual_seed(0)

# Create a minimal config class with required attributes
class Config:
    def __init__(self):
        self.info_weight = 0.5
        self.epsilon = 1e-8
        self.mu = 0.1
        self.truth_weight = 0.3
        # informative_map: maps values to informativeness scores
        self.informative_map = {
            'high': 0.9,
            'medium': 0.5,
            'low': 0.1,
            'none': 0.0
        }
        # verification_map: maps verification status to scores
        self.verification_map = {
            'verified': 1.0,
            'unverified': 0.5,
            'false': 0.0
        }

config = Config()

# Create sample outputs: List[List[Tensor]]
# 2 samples, each with 3 output tensors of shape (4,)
sample_outputs = [
    [torch.randn(4), torch.randn(4), torch.randn(4)],
    [torch.randn(4), torch.randn(4), torch.randn(4)]
]

# Create labels: List[dict]
# Each label is a dict where:
# - keys are sentence indices
# - values are dicts where:
#   - keys are statement indices
#   - values are tuples of (verification_status, informativeness_level)
labels = [
    {
        0: {0: ('verified', 'high'), 1: ('unverified', 'medium')},
        1: {0: ('verified', 'low')}
    },
    {
        0: {0: ('false', 'high'), 1: ('verified', 'medium')},
        1: {0: ('unverified', 'none')}
    }
]

# Tokenizer is not used in the function body, so None is acceptable
tokenizer = None

# Default score for outputs with no labels
default_for_none = 0.0

ARGS = {
    'sample_outputs': sample_outputs,
    'labels': labels,
    'tokenizer': tokenizer,
    'default_for_none': default_for_none,
    'config': config
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
Imports
torch, tqdm — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type list, ndim 2, 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

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

Hallucination occurs when large language models exhibit behavior that deviates from the boundaries of their knowledge during response generation. To address this critical issue, previous learning-based methods attempt to finetune models but are limited by off-policy sampling and coarse-grained feedback. In this paper, we present \textit{\b{R}einforcement \b{L}earning \b{f}or \b{H}allucination} (RLFH), an on-policy self-alignment approach that enables LLMs to actively explore their knowledge boundaries and self-correct generation behavior through fine-grained feedback signals. RLFH introduces a self-assessment framework where the policy serves as its own judge. Through this framework, responses are automatically decomposed into atomic facts and their truthfulness and informativeness are assessed against external knowledge sources. The resulting fine-grained feedback at the statement level are then converted into token-level dense reward signals. This enables online reinforcement learning to achieve precise and timely optimization without human intervention. Comprehensive evaluations on HotpotQA, SQuADv2, and Biography benchmarks validate RLFH's effectiveness in hallucination mitigation.

For agents

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

get_harvested_code_for_paper("2406.12221")
get_code_for_paper("2406.12221")
have("2406.12221")

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

curl https://syntology.ai/api/ran/2406.12221.json

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

[![Syntology run record](https://syntology.ai/api/ran/2406.12221.svg)](https://syntology.ai/paper/2406.12221)

Connect an agent — have() is free.