Factual Access States in Diffusion Language Models: From Mechanism to Knowledge Editing and Retrieval

Yuchen Guo; Zhuoyuan Mao; Yiwen Guo

cs.CLOct 20, 2026

Factual Access States in Diffusion Language Models: From Mechanism to Knowledge Editing and Retrieval

Authors: Yuchen Guo, Zhuoyuan Mao, Yiwen Guo

Organizations: The University of Tokyo · LIGHTSPEED · Independent Researcher

Abstract

Factual recall is a lookup: some representation inside the network serves as the key, and what that key is made of decides which queries can retrieve the fact, what an edit must cover, and what a retriever can read. We call it the factual access state. In autoregressive (AR) models the subject position enriches attributes largely independently of the relation, and relation-specific selection happens later, at the readout position. In masked diffusion language models (DLMs) we find the state is assembled differently, and much earlier: in the first attention layer the wording of the relation is routed into the position of whichever entity is visible and the early feed-forward layers look the fact up on this fused representation. The state therefore varies with the relation's surface form and with the query direction, so the queries about one fact occupy a region of state space rather than a point, and an operation on the fact succeeds only for queries that enter it. We treat the state as an interface to a DLM's factual knowledge, and put it to two uses. Writing to the state gives a knowledge editor: from one declarative statement, Multi-View Edit fits one shared value per direction across that region and installs it in the early-MLP block, preserving source-prompt reliability while substantially improving paraphrase and reverse question answering, where AR editors gain near zero. Reading it suggests a complementary use: the state at each entity position can serve as a relation-specific retrieval vector with no training, returning both pieces of evidence for two-entity queries on which pooled, prompted and lexical retrievers almost never return both.

Figures & tables

Figure 1: Factual access states. A masked DLM fuses an entity with the queried relation early; the resulting state supports knowledge editing and evidence retrieval.

Figure 2: Window-tracing on a masked DLM (LLaDA-8B). Sliding-window causal tracing with $w{=}10$ , showing the indirect effect across layers at subject, relation, and other token positions. Left: full block; middle: attention only; right: FF/MLP only. The subject signal is bimodal and MLP-dominated.

Figure 3: Single-layer causal effect. Per-layer IE inside the early ( $0$ – $8$ ) and late ( $18$ – $25$ ) peaks, split by attention vs. MLP. Only layer- $0$ attention stands out; no single MLP layer exceeds $\sim\!0.03$ .

Layer- $0$ restriction on subject tokens	IE (mean $\pm$ sem) $\scriptstyle\uparrow$	% of full L0 attn $\scriptstyle\uparrow$
sink removed ( sink_off )	$+0.187\pm 0.047$	$79\%$
subject $\to$ relation only	$\mathbf{+0.199\pm 0.044}$	$\mathbf{84\%}$
subject $\to$ subject-internal only	$+0.049\pm 0.035$	$21\%$
subject $\to$ self only	$+0.008\pm 0.010$	$3\%$
subject $\to$ [MASK] only	$-0.008\pm 0.010$	$0\%$

Table 1: Layer-

0

attention surgery on LLaDA-8B. Each row keeps only the listed class of edges out of the subject tokens; % is normalised by the full layer-

0

attention IE of

0.238

Model	Method	Rel. $\scriptstyle\uparrow$	Para. $\scriptstyle\uparrow$	Neigh. $\scriptstyle\uparrow$	Loc-Pres. $\scriptstyle\uparrow$
Dream-7B	FT+L	99.67	86.47	18.95	31.33
	Single-View	94.83	72.05	74.48	89.91
	Multi-View	99.88	99.24	73.04	88.29
LLaDA-8B	FT+L	99.27	51.94	47.28	65.04
	Single-View	96.87	53.65	57.13	68.31
	Multi-View	98.14	67.92	73.73	89.92

Table 2: CounterFact. ROME-based rows use a last-subject-token write; FT+L is a constrained fine-tuning baseline. Single-View uses

\mathcal{V}_{\mathrm{obj}}\cup\mathcal{V}_{\mathrm{ctx}}

; Multi-View adds

\mathcal{V}_{\mathrm{rel}}

. Published AR references are collected in App. D.2 .

Model	Method	Rel. $\scriptstyle\uparrow$	Para. $\scriptstyle\uparrow$	R-QA $\scriptstyle\uparrow$	R-Judge $\scriptstyle\uparrow$	$\bm{\Delta}$ (R-QA / R-Judge) $\scriptstyle\uparrow$
LLaDA-8B	no edit (baseline)	–	–	22.35	30.59	$0.00\,/\,0.00$
	FT+L	74.76	45.73	19.94	28.79	$-2.41$ / $-1.80$
	Single-Forward	95.53	73.03	22.52	38.88	$+0.17$ / $+8.29$
	Single-Binary	97.04	65.48	42.03	41.03	$+19.68$ / $+10.44$
	Multi-Binary	98.49	84.56	63.24	55.60	$\mathbf{+40.89}$ / $\mathbf{+25.01}$
Dream-7B	no edit (baseline)	–	–	20.26	18.15	$0.00\,/\,0.00$

Table 3: BAKE. R-QA and R-Judge are reverse probes;

\Delta

is measured from each DLM’s no-edit baseline. Single/Multi denotes one view versus the masked family; Forward/Binary denotes whether the explicit reverse group is installed. FT+L is a constrained fine-tuning baseline. Published AR references are collected in App. D.2 .

		Rel.			Gen.
Model	Method	$=\!1.0$ $\scriptstyle\uparrow$	$\geq\!0.5$ $\scriptstyle\uparrow$	Weighted $s$ $\scriptstyle\uparrow$	$=\!1.0$ $\scriptstyle\uparrow$	$\geq\!0.5$ $\scriptstyle\uparrow$	Weighted $s$ $\scriptstyle\uparrow$
LLaDA-8B	FT+L	15.0%	26.3%	0.206	11.9%	22.9%	0.174
	Single-View	14.9%	26.2%	0.205	11.7%	23.3%	0.175
	Multi-View	17.6%	32.5%	0.250	12.6%	25.5%	0.187
Dream-7B	FT+L	16.2%	32.4%	0.243	13.4%	26.8%	0.201
	Single-View	20.1%	38.1%	0.291	16.5%	33.1%	0.248

Table 4: QAEdit on pre-edit-wrong examples. Fraction of answers scored exact (

=1.0

) or at least partially correct (

\geq 0.5

), and weighted score

s=(n_{1.0}+0.5n_{0.5})/n

. Rel. denotes source prompts and Gen. paraphrased questions. Target-length results are in App. D.6 .

Figure 4: Multi-View trigger and editing effect. Left: mean trigger strength $\alpha_{\text{fire}}$ ; right: mean actual $\Delta\text{logit}$ . M1 = forward single-view, M2 = forward $+$ reverse single-view, M3 = full Multi-View (ours). LLaDA-8B. Query difficulty increases from Paraphrase (left) to Reverse-Judge (right).

Shared baselines
	Global corpus			Entity-conditioned			Time $\scriptstyle\downarrow$
Method	Single	Adj.	Cross.	Single	Adj.	Cross.	Global	Entity
	R@1 $\scriptstyle\uparrow$	All@2 $\scriptstyle\uparrow$	All@2 $\scriptstyle\uparrow$	R@1 $\scriptstyle\uparrow$	All@2 $\scriptstyle\uparrow$	All@2 $\scriptstyle\uparrow$	ms/query	ms/query
BM25	92.82	1.38	1.01	95.35	11.83	11.83	0.51	0.31
Qwen2.5 pooling	69.06	1.28	0.00	81.19	8.83	8.14	12.76	0.43
PromptReps (20/20), hybrid	92.53	1.58	1.56	93.47	16.35	16.26	264.98	263.20

Table 5: Controlled benchmark: global search and entity-conditioned retrieval (%). All methods search the same

5{,}528

cards for the same

10{,}800

test queries. Single is R@1; Adj./Cross. are dual-target All@2. Time is measured in ms/query. Shaded cells mark the best value in each column.

Appendix figures & tables24 assets

Supplementary material from the paper’s appendix.

Appendix

Setting	Backbone	Update mechanism	Variation explicitly covered	Reverse edit
ROME / MEMIT	AR	Backbone MLP weights	Context templates around one forward conditional	Not native
BIRD / BAKE	AR	Bidirectional objectives / evaluation	Forward–reverse objectives under AR factorisation	Explicit
Park & Jo	Masked DLM	Early–middle MLP weights	Partially unmasked target states during decoding	Not central
TimeROME-DLM	Masked DLM	Inference-time residual memory	Diffusion-time / temporal retrieval states	Not central
Multi-View Edit	Masked DLM	Early-MLP down-projections	Relation form, bidirectional context, and forward–reverse views from one statement	Explicit

Appendix

Table 6: Positioning of closely related editing settings. Existing masked-DLM editors primarily address robustness across diffusion states or introduce inference-time memories. Multi-View Edit instead expands one declarative statement into a family of conditional retrieval views.

Figure 5: Early-block decomposition (referenced in § 2.2 ). IE recovered by restoring different subsets of the early block. Layer- $0$ attention combined with MLPs $1$ – $8$ ( $0.943$ ) matches the full early block ( $0.936$ ), while either sub-component alone falls short.

Figure 6: Selective peak ablation with channel decomposition ( $n{=}550$ ). Restoring the early peak alone already recovers most of the causal effect on the MLP channel ( $0.917$ vs. $0.967$ for all layers), while restoring the late peak alone leaves the attention channel almost silent ( $0.073$ ).

Figure 7: Cross-peak conditional patching ( $n{=}550$ ). Late $+$ Block Early preserves the late peak’s IE ( $1.012\to 1.012$ on Full Block), whereas Early $+$ Block Late destroys the early peak’s effect ( $0.946\to 0.019$ ). The asymmetry is strongest on the MLP channel ( $0.917\to 0.148$ ).

Figure 8: Subject position (Exp 3) vs. [MASK] position (Exp 7), $n{=}194$ . Left: full-block IE per layer. Middle / Right: attention-only and MLP-only splits. The two positions are almost mutually exclusive as causal loci: subject-side dominance gives way to [MASK] -side dominance only in the last few layers.

Figure 9: Dream-7B 1-D causal trace, $n{=}432$ . Left: hidden states. Middle: attention only. Right: FF/MLP only. Four curves correspond to the four role-typed positions. Dream-7B reproduces every qualitative feature reported on LLaDA-8B: a subject-anchored early peak in Hidden and MLP, a very large low-layer spike on the Attention channel at the subject positions (layers $0$ – $5$ , $\mathrm{AIE}\!\approx\!0.25$ ), a Relation-side attention peak around layers $18$ – $23$ , and a [MASK] -side takeover in the top layers.

Figure 10: LLaDA-8B subject-side 1-D causal trace, $n{=}156$ . Setting: the object is kept clean; the subject span is replaced by [MASK_subj] . Curves show AIE at each role-typed position across layers. The Object position now plays the causal role that the subject played in the tail-mask setting: it dominates Hidden / MLP in the early-to-mid layers and its attention spikes at the lowest layers. The [MASK_subj] position takes over in the top layers, exactly mirroring the [MASK] takeover of App. B.3 .

Category	Parameter	Value
$z^{*}$ optimisation	Optimiser	Adam
	Learning rate	$5\times 10^{-2}$
	Max steps	100 (early stop at loss $<0.09$ )
	Weight decay $v_{\text{wd}}$	$0.5$
	Clamp factor $\kappa$	$3.0$ ( $\\|\delta\\|\leq\kappa\\|z_{0}\\|$ )
Weight update	Cov. regularisation $\lambda$	$10^{-5}$

Appendix

Table 7: Editing hyperparameters.

Prefix templates (prepended before prompt):
#	Template (prefix \|\| suffix)
1	"" (identity, no wrapper)
2	"In one context, "
3	"According to records, "
4	"In this case, "
5	"For reference, "

Appendix

Table 8: Context templates. Each prompt is wrapped as [prefix] + prompt + [MASK] + [suffix] . Suffix templates exploit the DLM’s bidirectional attention to condition on context after the prediction slot.

		CounterFact			BAKE
Model	Method	Rel. $\scriptstyle\uparrow$	Para. $\scriptstyle\uparrow$	Neigh. $\scriptstyle\uparrow$	Rel. $\scriptstyle\uparrow$	Para. $\scriptstyle\uparrow$	R-QA $\scriptstyle\uparrow$	R-Judge $\scriptstyle\uparrow$
GPT-J (6B)	ROME	100.00	97.95	81.46	98.01	94.60	2.80	33.89
	MEMIT	100.00	97.10	81.96	96.21	86.99	0.16	13.47
	BIRD	–	–	–	99.33	96.00	2.90	34.96
LLaMA-2 (7B)	ROME	99.68	98.10	77.59	99.70	94.74	0.26	40.78
	MEMIT	99.58	97.99	77.80	99.85	96.92	0.14	30.10

Appendix

Table 9: Published autoregressive references. CounterFact and BAKE results reported by the original papers for GPT-J (6B) and LLaMA-2 (7B). BAKE reports a pre-edit score of

0.00

on both reverse probes for every base LLM, so the published

\Delta

equals the post-edit value. BIRD does not report CounterFact numbers.

Split	Position	Rel. $\scriptstyle\uparrow$	Para. $\scriptstyle\uparrow$	R-QA $\scriptstyle\uparrow$	R-Judge $\scriptstyle\uparrow$
Development ( $n{=}4694$ )	first	97.23	67.22	60.94	52.53
	mean	98.27	73.63	61.02	50.79
	last	98.49	84.56	63.24	55.60
	spread	1.26	17.34	2.30	4.81
Held-out ( $n{=}1000$ )	first	97.50	67.60	58.60	56.60
	mean	97.90	75.40	58.60	55.90

Appendix

Table 10: Subject-position sweep on LLaDA-8B (BAKE, ROME base editor). first / mean / last write the update at the first subject token, at the mean over the subject span, and at the last subject token respectively. The top block is the development pool on which the position was selected; the bottom block is a disjoint held-out split evaluated once after the choice was fixed. Reliability and both reverse probes are insensitive to the choice on both splits; Paraphrase is not. last is best on all four columns on both splits, and the three-way ordering is identical column by column. The last setting is the configuration reported throughout the main text.

Variant	#Views	Rel. $\scriptstyle\uparrow$	Para. $\scriptstyle\uparrow$	R-QA $\scriptstyle\uparrow$	R-Judge $\scriptstyle\uparrow$
Single-Binary	14	97.04	65.48	42.03	41.03
Multi-Binary (matched)	14	97.81	82.24	63.81	55.92
Multi-Binary	20	98.49	84.56	63.24	55.60

Appendix

Table 11: Matched view budget on BAKE (LLaDA-8B). #Views counts the supervised views per direction. Multi-Binary (matched) uses the relation-masked and reverse construction with six context views removed, giving it the same budget as Single-Binary . All other components are held fixed.

Variant	#Views	Rel. $\scriptstyle\uparrow$	Para. $\scriptstyle\uparrow$	Neigh. $\scriptstyle\uparrow$	Loc-Pres. $\scriptstyle\uparrow$
Single-View	14	96.87	53.65	57.13	68.31
Multi-View (matched)	14	98.03	66.85	71.05	89.64
Multi-View	20	98.14	67.92	73.73	89.92

Appendix

Table 12: Matched view budget on CounterFact (LLaDA-8B). Same three configurations as Table 11 , evaluated on the CounterFact edit pool.

Model	Variant	Rel. $\scriptstyle\uparrow$	Para. $\scriptstyle\uparrow$	Neigh. $\scriptstyle\uparrow$	Loc-Pres. $\scriptstyle\uparrow$
Dream-7B	Multi-View	99.88	99.24	73.04	88.29
Dream-7B	Multi-View $+$ KL	98.27	92.47	74.89	89.21
LLaDA-8B	Multi-View	98.14	67.92	73.73	89.92
LLaDA-8B	Multi-View $+$ KL	96.83	66.38	74.03	90.18

Appendix

Table 13: KL regularisation ablation on CounterFact. All rows use the ROME base editor.

Whitening statistics	Rel. $\scriptstyle\uparrow$	Para. $\scriptstyle\uparrow$	Neigh. $\scriptstyle\uparrow$	Loc-Pres. $\scriptstyle\uparrow$
No whitening	97.20	59.55	70.94	83.41
Clean-text covariance	97.20	65.32	72.85	87.96
Random-mask covariance (ours)	98.14	67.92	73.73	89.92

Appendix

Table 14: Whitening covariance ablation on CounterFact (LLaDA-8B). The model, base editor, data split, and all non-covariance hyperparameters are held fixed.

Model	Context views	Rel. $\scriptstyle\uparrow$	Para. $\scriptstyle\uparrow$	Neigh. $\scriptstyle\uparrow$	Loc-Pres. $\scriptstyle\uparrow$
LLaDA-8B	None	91.60	38.20	76.08	92.48
	Prefix	95.40	49.40	75.10	91.90
	Prefix $+$ suffix	98.14	67.92	73.73	89.92
Dream-7B	None	84.72	68.20	76.44	90.50
	Prefix	96.98	85.40	74.58	89.84
	Prefix $+$ suffix	99.88	99.24	73.04	88.29

Appendix

Table 15: Prefix/suffix context ablation on CounterFact. None uses no contextual templates, Prefix uses only templates before the prompt, and Prefix

+

suffix combines prefix and suffix templates; the last row of each block is the Multi-View configuration of Table 2 . All other components of the editor are held fixed. Bold indicates the best result within each model and metric; note that Neighborhood and Loc-Pres. are best without contextual views, i.e. the gains in Rel. and Para. are traded against a small locality cost.

Figure 11: ROME sequential-edit stability. Single-Binary and Multi-Binary are applied cumulatively without restoring the model weights. Scores are evaluated after

T\in\{1,10,25,50,100\}

sequential edits.

Target length	$n$	$=\!1.0$ $\scriptstyle\uparrow$	$=\!0.5$	$\geq\!0.5$ $\scriptstyle\uparrow$	Weighted $s$ $\scriptstyle\uparrow$
1 word	677	21.3%	16.8%	38.1%	0.297
2 words	820	16.2%	12.4%	28.7%	0.224
3 words	437	16.2%	15.3%	31.6%	0.239
4 words	176	18.2%	17.0%	35.2%	0.267
$\geq\!5$	162	11.7%	16.0%	27.8%	0.198

Appendix

Table 16: QAEdit LLaDA-8B breakdown by target length (source prompt, Multi-View ).

Method	Split	$=\!1.0$ $\scriptstyle\uparrow$	$=\!0.5$	$=\!0.0$ $\scriptstyle\downarrow$	$\geq\!0.5$ $\scriptstyle\uparrow$	Weighted $s$ $\scriptstyle\uparrow$
Multi-View	Rel.	17.6%	14.9%	67.5%	32.5%	0.250
Multi-View	Gen.	12.6%	12.9%	74.5%	25.5%	0.187
Multi-View $+$ partial reveal	Rel.	35.0%	16.1%	48.9%	51.1%	0.430
Multi-View $+$ partial reveal	Gen.	26.7%	17.0%	56.3%	43.7%	0.352

Appendix

Table 17: Composing decoding-state views with the multi-view family on QAEdit (LLaDA-8B, pre-edit-wrong examples). The Multi-View rows repeat Table 4 .

+

partial reveal adds the partially revealed target states of Park and Jo (2026) to

\mathcal{V}

as an additional view type; all other components of the editor are held fixed.

Target length	$=\!1.0$ $\scriptstyle\uparrow$	$=\!0.5$	$\geq\!0.5$ $\scriptstyle\uparrow$	Weighted $s$ $\scriptstyle\uparrow$
1 word	34.3%	18.2%	52.5%	0.434
2 words	25.9%	12.0%	38.0%	0.319
3 words	21.3%	21.3%	42.7%	0.320
4 words	21.7%	17.4%	39.1%	0.304
$\geq\!5$	20.0%	20.0%	40.0%	0.300

Appendix

Table 18: QAEdit LLaDA-8B breakdown by target length for the composed editor (source prompt, Multi-View

+

partial reveal). Table 16 reports the same breakdown for Multi-View alone.

Domain	Relations	Entities
Film production	director, composer, screenwriter, producer	590
Person attributes	occupation, citizenship, birthplace, language	351
Sports people	sport, team, playing position, citizenship	236
Family relations	father, mother, spouse, sibling	116
Written works	author, genre, publisher, language	89

Appendix

Table 19: Domains of the controlled benchmark.

Shared baselines
	HotpotQA ( $n=1{,}296$ )					2WikiMultiHopQA ( $n=2{,}933$ )
Method	Sup. R@2 $\uparrow$	Both@2 $\uparrow$	EM $\uparrow$	F1 $\uparrow$	ms $\downarrow$	Sup. R@2 $\uparrow$	Both@2 $\uparrow$	EM $\uparrow$	F1 $\uparrow$	ms $\downarrow$
BM25	52.10	25.15	52.31	61.23	0.38	56.50	26.93	61.61	62.18	0.38
Qwen2.5 pooling	18.46	3.32	45.06	53.19	31.43	14.31	1.94	38.56	43.20	31.44
PromptReps (20/20), hybrid	41.86	11.34	49.00	56.94	456.18	44.85	13.16	55.74	56.33	456.94
Dream-7B-Instruct

Appendix

Table 20: Natural comparison-question retrieval and end-to-end RAG. Every retriever receives the same per-question distractor corpus and a total budget of two sentences; all retrieved evidence goes to the same frozen Qwen2.5-7B-Instruct reader. Accuracy in percent; time in ms per query, from question to retrieved evidence, excluding answer generation. Shaded cells mark the best non-oracle method in each column.

Figure 12: Relation messages support early evidence selection. Query-side layer-

0

interventions on Dream with a fixed document index. Bars: within-entity relation R@1 averaged over targets; diamonds: global entity identification; error bars: entity/pair-clustered

95\%

confidence intervals. C: clean; R: relation messages removed; M: mass-matched random messages (three-seed mean); N: norm-matched random messages.

Query-side L0 intervention	Single	Adjacent	Crossed
Clean	54.31	59.11	42.53
Remove relation messages	27.07	25.26	25.35
Native random removal, mean of three	54.73	59.07	42.39
Exact-mass random message, mean of three	57.61	59.51	41.97
Norm-matched random message	54.37	58.56	41.81
Clean relation add-back	54.31	59.11	42.53

Appendix

Table 21: Causal readout results: within-entity relation R@1 (%). The first nine rows read the L0 residual. Random-seed rows are averaged only for display; all thirteen conditions are stored individually. Former/latter removal is a no-op in layouts without these tokens.

Factual Knowledge in Language Models

Factual knowledge in language models concerns what facts they store, how reliably they retrieve them, and how that knowledge can be tested or changed.

See all 193 papers

Knowledge Editing for Masked Diffusion Language Models
Jun 2, 2026Haewon Park, Yohan JoKnowledge Editing Masked Diffusion Models
Graduate School of Data Science, Seoul National University
DiffRetriever: Parallel Representative Tokens for Retrieval with Diffusion Language Models
May 8, 2026Shuai Wang, Yu Yin, Shengyao Zhuang +2Multi-Vector Retrieval Diffusion Language Models
The University of Queensland, Brisbane, QLD, Australia · CSIRO
Self-Augmenting Retrieval for Diffusion Language Models
Jun 4, 2026Paul Jünger, Justin Lovelace, Linxi Zhao +2Adaptive RAG Retrieval-Augmented Generation
Department of Computer Science, Cornell University.

Jun 2, 2026J/K move · Enter open · S save

Knowledge Editing for Masked Diffusion Language Models

Haewon Park, Yohan Jo

Graduate School of Data Science, Seoul National University

Knowledge Editing Masked Diffusion Models Masked Language Modeling Model Editing Diffusion Language Models Language Model Editing

Open paper AskPDF arXiv

Knowledge editing aims to update or correct factual knowledge in a language model. A widely used approach, locate-then-edit, first localizes a fact within the model and then edits the weights there. To date, such methods have been developed exclusively for autoregressive models (ARMs). Whether they work for masked diffusion models (MDMs), which model text bidirectionally and generate by iterative denoising rather than next-token prediction, remains an open question. We address it by transferring locate-then-edit to MDMs and comparing multiple MDMs with their matched ARMs. Our central finding has two parts. First, where an edit should be applied transfers between them: the same early-to-mid-layer MLP at the last subject token is most effective for both. Second, this shared location does not guarantee a shared outcome. Single-token edits succeed in both, but as targets grow longer, editing degrades far more sharply in the MDMs than in the ARMs. The failure stems from how the edited fact is generated: producing a multi-token target passes through intermediate states in which the target is partially unmasked, for which the edit was never optimized. Guided by this diagnosis, we introduce a simple correction that optimizes the edit including such states, substantially restoring multi-token performance. Our code is available at https://github.com/holi-lab/MDM-KE.

Abstract

Factual Knowledge in Language Models

Explore similar work

Knowledge Editing for Masked Diffusion Language Models

DiffRetriever: Parallel Representative Tokens for Retrieval with Diffusion Language Models

Self-Augmenting Retrieval for Diffusion Language Models