Large Decision Models: Repurposing Large Language Models as Single-Pass Decision Makers

Ziyi Chen; Yan Zhang; Jianhui Wei; Jianing Yin; Daoan Zhang

cs.AIOct 11, 2026

Large Decision Models: Repurposing Large Language Models as Single-Pass Decision Makers

Authors: Ziyi Chen, Yan Zhang, Jianhui Wei, Jianing Yin, Daoan Zhang

Organizations: University of Illinois Urbana-Champaign · National University of Singapore · Zhejiang University · University of Pennsylvania ·

Abstract

Large language models make decisions through parallel prefill and sequential decoding. Although effective, this approach incurs decoding costs and does not explicitly expose decision probabilities. Existing task-specific classifiers require per-task training and offer limited generalization to unseen decision schemas. We propose Large Decision Models (LDMs), repurposing language models to return explicit probabilities over labels specified at run time using one prefill forward pass. We first adapt models to this interface through Diverse Decision Supervision (DDS), covering varied domains, rules and labels. To further narrow the performance gap to a reasoning model, Self-Posed Reasoning Distillation (SPRD) adds tasks generated by a frozen teacher, filters them for answer consistency, and trains on label distributions obtained from repeated teacher reasoning. We evaluate on DecisionVista, pairing new-schema tests with held-out human-labelled sources. Across three base models, the recipe improves accuracy over a step-matched baseline by 1.8 and 1.5 percentage points, respectively, achieving a 182-fold speedup in latency over reasoning rollout on Gemma-4-31B. Reasoning remains stronger on the new-schema test, motivating a confidence router for uncertain decisions. On development sets, routing recovers a median 92% of reasoning's accuracy advantage while generating 27-58% of its tokens. Code is available at https://github.com/JianhuiWei7/LDM.

Figures & tables

Figure 1: Overview. (a) The LDM processes the task in one prefill, projects the last input hidden state through the LM head, and normalises only the allowed option logits. (b) Baseline and DDS examples supply hard-label cross-entropy supervision. The frozen Qwen3.8-27B teacher generates four reasoning-and-answer samples per self-posed task. SPRD-C filters for consistency and distils answer-vote distributions with

\lambda D_{\mathrm{KL}}(q\|p_{\theta})

. (c) A margin router accepts confident LDM predictions and otherwise invokes the student’s untuned base for one reasoning rollout.

Panel of Figure 2 — Figure 2: Confidence routing: accuracy (mean of the transfer suite, DV-Dev-A and DV-Dev-B; seed means) against generated tokens as a share of reasoning on every item. Router thresholds are fixed on the validation suite at routing budgets of 5–70%; large marker: 30%. Dashed: random routing, expected accuracy at the same generated-token share. Star: reasoning on every item.

Mode	Median	Rate	Speedup	GPU-h
	(ms)	(req./s)	( $\times$ )	(/1M)
Reasoning rollout	7,380	0.787	1.0	706.2
Tuned readout	40.6	44.2	181.8	12.6

Table 4: Gemma-4-31B serving efficiency. Speedup uses median latency relative to reasoning; GPU-hours per million requests are extrapolated.

Appendix figures & tables17 assets

Supplementary material from the paper’s appendix.

Appendix

Step	Event
1	DV-Dev-A and DV-Dev-B built and used for earlier recipes (Appendix B ); DV-Stress built and sealed.
2	Pool generated; DV-Stress family descriptions used only in the exclusion list.
3	$\lambda$ , $\tau$ chosen on the validation suite and DV-Dev-A (9B); SPRD variants and DV-Stress checkpoints frozen and recorded.
4	DV-Stress read once (+1.0 over baseline).
5	Third-family (Magistral) labels for DV-Dev-A, DV-Dev-B and DV-Stress; gains concentrate on contested items.
6	Step-matched baseline, SPRD-C and other variants trained; DV-Pilot families proposed and audited, protocol recorded.

Appendix

Table 6: Order of decisions relative to test-set reads.

Validation source	Readout	Reasoning	Gap
WinoGrande	67.5	93.5	+26.0
MedMCQA	65.0	72.5	+7.5
TweetEval irony	69.5	76.0	+6.5
SMS spam	97.5	98.5	+1.0
ARC-Challenge	96.0	96.5	+0.5
CaseHOLD	78.5	78.0	$-0.5$

Appendix

Table 7: Rollout–readout gap of untuned Qwen3.8-27B. Reasoning = majority of four thinking-mode samples; readout = single-pass letter readout.

Vote share	Items	Readout	gpt-5.5	Gen. intent
1.0	13,191	85.9	93.9	73.9
$[0.75,1)$	1,638	54.8	64.0	43.2
$<0.75$	1,151	47.3	51.9	40.0
All	15,980	79.9	87.8	68.3

Appendix

Table 8: Agreement (%) of other label sources with the reasoning-teacher majority on the self-posed pool, by the teacher’s vote share (15,980 of the 16,040 items have at least one parseable teacher answer).

Item	Setting
Bases	Qwen3.5-9B; Qwen3.8-27B (rev. 1d4bf0f2); Gemma-4-31B-it
Adapter	LoRA rank 16, $\alpha{=}32$ , dropout 0.05, on every linear layer of the language model; bf16; final step, no checkpoint selection
Optimiser	AdamW, lr $5{\times}10^{-5}$ , linear schedule with 5% warm-up, gradient clipping 1.0, effective batch 16 (9B: 4 $\times$ 4 accumulation; 27B/31B: 2 $\times$ 8), max length 2,048
Objective	Eq. 3 : cross-entropy on baseline and DDS records; $\lambda{=}3$ KL to SPRD-C vote targets; option order re-shuffled per draw
Data	baseline = 20k public records + Nimble’s 2.7k training items (22.4k); DDS = half of the 63.4k mixture (31.7k, fixed sample); SPRD-C = 10,331 items labelled by Qwen3.8-27B for every base
Steps	baseline, step-matched: 4,032 (2.9 passes); +DDS: 3,386; +SPRD-C: 2,050; +both: 4,031 (one pass each); one-pass baseline: 1,405

Appendix

Table 10: Configuration of every variant in the controlled study.

Configuration	Training data	Steps	Seeds	Status	Where
Baseline, step-matched	baseline 22.4k	4,032	3 / 3 / 3	pre-reg. comparator	Tables 3 , 2 , 9
+DDS + SPRD-C	baseline + DDS 31.7k + SPRD-C 10.3k	4,031	3 / 3 / 3	pre-reg. primary	Tables 3 , 2 , 9
+DDS	baseline + DDS	3,386	3 / 3 / 3	pre-reg. secondary; descr. vs. baseline	Tables 2 , 9
+SPRD-C	baseline + SPRD-C	2,050	3 / 3 / 3	pre-reg. secondary; descr. vs. baseline	Tables 2 , 9
Baseline, one pass	baseline	1,405	2 / 2 / 2	descr.	Table 9
Validation-selected baseline and recipe	as above, update scale 1 or 0.65	–	3 / 2 / 2	post hoc	Tables 2 , 35

Appendix

Table 11: Experiment summary. Steps at effective batch 16. Seeds: 9B / 27B / 31B. Status: pre-reg. , primary or secondary contrast fixed before the sealed reads; descr. , descriptive; post hoc , specified after the sealed reads.

Component	Records	Content
additional public data	35.6k	54 public sources (medicine, science, math/logic incl. AQuA and 26 BIG-bench tasks, reading, commonsense, preference: HH-RLHF, SHP, PKU-SafeRLHF, safety, code, fact-checking, finance)
Super-NaturalInstructions	15.4k	258 categorical tasks; tasks derived from evaluation sources excluded by name and by content (a BoolQ-derived task was removed after inspection)
grid-based synthesis	13.0k $\to$ 8.4k	gpt-5.6 (sol/terra/luna) invents tasks over 44 fields $\times$ 17 decision types; kept only if gpt-6-astra agrees blind; audited against evaluation families
grounded synthesis	10.7k $\to$ 3.9k	new decision questions about real public documents, audited likewise
DDS mixture	63.4k	additional public data + instruction tasks + audited synthesis; no benchmark-source data; half (31.7k) sampled per run

Appendix

Table 14: DDS components (“

\to

”: before and after the evaluation-family audit). All records are 13-gram de-contaminated against evaluation items available when the mixture was built; the later DecisionVista-Synth and DecisionVista-Human sets were checked against the frozen mixture.

Model	Role in training data	Role in evaluation data
Qwen3.8-27B	SPRD schema and item author; reasoning teacher for every SPRD-C variant; 27B student	evaluated
Qwen3.5-9B, Gemma-4-31B	students	evaluated
gpt-5.5	none in any recipe of Table 9 (labels of the “gpt-5.5 labels” control only)	author of DV-Dev-A–DV-Pilot
gpt-6-astra	blind verifier of DDS grid-based synthetic labels; auditor retaining synthesised DDS tasks and SPRD schemas only when judged distinct from evaluation families	blind verifier of DV-Dev-A–DV-Pilot; author of 10 DV-Synth families and blind labeller of the other 25
gpt-5.6 (sol/terra/luna)	author of DDS grid-based synthetic tasks	none
Magistral-Small	none	third-family labels on DV-Dev-A, DV-Dev-B and DV-Stress; required agreement on DV-Pilot

Appendix

Table 18: Every model that touched training or evaluation data, and how. gpt-6-astra is the one model with a role on both sides: its blind labels decide which synthesised DDS records are kept, and it wrote or labelled DecisionVista-Synth items. We therefore report DecisionVista-Synth results separately for the 25 families it did not write (Table 16 ).

Qwen3.5-9B, 3 seeds	Transfer	DV-Dev-A	DV-Dev-B	DV-Synth †	DV-Human †
Readout, no routing	71.8	87.1	84.3	66.9	74.9
Margin router (main method)	76.2	95.8	91.5	85.2	76.5
Learned rejection score	76.0	95.5	90.8	87.3	76.3
learned $-$ margin	$-$ 0.2	$-$ 0.3	$-$ 0.7	+2.1	$-$ 0.2
95% interval	[ $-$ 0.8, +0.2]	[ $-$ 1.2, +0.4]	[ $-$ 1.7, +0.3]	[+1.2, +2.9]	[ $-$ 0.6, +0.2]
Learned rejection, argmax only	74.7	92.6	87.7	80.4	75.6

Appendix

Table 21: Learned rejection vs. the margin router, Qwen3.5-9B, seed means over three seeds. Budget-based routers route 30% of the validation suite items; the argmax-only router routes 10–14% of transfer-suite and DV-Dev-A items. Intervals: paired bootstrap over items (task families for synthetic DecisionVista sets). † Sealed, descriptive. ‡ DDS+SPRD-C readout shown the option without training on it, seed 17 only.

	Val.			Transfer suite
Base	Readout	Reasoning	Gap	Readout	Reasoning	Gap
Qwen3.5-9B (dense)	71.7	79.1	7.4	66.2	75.0	8.8
Qwen3.8-27B (dense)	79.3	82.7	3.4	75.4	80.6	5.2
Qwen3.5-35B-A3B (MoE)	77.3	83.4	6.1	73.6	80.1	6.5
Qwen3.5-122B-A10B (MoE)	79.8	85.9	6.1	74.0	81.6	7.6
Qwen3.5-397B-A17B (MoE) †	81.9	87.6	5.7	78.2	83.4	5.2

Appendix

Table 22: Readout versus reasoning rollout for untuned dense and MoE bases (accuracy, %). Gap: reasoning minus readout. The 27B dense base is Qwen3.8; the others are Qwen3.5. 122B-A10B reasoning ran in FP8, as did the 27B. † FP8 checkpoint in both modes (routed experts FP8, other weights bf16); on 35B-A3B, FP8 versus bf16 changes validation readout accuracy by 0.7 points.

	All tokens				Decision token only
$k$	Val.	Transfer	DV-Dev-A	DV-Dev-B	Val.	Transfer	DV-Dev-A	DV-Dev-B
8 (trained)	77.3	73.6	86.5	81.9	77.3	73.6	86.5	81.9
16	76.5	72.1	84.8	80.7	77.4	73.5	86.5	82.4
32	72.7	67.2	81.6	76.7	–	–	–	–
64	66.6	61.1	76.2	72.0	76.4	73.4	84.7	80.9
128	59.5	50.7	68.8	65.4	–	–	–	–

Appendix

Table 23: Untuned Qwen3.5-35B-A3B readout with more experts per token (accuracy, %). All tokens : every token uses its top-

k

experts. Decision token : only the final position uses top-

k

; all other tokens keep top-8.

$k$	Val.	Transfer	DV-Dev-A	DV-Dev-B
8 (default)	80.2	75.0	84.5	81.5
16	80.5	74.8	87.7	82.8
32	77.9	73.6	86.3	81.8
64	75.4	70.9	86.1	79.3
128	74.1	68.9	84.8	77.7
256 (all)	71.1	65.0	80.8	73.5

Appendix

Table 24: Qwen3.5-35B-A3B LoRA-tuned with top-

k

routing and evaluated at the same

k

(accuracy, %; same 5.5k-record data slice, recipe and trainer for every

k

; LoRA covers attention, the shared expert and the router, not the routed experts). Means over seeds 17 and 29.

Base	Val.	Transfer	DV-Synth	DV-Human	Typed-dec.
Qwen3.8-27B (dense)	83.7	77.5	78.2	77.0	74.7
Gemma-4-31B (dense)	84.1	79.0	80.8	80.2	75.4
Qwen3.5-397B-A17B (MoE)	86.5	80.7	75.8	79.4	73.8

Appendix

Table 25: Full recipe on dense and MoE bases (accuracy, %; DV-Synth family-macro, DV-Human source-macro). Dense rows: three seeds, Qwen-27B SPRD-C labels. 397B-A17B: one seed, own labels, FP8 checkpoint; post hoc and descriptive.

Variant	Val.	DV-Dev-A	DV-Dev-B	Transfer
Baseline, step-matched	82.4	88.9	86.3	77.6
+ SPRD-C, Qwen-27B labels	83.2	91.1	89.6	78.0
+ SPRD-C, own labels	83.7	91.4	89.9	78.6
+ DDS + SPRD-C, Qwen-27B labels	84.1	91.1	90.7	79.0
+ DDS + SPRD-C, own labels	83.3	92.1	91.3	78.4

Appendix

Table 26: Gemma-4-31B-it, three-seed means (accuracy, %), validation sets. Own labels: SPRD-C relabelled by Gemma’s own reasoning.

Variant	Nimble-13	Typed-dec.	DV-Synth	DV-Human	ECE
Baseline, step-matched	75.3	74.1	79.9	79.5	22.1 $\rightarrow$ 5.4
+ SPRD-C, Qwen-27B labels	76.6	75.1	79.6	79.4	21.6 $\rightarrow$ 6.0
+ SPRD-C, own labels	76.7	75.3	80.6	80.0	21.0 $\rightarrow$ 5.9
+ DDS + SPRD-C, Qwen-27B labels	76.3	75.4	80.8	80.2	20.7 $\rightarrow$ 5.4
+ DDS + SPRD-C, own labels	77.1	75.3	80.6	80.2	21.2 $\rightarrow$ 6.1

Appendix

Table 27: The same Gemma-4-31B-it variants on public and test sets (three-seed means, accuracy %; ECE on the transfer suite before

\rightarrow

after temperature scaling). All reads of the own-label variants are post hoc and descriptive; they are not part of the pre-registered contrasts.

Base	Untuned	1/16 data	Full data	Full, $s{=}0.65$	Gain	Share
0.8B	50.9	60.8	66.7	65.8	+15.8	76%
2B	59.9	68.3	73.0	72.0	+13.1	82%
4B	72.1	76.9	79.9	80.2	+7.8	90%
9B	76.0	81.2	81.9	82.6	+5.9	93%
27B	84.3	–	86.6	87.4	+2.3	97%

Appendix

Table 28: Untuned readout versus decision tuning across Qwen3.5 sizes (accuracy, %, mean of four development sets; scaling study, one seed). Share: untuned as a fraction of fully tuned accuracy. 1/16: tuned on 1/16 of the data.

Variant	Val.	Transfer	Nimble-13	Typed-dec.	DV-Synth	DV-Human	ECE
Untuned	71.8	66.1	73.7	63.5	52.5	69.4	17.1 $\rightarrow$ 6.5
Head only, baseline	72.7	66.2	73.4	64.2	52.7	69.5	3.1 $\rightarrow$ 6.2
LoRA, baseline	75.9	69.7	74.1	69.8	64.5	72.8	23.8 $\rightarrow$ 4.3
Head only, DDS mix	72.6	66.2	73.8	64.9	52.2	69.5	4.3 $\rightarrow$ 6.2
LoRA, DDS mix	77.1	71.6	71.6	72.2	66.1	75.0	10.0 $\rightarrow$ 2.7

Appendix

Table 29: Head-only training versus LoRA on Qwen3.5-9B (three-seed means, accuracy %; ECE on the transfer suite before

\rightarrow

after temperature scaling).

Source	Clust.	Mean $\Delta$	Seed-pair range	$>0$
ContractNLI	242	+2.7	[+0.7, +7.1]	7/7
CSQA2	300	+3.1	[ $-$ 0.3, +5.7]	6/7
ETHICS	300	$-$ 1.3	[ $-$ 3.7, +1.3]	2/7
Fig-QA	300	+0.9	[ $-$ 0.7, +2.3]	6/7
FOLIO	73	+1.5	[ $-$ 2.0, +6.4]	3/7
LIAR	299	+0.4	[ $-$ 4.7, +6.4]	3/7

Appendix

Table 32: DecisionVista-Human gain of DDS+SPRD-C over the step-matched baseline by source (points), over the seven pre-registered seed pairs (three at 9B, two on the 27B and 31B bases). Clusters: distinct documents or premise sets; a cluster bootstrap over them gives the same pooled interval as the item bootstrap ([0.9, 2.2]).

LLM Decision-Making

LLM decision-making studies how language models choose, act, seek information, and allocate resources under goals, uncertainty, constraints, and social pressures.

See all 216 papers

Diagnosing and Improving Probabilistic Reasoning in Large Language Models
Sep 29, 2026Huaman Sun, Dingcheng Wang, Jason Hartline +1RL for Language Model Reasoning LLM Decision-Making
Department of Computer Science Northwestern University Evanston, IL 60201, USA
StartLux-Decision: AutoThink Decision Models with Reinforcement Learning for Calibrated Deliberative Decisions
Oct 12, 2026Xialie Zhuang, Zunhai Su, Jia Pan +3LLM Decision-Making RL for Language Model Reasoning
StartLux · The University of Hong Kong · Southern University of Science and Technology
Releasing the Inner Jev: A Simple Recipe for Open Decision Models
Oct 12, 2026Jonathan Geuter, Andrew Cheng, David Alvarez-MelisLLM Decision-Making LLM Fine-Tuning
Harvard University · Kempner Institute

Sep 29, 2026J/K move · Enter open · S save

Diagnosing and Improving Probabilistic Reasoning in Large Language Models

Huaman Sun, Dingcheng Wang, Jason Hartline, Jessica Hullman

Department of Computer Science Northwestern University Evanston, IL 60201, USA

RL for Language Model Reasoning LLM Decision-Making

Open paper AskPDF arXiv

Large language models (LLMs) are increasingly proposed as decision assistants who must reason probabilistically from available evidence under explicit decision costs. We propose a decision-theoretic framework that decomposes LLMs' decision loss into two components: forming accurate beliefs from provided evidence and translating those beliefs into actions that optimize a provided utility function. Using a synthetic benchmark with known ground truth, we apply the decomposition to characterize probabilistic reasoning in frontier and open-sourced models. We further evaluate whether RL interventions targeting beliefs, decisions, or both improve these components across three domains, whether improvements transfer across components and elicitation formats, and whether decision performance can improve without improvement in belief formation. We find that targeting one component of probabilistic reasoning redistributes decision loss, improving the target without necessarily transferring to others, and that jointly targeting belief formation and decision-making improves both but hinges on matched formats between training and evaluation.

Abstract

LLM Decision-Making

Explore similar work

Diagnosing and Improving Probabilistic Reasoning in Large Language Models

StartLux-Decision: AutoThink Decision Models with Reinforcement Learning for Calibrated Deliberative Decisions

Releasing the Inner Jev: A Simple Recipe for Open Decision Models