SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2601.07894 · 2026

Revealing the Attention Floating Mechanism in Masked Diffusion Models

Shuo Wang, Maosong Sun, Zhenghao Liu, Yu Gu, Yukun Yan, Chaojun Xiao, Pengcheng Huang, Xin Dai

arXiv · PDF · Open in the Atlas

Code that ran

12 of 13 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; 3 honoured the contract we drafted, 1 violated it and 0 have no contract verdict; 8 failures 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/2601.07894.json · badge: /api/ran/2601.07894.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
NEUIR/Attention-Floating canonical (classifier label)
classifier basis: census
12 ran · 1 no recorded run
FunctionStatusWhere it lives
accuracy_score Ran
ran, honoured the contract we drafted
NEUIR/Attention-Floating/evaluate/evaluate_dream.py
pointer only (licence: NONE) · get_code("6a4180528ef351f2")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, NEUIR/Attention-Floating 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.

add_gumbel_noise Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
NEUIR/Attention-Floating/src/attention_extraction/Llada.py
pointer only (licence: NONE) · get_code("2340e6453f04571f")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, NEUIR/Attention-Floating 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.

attention_rollout_region Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
NEUIR/Attention-Floating/src/visualization/attention_rollout.py
pointer only (licence: NONE) · get_code("f1db726d53551dd6")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, NEUIR/Attention-Floating 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.

batch_analyze_ar Ran
ran; our contract draft was wrong, not the code
NEUIR/Attention-Floating/src/visualization/absorption_comparison.py
pointer only (licence: NONE) · get_code("38d374516cad7ce5")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, NEUIR/Attention-Floating 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.

batch_analyze_dream Ran
ran; our contract draft was wrong, not the code
NEUIR/Attention-Floating/src/visualization/absorption_comparison.py
pointer only (licence: NONE) · get_code("d4491256869f88bb")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, NEUIR/Attention-Floating 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.

batch_analyze_llada Ran
ran; our contract draft was wrong, not the code
NEUIR/Attention-Floating/src/visualization/absorption_comparison.py
pointer only (licence: NONE) · get_code("eb1895b1ae5f22aa")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, NEUIR/Attention-Floating 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.

compute_qk_cos_normprod Ran
ran, honoured the contract we drafted
NEUIR/Attention-Floating/src/visualization/QK_decomposition.py
pointer only (licence: NONE) · get_code("d230ff14a05ed8e6")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, NEUIR/Attention-Floating 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.

detect_model_type Ran
ran; our contract draft was wrong, not the code
NEUIR/Attention-Floating/src/visualization/QK_decomposition.py
pointer only (licence: NONE) · get_code("b86a2fefc81c48b6")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, NEUIR/Attention-Floating 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.

f1_score Ran
ran, honoured the contract we drafted
NEUIR/Attention-Floating/evaluate/evaluate_dream.py
pointer only (licence: NONE) · get_code("cb0727ddfdbb00d2")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, NEUIR/Attention-Floating 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.

find_sinks_from_attention Ran
ran, violated the contract we drafted
NEUIR/Attention-Floating/src/visualization/QK_decomposition.py
pointer only (licence: NONE) · get_code("6d8719bde8059d0a")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, NEUIR/Attention-Floating 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.

get_num_transfer_tokens Ran
ran; our contract draft was wrong, not the code
NEUIR/Attention-Floating/src/attention_extraction/Llada.py
pointer only (licence: NONE) · get_code("623e32f834b5c999")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, NEUIR/Attention-Floating 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.

normalize_answer Ran
ran; our contract draft was wrong, not the code
NEUIR/Attention-Floating/evaluate/evaluate_dream.py
pointer only (licence: NONE) · get_code("ff3b542378f5d19c")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, NEUIR/Attention-Floating 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.

visualize_rollout_attention Not yet run
unverified (harvested, no recorded run)
NEUIR/Attention-Floating/src/visualization/attention_rollout.py
pointer only (licence: NONE) · get_code("3fff7d663b106945")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, NEUIR/Attention-Floating 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 13 samples lifted from this paper's linked repositories, 13 carry a function identity and 0 do not. Of the 13 distinct functions with an identity, 3 are also present in other papers whose code we hold (3 with at least one paper dated earlier, 2 the same month, 3 later) and 10 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
normalize_answer 133 other papers
116 dated earlier, 3 the same month, 13 later, 1 undated · 1 only as an assembly brick
1508.06615 (2015-08)
Character-Aware Neural Language Models
1606.05250 (2016-06)
SQuAD: 100,000+ Questions for Machine Comprehension of Text
1608.07905 (2016-08)
Machine Comprehension Using Match-LSTM and Answer Pointer
and 130 more not named here
add_gumbel_noise 18 other papers
9 dated earlier, 1 the same month, 8 later · 3 only as an assembly brick
2502.09992 (2025-02)
Large Language Diffusion Models
2505.15809 (2025-05)
MMaDA: Multimodal Large Diffusion Language Models
2506.06295 (2025-06)
dLLM-Cache: Accelerating Diffusion Large Language Models with Adaptive Caching
and 15 more not named here
get_num_transfer_tokens 15 other papers
9 dated earlier, 6 later · 1 only as an assembly brick
2502.09992 (2025-02)
Large Language Diffusion Models
2505.15809 (2025-05)
MMaDA: Multimodal Large Diffusion Language Models
2506.06295 (2025-06)
dLLM-Cache: Accelerating Diffusion Large Language Models with Adaptive Caching
and 12 more not named here

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

Masked diffusion models (MDMs), which leverage bidirectional attention and a denoising process, are narrowing the performance gap with autoregressive models (ARMs). However, their internal attention mechanisms remain under-explored. This paper investigates the attention behaviors in MDMs, revealing the phenomenon of Attention Floating. Unlike ARMs, where attention converges to a fixed sink, MDMs exhibit dynamic, dispersed attention anchors that shift across denoising steps and layers. Further analysis reveals its Shallow Structure-Aware, Deep Content-Focused attention mechanism: shallow layers utilize floating tokens to build a global structural framework, while deeper layers allocate more capability toward capturing semantic content. Empirically, this distinctive attention pattern provides a mechanistic explanation for the strong in-context learning capabilities of MDMs, allowing them to double the performance compared to ARMs in knowledgeintensive tasks. All codes and datasets are available at https://github.com/NEUIR/Attention-Floating.

For agents

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

get_harvested_code_for_paper("2601.07894")
get_code_for_paper("2601.07894")
have("2601.07894")

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

curl https://syntology.ai/api/ran/2601.07894.json

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

[![Syntology run record](https://syntology.ai/api/ran/2601.07894.svg)](https://syntology.ai/paper/2601.07894)

Connect an agent — have() is free.