Steering Language Models Before They Speak: Logit-Level Interventions

Hyeseon An; Shinwoo Park; Hyundong Jin; Yo-Sub Han

cs.CLJan 16, 2026

Steering Language Models Before They Speak: Logit-Level Interventions

Authors: Hyeseon An, Shinwoo Park, Hyundong Jin, Yo-Sub Han

Organizations: Department of Computer Science, Yonsei University, Seoul, South Korea

Abstract

Controllable generation requires language models to realize output characteristics such as reading level, politeness, and toxicity. Existing steering methods are often indirect, require access to internal activations, or depend on auxiliary trained models. We propose SWAI, a training-free inference-time method that addresses these limitations by steering directly in logit space using corpus-derived token statistics. SWAI computes z-normalized one-vs-rest log-odds scores from labeled corpora and biases high-scoring tokens only within the model's top-K candidate set, allowing control to favor target-characteristic tokens while preserving contextually plausible choices. Across readability, politeness, and toxicity control, SWAI consistently improves over prompt-based and prior logit-level baselines without modifying model parameters, accessing internal layers, or training an auxiliary model. Selectivity and lookup-table ablations show that the gains come from target-specific statistical scores rather than generic logit perturbation. These results indicate that effective steering does not require learned controllers when the logit intervention is guided by target-specific statistics under high-probability candidates.

Figures & tables

Figure 1: Overview of SWAI . It first derives token-level statistical scores from labeled corpora via smoothed one-vs-rest log-odds with variance normalization. During decoding, intervention is confined to a top- $K$ candidate set, where a fixed logit offset is applied to the highest-scoring tokens before sampling from the modified distribution.

OSE (Rewriting)
Model	Method	Acc	$F_{1}$	Precision	Recall	Perplexity	$\kappa$	MCC
Llama3.1 8B	Prompting	37.11%	0.363	36.20%	36.48%	11.99	0.052	0.052
	GeDi	38.96%	0.392	41.14%	39.16%	11.88	0.091	0.093
	DExperts	35.06%	0.173	33.33%	11.69%	13.12	0.000	0.000
	SWAI (ours)	84.53%	0.845	84.53%	84.53%	8.88	0.768	0.768
Llama3.2 1B	Prompting	28.87%	0.289	29.36%	28.99%	13.49	-0.063	-0.064

Table 1: Main results on OSE and WikiPol datasets. Across both datasets, SWAI consistently outperforms prompt-based baselines.

Figure 2: Class-wise Accuracy on OSE and WikiPol datasets. Reference denotes the Judge LLM performance on the original data before logit steering.

Method	Acc.	$F_{1}$	Prec.	Recall
Prompting	50.25%	0.010	100.00%	0.50%
GeDi	52.56%	0.461	54.92%	52.56%
DExperts	76.92%	0.585	72.22%	58.28%
SWAI (ours)	78.33%	0.552	97.56%	38.46%

Table 2: Results on the RealTox dataset. Performance is reported for prompt-based toxicity control and SWAI .

Figure 3: Effect of the steering bias $\delta$ on OSE (Llama3.1 8B). Accuracy and F1 $(\times 100)$ peak at $\delta=1.5$ , showing a clear trade-off between steering strength and generation quality.

Figure 4: Distinctive-token share of each class in generated text, as a function of set size $N$ . Across all $N$ , SWAI raises the target-class share (Target) above Baseline while pushing the others (Off-target) below it.

Figure 5: Lookup-table ablation on OSE and WikiPol . Each cell reports per-class one-vs-rest accuracy. Random shuffles the score values across token IDs, preserving the score distribution while breaking the token-to-score mapping; Flipped steers with another class’s table and is run on one class pair per dataset ( OSE : E and A; WIKIPOL : I and P), so the remaining class is left blank.

Dataset	$z_{r}(v)<0$	$z_{r}(v)=0$	$z_{r}(v)>0$
OSE	0.1–2.4%	14.3–19.4%	79.8–85.0%
WikiPol	1.1–2.2%	28.3–36.1%	62.8–70.6%
RealTox	0–7.0%	12.6–17.9%	80.4–82.1%

Table 3: Distribution of statistical score signs among favored tokens across datasets. Ranges are taken across target classes.

Judge	Acc.	F1	Prec.	Recall	$\kappa$	MCC
GPT-4o	84.53%	0.845	84.53%	84.53%	0.768	0.768
Gemini 3.5 Flash	85.19%	0.852	85.19%	85.19%	0.778	0.779

Table 4: Cross-LLM evaluation on OSE. Both judge models yield highly comparable performance on SWAI -generated text.

	Elementary	Intermediate	Advanced
Annotator 1	88.24	76.47	76.47
Annotator 2	76.47	52.94	70.59
Annotator 3	70.59	58.82	70.59
Overall	78.43	62.75	72.55

Table 5: Human evaluation accuracy (%) for SWAI -generated text across the three reading levels in OSE.

Appendix figures & tables6 assets

Supplementary material from the paper’s appendix.

Appendix

Model	Level	Accuracy	$F_{1}$	Precision	Recall	$\kappa$	MCC
Llama3.1 8B	Ele	88.93%	0.834	83.40%	83.40%	0.672	0.672
	Int	78.93%	0.684	68.40%	68.40%
	Adv	88.40%	0.826	82.60%	82.60%
	Total	78.13%	0.781	78.13%	78.13%
Llama3.2 1B	Ele	85.56%	0.783	78.52%	78.00%	0.553	0.553
	Int	72.00%	0.583	57.89%	58.67%

Appendix

Table 6: Results of open-ended generation on OSE . The model freely continues the input text and the generated outputs are evaluated for reading difficulty using the Judge LLM.

Figure 6: Per-condition selectivity. Change in each class’s distinctive-token share relative to Baseline ( $\Delta$ , top- $N$ =500). Steered rows show positive on-diagonal and negative off-diagonal entries, whereas prompt-only rows are near zero.

Rank	T	NT	Rank	T	NT
1	·you (39.69)	·· (38.40)	26	·idiot (16.86)	·state (9.87)
2	·shit (34.99)	·the (28.48)	27	·breasts (16.22)	200 (9.86)
3	·“ (34.68)	. (22.60)	28	·raped (16.02)	3 (9.48)
4	·sex (31.20)	·in (22.14)	29	·rape (15.89)	·season (9.38)
5	·stupid (30.47)	201 (18.74)	30	·suck (15.72)	9 (9.32)
6	·her (29.19)	·its (14.47)	31	·crap (15.63)	.S (9.26)

Appendix

Table 7: Top-50 tokens by statistical score for RealTox . T : toxic; NT : non-toxic. The leading dot (·) marks a token-internal word boundary. Scores are z-normalized one-vs-rest log-odds.

Rank	P	I	N	Rank	P	I	N
1	Thanks (11.89)	·why (10.07)	·< (5.08)	26	·happy (3.64)	And (3.50)	·one (2.38)
2	·help (10.29)	·Why (8.59)	url (4.71)	27	. (3.59)	How (3.48)	·Do (2.36)
3	·for (9.31)	’t (7.29)	·has (4.21)	28	·OK (3.57)	·exactly (3.45)	·( (2.35)
4	·please (8.69)	Why (7.11)	>, (3.91)	29	·By (3.52)	·all (3.45)	·where (2.34)
5	·you (8.20)	·not (6.09)	> (3.59)	30	·good (3.51)	·off (3.42)	I (2.30)
6	·Would (7.53)	·And (5.52)	·changed (3.49)	31	·me (3.51)	·" (3.40)	·official (2.28)

Appendix

Table 8: Top-50 tokens by statistical score for WikiPol . P : polite; I : impolite; N : neutral. The leading dot (·) marks a token-internal word boundary.

Rank	E	I	A	Rank	E	I	A
1	·people (13.74)	·However (3.02)	·being (6.86)	26	·It (4.77)	·due (1.60)	·across (3.34)
2	. (12.93)	m (2.87)	·as (6.65)	27	·good (4.75)	·celebrate (1.58)	·set (3.28)
3	·very (8.46)	bn (2.62)	·by (6.12)	28	·started (4.70)	·confidential (1.58)	·having (3.26)
4	·they (8.36)	·mainly (2.11)	·out (5.76)	29	·big (4.69)	·been (1.57)	·to (3.25)
5	·million (7.96)	·charge (1.98)	, (5.60)	30	·give (4.60)	·chosen (1.55)	·those (3.24)
6	·says (7.60)	·had (1.96)	ed (5.44)	31	·lot (4.56)	·corner (1.54)	·into (3.15)

Appendix

Table 9: Top-50 tokens by statistical score for Ose . E : elementary; I : intermediate; A : advanced. The leading dot (·) marks a token-internal word boundary.

Panel of Figure 7 — Figure 7: Qualitative promoted-token examples for OSE on the shared topic of ‘Amazon’. Dark highlights indicate tokens promoted by the logit bias; light highlights indicate biased tokens that were already top-ranked before intervention.

Language Model Steering

Language model steering controls a model’s behavior at inference time by intervening in its representations, components, or decoding process.

See all 279 papers

DLM-SWAI: Steering Diffusion Language Models Before They Unmask
May 28, 2026Hyeseon An, Yo-Sub HanControllable Text Generation Diffusion Language Model Inference
Department of Computer Science, Yonsei University, Seoul, South Korea
Steer Like the LLM: Activation Steering that Mimics Prompting
May 5, 2026Geert Heyman, Frederik VandeputteLanguage Model Steering LLM Prompting
Nokia Bell Labs, Belgium.
A Coin Flip Per Token: Bernoulli Sparse Steering of Large Language Models
Jul 6, 2026Nima Eshraghi, Lovedeep Gondara, Yuqing Huang +5Language Model Steering Sparse Autoencoders
The Vanguard Group, Inc. · University of Toronto

May 28, 2026J/K move · Enter open · S save

DLM-SWAI: Steering Diffusion Language Models Before They Unmask

Hyeseon An, Yo-Sub Han

Department of Computer Science, Yonsei University, Seoul, South Korea

Controllable Text Generation Diffusion Language Model Inference Diffusion Language Models

Open paper AskPDF arXiv

Steering language model generation toward desired textual properties is essential for practical deployment, and inference-time methods are particularly appealing because they enable controllable generation without retraining. Recent work has also highlighted diffusion language models as an emerging generation paradigm with distinct decoding properties. However, most existing steering approaches either rely on auxiliary models or are designed for autoregressive next-token decoding, making them difficult to apply to diffusion language models DLMs, which generate text through iterative denoising of partially masked sequences. Therefore, we propose DLM-SWAI, a simple training-free steering method that biases the token distribution at each denoising step using pre-computed token-level style scores. Experiments on style and safety control tasks show that DLM-SWAI effectively steers diffusion language models while preserving generation quality and requiring minimal computational overhead. Ablations further reveal a controllable trade-off between steering strength and fluency, and our analysis links class-wise steerability to the strength of token-level attribute cues.

Abstract

Language Model Steering

Explore similar work

DLM-SWAI: Steering Diffusion Language Models Before They Unmask

Steer Like the LLM: Activation Steering that Mimics Prompting

A Coin Flip Per Token: Bernoulli Sparse Steering of Large Language Models