SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2604.06155 · 2026

Toward Consistent World Models with Multi-Token Prediction and Latent Semantic Enhancement

Wei Chen, Rui Mao, Mingyang Zhou, Hao Liao, Haiming Qin, Qimin Zhong, Naipeng Chao

arXiv · PDF · Open in the Atlas

Code that ran

5 of 6 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, 0 violated it and 4 have no contract verdict; 1 failure of our instrument; 1 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/2604.06155.json · badge: /api/ran/2604.06155.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
QiminZhong/LSE-MTP community
Papers with Code archive
5 ran · 1 no recorded run
FunctionStatusWhere it lives
Block Ran
ran on a synthesized input · constructed an object; no result was computed
QiminZhong/LSE-MTP/model.py
pointer only (licence: NONE) · get_code("9b480329d99cb4a9")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, QiminZhong/LSE-MTP when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

CausalSelfAttention Ran
ran on a synthesized input · constructed an object; no result was computed
QiminZhong/LSE-MTP/model.py
pointer only (licence: NONE) · get_code("783559fa04a5409b")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, QiminZhong/LSE-MTP when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

LayerNorm Ran
ran on a synthesized input · constructed an object; no result was computed
QiminZhong/LSE-MTP/model.py
pointer only (licence: NONE) · get_code("84e7fba866437fc7")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, QiminZhong/LSE-MTP when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

MLP Ran
ran on a synthesized input · constructed an object; no result was computed
QiminZhong/LSE-MTP/model.py
pointer only (licence: NONE) · get_code("47d551d04948384a")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, QiminZhong/LSE-MTP when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

new_gelu Ran
ran; our contract draft was wrong, not the code
QiminZhong/LSE-MTP/model.py
pointer only (licence: NONE) · get_code("d26a83b5dbd5eec6")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, QiminZhong/LSE-MTP when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

GPT Not yet run
unverified (harvested, no recorded run)
QiminZhong/LSE-MTP/model.py
pointer only (licence: NONE) · get_code("1efd08ec652fde62")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, QiminZhong/LSE-MTP when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. No invocation recorded.

Also present in other papers' code

Of 6 samples lifted from this paper's linked repositories, 6 carry a function identity and 0 do not. Of the 6 distinct functions with an identity, 4 are also present in other papers whose code we hold (4 with at least one paper dated earlier, 0 the same month, 0 later) and 2 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
LayerNorm
brick of this paper's assembly: helper
10 other papers
10 dated earlier · 6 only as an assembly brick
1706.03762 (2017-06)
Attention Is All You Need
2202.09244 (2022-02)
Transfer and Marginalize: Explaining Away Label Noise with Privileged Information
2309.07207 (2023-09)
EarthPT: a time series foundation model for Earth Observation
and 7 more not named here
new_gelu
brick of this paper's assembly: helper
9 other papers
9 dated earlier
2210.13319 (2022-10)
MARS: Meta-Learning as Score Matching in the Function Space
2307.03381 (2023-07)
Teaching Arithmetic to Small Transformers
2309.07207 (2023-09)
EarthPT: a time series foundation model for Earth Observation
and 6 more not named here
Block
brick of this paper's assembly: model
3 other papers
3 dated earlier
1706.03762 (2017-06)
Attention Is All You Need
2309.07207 (2023-09)
EarthPT: a time series foundation model for Earth Observation
2603.24298 (2026-03)
SpinGQE: A generative quantum eigensolver for spin Hamiltonians
MLP
brick of this paper's assembly: model
1 other paper
1 dated earlier
2309.07207 (2023-09)
EarthPT: a time series foundation model for Earth Observation

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

Whether Large Language Models (LLMs) develop coherent internal world models remains a core debate. While conventional Next-Token Prediction (NTP) focuses on one-step-ahead supervision, Multi-Token Prediction (MTP) has shown promise in learning more structured representations. In this work, we provide a theoretical perspective analyzing the gradient inductive bias of MTP, supported by empirical evidence, showing that MTP promotes the convergence toward internal belief states by inducing representational contractivity via gradient coupling. However, we reveal that standard MTP often suffers from structural hallucinations, where discrete token supervision encourages illegal shortcuts in latent space that violate environmental constraints. To address this, we propose a novel method Latent Semantic Enhancement MTP (LSE-MTP), which anchors predictions to ground-truth hidden state trajectories. Experiments on synthetic graphs and real-world Manhattan Taxi Ride show that LSE-MTP effectively bridges the gap between discrete tokens and continuous state representations, enhancing representation alignment, reducing structural hallucinations, and improving robustness to perturbations.

For agents

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

get_harvested_code_for_paper("2604.06155")
get_code_for_paper("2604.06155")
have("2604.06155")

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

curl https://syntology.ai/api/ran/2604.06155.json

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

[![Syntology run record](https://syntology.ai/api/ran/2604.06155.svg)](https://syntology.ai/paper/2604.06155)

Connect an agent — have() is free.