Retrieval as Generation: A Unified Framework with Self-Triggered Information Planning

Bo Li; Mingda Wang; Gexiang Fang; Shikun Zhang; Wei Ye

cs.CLApr 13, 2026

Retrieval as Generation: A Unified Framework with Self-Triggered Information Planning

Authors: Bo Li, Mingda Wang, Gexiang Fang, Shikun Zhang, Wei Ye

Organizations: National Engineering Research Center for Software Engineering, Peking University · School of Computer Science, Peking University · School of Health Sciences and Biomedical Engineering, Hebei University of Technology

Abstract

We revisit retrieval-augmented generation (RAG) by embedding retrieval control directly into generation. Instead of treating retrieval as an external intervention, we express retrieval decisions within token-level decoding, enabling end-to-end coordination without additional controllers or classifiers. Under the paradigm of Retrieval as Generation, we propose GRIP (Generation-guided Retrieval with Information Planning), a unified framework in which the model regulates retrieval behavior through control-token emission. Central to GRIP is Self-Triggered Information Planning, which allows the model to decide when to retrieve, how to reformulate queries, and when to terminate, all within a single autoregressive trajectory. This design tightly couples retrieval and reasoning and supports dynamic multi-step inference with on-the-fly evidence integration. To supervise these behaviors, we construct a structured training set covering answerable, partially answerable, and multi-hop queries, each aligned with specific token patterns. Experiments on five QA benchmarks show that GRIP surpasses strong RAG baselines and is competitive with GPT-4o while using substantially fewer parameters.

Figures & tables

Figure 1: Overview of the GRIP framework. The top half illustrates the training process: four behavior types are constructed for supervised fine-tuning, followed by reinforcement learning to improve answer quality and retrieval control. The bottom half depicts self-triggered information planning, where GRIP dynamically emits control tokens to decide whether to retrieve, generate new queries, or terminate reasoning, all within a unified decoding trajectory.

Figure 2: Four types of structured training samples in GRIP, each aligned with specific token patterns for controlling retrieval behavior. This design enables the model to learn when to retrieve, plan new queries, or finalize answers via language-native supervision.

	HotpotQA			PopQA			NQ			WebQ			TriviaQA			Avg. Score
	EM	ROUGE	F1	EM	ROUGE	F1	EM	ROUGE	F1	EM	ROUGE	F1	EM	ROUGE	F1	Avg. Score
Training-free Method
Instruct	17.2	21.6	25.9	17.4	19.6	23.2	14.4	23.8	20.3	14.8	25.1	29.4	46.2	45.3	55.1	26.6
GPT-3.5 Turbo	26.2	32.4	38.2	29.1	31.1	35.3	20.7	34.3	30.0	15.9	27.4	31.9	55.6	57.8	69.7	35.7
GPT-4o	33.2	40.2	47.0	30.6	32.3	39.9	26.5	42.7	28.3	23.5	32.3	37.0	65.7	64.3	78.2	41.4
Single RAG	26.1	31.6	37.2	22.8	26.6	31.1	19.3	28.4	24.8	14.0	22.7	26.6	46.4	47.0	56.8	30.8

Table 1: Main results on five QA benchmarks. We report EM, ROUGE (average of ROUGE-1/2/L), and F1; Avg.Score is the unweighted mean over all datasets and metrics. All open-source baselines are reproduced using official implementations and released checkpoints under the same evaluation setting, and results are averaged over three random seeds. Results on a domain-specific dataset are provided in Appendix B .

	NQ (54.3%)			WebQ (24.3%)			Avg. Score
	EM	ROUGE	F1	EM	ROUGE	F1	Avg. Score
DRAGIN	30.3	33.9	39.3	42.5	48.4	52.9	41.2
GainRAG	31.6	34.4	39.6	33.0	40.7	45.6	37.4
GRIP	31.1	34.0	39.8	44.9	50.7	55.6	42.7

Table 2: Subset where DRAGIN/GainRAG retrieve at least once while GRIP answers without retrieval. Percentages denote the subset proportion per dataset.

	NQ (23.1%)			WebQ (43.8%)			Avg. Score
	EM	ROUGE	F1	EM	ROUGE	F1	Avg. Score
DRAGIN	11.4	16.7	19.0	18.6	23.8	28.3	19.6
GainRAG	8.8	14.8	17.4	11.1	18.4	22.4	15.5
GRIP	15.4	20.3	23.4	19.9	24.6	28.9	22.1

Table 3: Subset where GRIP performs two retrievals while DRAGIN/GainRAG retrieve at most once. Percentages denote the subset proportion per dataset.

	Hotpot	PopQA	NQ	WebQ	Trivia	Avg. Count
DRAGIN	1.13	1.26	1.06	1.02	1.08	1.11
GainRAG	1.0	1.0	1.0	1.0	1.0	1.0
R1-Searcher	5.75	5.34	5.09	4.67	4.76	5.12
GRIP	1.44	1.58	0.76	1.15	1.25	1.24
w/o RL	1.99	1.98	0.77	1.26	1.99	1.60

Table 4: Mean retrieval count per dataset. GRIP adapts retrieval frequency across tasks, and RL further reduces retrieval calls while preserving this adaptivity. Hotpot and Trivia denote HotpotQA and TriviaQA, respectively.

Figure 3: The percentage of samples where the gold answer span is found in the top-1 or top-3 retrieved passages. GRIP-generated queries significantly improve answer presence in retrieved contexts across both NQ and WebQ, indicating more effective retrieval behavior.

Max $B$	Avg.Count (per dataset)						Avg. Score
Max $B$	Hotpot	PopQA	NQ	WebQ	Trivia	Avg	Avg. Score
3	1.44	1.58	0.76	1.15	1.25	1.24	41.0
5	1.56	1.73	1.01	1.45	1.31	1.41	41.2
7	1.66	1.79	1.12	1.65	1.35	1.51	41.5
10	1.74	1.86	1.22	1.90	1.39	1.62	41.8

Table 5: Effect of the maximum retrieval budget

B

at inference time. Hotpot and Trivia denote HotpotQA and TriviaQA, respectively.

Teacher	Training	Avg.Score
GPT-4o-mini	GRIP (SFT)	40.7
GPT-4o-mini	GRIP (SFT+RL)	41.0
LLaMA3-8B-Instruct	GRIP (SFT)	38.8
LLaMA3-8B-Instruct	GRIP (SFT+RL)	39.0
Qwen3-32B	GRIP (SFT)	40.2
Qwen3-32B	GRIP (SFT+RL)	40.3

Table 6: Teacher replacement results, where only the GPT-labeled subset is re-annotated.

Model	MMLU (Acc)	MBPP (Pass@1)
Instruct	66.56	54.6
SFT-RAG	62.93	47.8
GRIP	65.73	53.8

Table 7: General capability evaluation on closed-form non-RAG benchmarks.

Pairwise comparison (GRIP vs.)	Win	Equal	Loss
Instruct	51.0	37.5	11.5
SFT-RAG	90.5	8.5	1.0
R1-Searcher	59.5	32.5	8.0

Table 8: Pairwise GPT-4o evaluation on CNN/DailyMail summarization. Results are averaged over both output orders to mitigate position bias.

Appendix figures & tables18 assets

Supplementary material from the paper’s appendix.

Appendix

	#Samples	Type
HotpotQA	7,405	Multi-hop
PopQA	14,267	Open-domain
NQ	3,610	Open-domain
WebQ	2,032	Open-domain
TriviaQA	11,313	Reading Comprehension
BioASQ	885	Domain-specific

Appendix

Table 9: Statistics and types of the test datasets used in our evaluation. The datasets vary in size and reasoning focus, covering multi-hop, open-domain, reading comprehension, and biomedical QA tasks.

	ROUGE	F1	Avg.Score
FLARE	30.2	69.4	49.8
DRAGIN	47.1	81.2	64.2
SFT-RAG	39.1	58.6	48.9
Self-RAG	45.9	68.8	57.4
INFO-RAG	39.4	59.2	49.3
RobustRAG	53.9	80.9	67.4

Appendix

Table 10: Performance on the domain-specific BioASQ benchmark (higher is better). Avg.Score is the unweighted mean of ROUGE and F1.

	Hotpot	PopQA	NQ	WebQ	Trivia	Avg. CoverEM
GPT-4o	38.6	45.2	48.5	43.6	82.1	51.6
DRAGIN	31.6	23.0	38.4	41.5	68.0	40.5
GainRAG	38.0	44.4	39.1	36.9	64.2	44.5
GRIP	45.0	49.1	41.2	52.4	74.9	52.5

Appendix

Table 11: CoverEM scores across five QA datasets. GRIP achieves the best average CoverEM. Hotpot and Trivia denote HotpotQA and TriviaQA, respectively.

Figure 4: Type- $\alpha$ : Samples that are directly answerable using the model’s internal knowledge alone, where the model emits [ANSWER] followed by the answer and concludes with [SOLVED] without any retrieval.

Figure 5: Type- $\beta$ : Samples where the model produces partial or noisy answers containing the gold answer but lacking clarity, encouraging it to recognize uncertainty and issue a retrieval request through [INTERMEDIARY] followed by [RETRIEVE] .

Figure 6: Type- $\gamma$ : Multi-hop or complex knowledge samples where both parametric memory and basic retrieval fail, requiring the model to iteratively construct better sub-questions and reason across multiple planning steps before concluding with [ANSWER] and [SOLVED] .

Figure 7: Type- $\theta$ : Cases where the retrieved content covers the answer superficially but demands further reasoning and content compression, training the model to selectively synthesize key information via [INTERMEDIARY] $\rightarrow$ [RETRIEVE] $\rightarrow$ [ANSWER] transitions.

Reward metric for $r_{\text{ans}}$	Avg.Score	Avg.Count
BLEU (default)	41.0	1.24
ROUGE	40.5	1.24
F1	40.8	1.25
EM	40.6	1.24

Appendix

Table 12: Reward-metric ablation for RL: we only change the overlap metric used in the answer-fidelity reward

r_{\text{ans}}

; all other settings are fixed.

Figure 8: Comparison of retrieval behavior distributions before and after reinforcement learning on NQ and WebQ.

Figure 9: Example of GRIP falling back to internal knowledge and deferring retrieval when no context is available.

Figure 10: Example of GRIP refining its retrieval query over multiple rounds to obtain a better answer.

Figure 11: Example of GRIP refining, reformulating, and compressing retrieved knowledge into a final answer using [ANSWER] .

Figure 12: PCA visualization of the model’s hidden states before emitting [ANSWER] (blue) and [RETRIEVE] (orange) on the NQ dataset. Each point corresponds to the hidden representation of the last token in the input prompt. The distinct clusters indicate that the model encodes separable internal states for different control decisions.

Figure 13: Prompt used for data generation with GPT-4o-mini.

Method	BM25	DPR	Hybrid
Single-RAG (training-free)	30.8	31.9	33.7
RobustRAG (training-based)	33.2	32.1	33.8
GRIP	41.0	39.2	41.0

Appendix

Table 13: Retriever ablation on the main evaluation suite. We replace the default BM25 retriever with a dense DPR retriever and a BM25+DPR hybrid retriever (denoted Hybrid ), while keeping the backbone (LLaMA3-8B), corpus (Wikipedia), and top-

k

passages (

k{=}3

) fixed. Single-RAG and RobustRAG are representative training-free and training-based baselines from the main table, respectively.

Figure 14: Realized retrieval-count distributions of GRIP under different maximum retrieval budgets

B

. Each subplot corresponds to one benchmark; curves show the fraction of examples that trigger

r

retrieval calls at inference time. The dashed vertical line marks

r{=}3

, the maximum retrieval depth observed in our training trajectories. As

B

increases, GRIP places more probability mass beyond

r{>}3

, indicating extrapolation beyond the training depth, while the distribution remains concentrated at small

r

, suggesting selective and task-aware retrieval rather than mechanically consuming the available budget.

Component	Status	Key settings
CLIP-HIGHER	on	clip_ratio_low=0.2, clip_ratio_high=0.28, clip_ratio_c=10.0
Token-level PG loss	on	loss_agg_mode= token-mean
Dynamic sampling	on	enabled (default DAPO setting)
Soft overlong shaping	on	overlong_buffer.len=8192, penalty_factor=1.0
KL in reward	off	use_kl_in_reward=False, kl_coef=0.0
KL loss	off	use_kl_loss=False, kl_loss_coef=0.0

Appendix

Table 14: Key DAPO components used in our RL stage. All unspecified hyperparameters follow the referenced DAPO recipe.

Qwen-3-4B-Instruct
	HotpotQA			PopQA			NQ			WebQ			TriviaQA			Avg. Score
	EM	ROUGE	F1	EM	ROUGE	F1	EM	ROUGE	F1	EM	ROUGE	F1	EM	ROUGE	F1	Avg. Score
Instruct	8.0	13.6	16.5	5.3	8.0	10.3	6.2	11.3	13.7	7.7	16.3	20.0	22.1	24.1	30.1	14.2
Single-RAG	18.4	24.6	28.8	18.3	21.7	25.5	12.9	18.9	21.9	8.7	16.7	20.2	36.0	38.0	46.0	23.8
Robust-RAG	27.3	30.8	36.7	27.3	28.2	33.2	20.9	23.8	28.1	19.8	24.5	28.7	47.7	45.2	55.1	31.8
R1-Searcher	18.2	21.2	25.4	12.5	14.4	18.0	11.8	16.8	20.1	12.9	21.6	25.4	36.7	35.1	43.5	22.3

Appendix

Table 15: Main results on five QA benchmarks under two backbones. We report EM, ROUGE, F1, and Avg. Score.

Retrieval-Augmented Generation

Retrieval-augmented generation connects language models to external knowledge so answers can use current, domain-specific evidence.

See all 761 papers

GRIP: Grounded Reasoning via Information-Restricted Premises
Aug 17, 2026Lirui TengRetrieval-Augmented Generation LLM Grounding
University of Waterloo, Waterloo ON, Canada
Retrieval from Within: An Intrinsic Capability of Attention-Based Models
May 7, 2026Elad Hoffer, Yochai Blau, Edan Kinderman +3Retrieval-Augmented Generation Question Answering
NVIDIA · Department of Electrical Engineering, Technion, Haifa, Israel
OThink-SRR1: Search, Refine and Reasoning with Reinforced Learning for Large Language Models
Mar 27, 2026Haijian Liang, Zenghao Niu, Junjie Wu +3RL for RAG Agentic RAG
Shenzhen University · OPPO Research Institute

Aug 17, 2026J/K move · Enter open · S save

GRIP: Grounded Reasoning via Information-Restricted Premises

Lirui Teng

University of Waterloo, Waterloo ON, Canada

Retrieval-Augmented Generation LLM Grounding Retrieval-Augmented Reasoning Hallucination in Language Models Information Bottleneck

Open paper AskPDF arXiv

High-capacity encoders in retrieval-augmented generation (RAG) can let the query dominate the latent state, leaving retrieved evidence functionally irrelevant. We call this failure mode query dominance. To address it, we introduce GRIP (Grounded Reasoning via Information-Restricted Premises), which imposes capacity asymmetry: the decoder keeps full-dimensional access to the query, while retrieved evidence passes through a severe stochastic bottleneck. This forces the evidence channel to encode only the residual information unavailable from the query. Across five reasoning benchmarks, GRIP outperforms strong iterative baselines, cuts a query--latent mutual-information diagnostic by roughly 30

\times

(14.8

\to

0.47 bits), and reduces hallucination by 73%. Residual-alignment analysis further shows that the bottleneck output occupies subspaces less aligned with the query than baseline representations.

Abstract

Retrieval-Augmented Generation

Explore similar work

GRIP: Grounded Reasoning via Information-Restricted Premises

Retrieval from Within: An Intrinsic Capability of Attention-Based Models

OThink-SRR1: Search, Refine and Reasoning with Reinforced Learning for Large Language Models