# On The Effectiveness-Fluency Trade-Off In LLM Conditioning: A Systematic Study

arXiv: https://arxiv.org/abs/2606.12234 · HTML: https://mlsanity.com/paper/2606.12234/

## Key facts

- Cited by 1 papers in the map (data through 14 Oct 2026).
- Builds on 24 of its 46 references that are in the map.
- First posted on arXiv 10 Jun 2026.

## Builds on (most cited first, 24 of 24)

- [Language Models are Few-Shot Learners](https://mlsanity.com/paper/2005.14165/) (2020)
- [Training language models to follow instructions with human feedback](https://mlsanity.com/paper/2203.02155/) (2022)
- [Exploring the Limits of Transfer Learning with a Unified Text-to-Text Transformer](https://mlsanity.com/paper/1910.10683/) (2019)
- [Sentence-BERT: Sentence Embeddings using Siamese BERT-Networks](https://mlsanity.com/paper/1908.10084/) (2019)
- [Measuring Massive Multitask Language Understanding](https://mlsanity.com/paper/2009.03300/) (2020)
- [Finetuned Language Models Are Zero-Shot Learners](https://mlsanity.com/paper/2109.01652/) (2021)
- [The Curious Case of Neural Text Degeneration](https://mlsanity.com/paper/1904.09751/) (2019)
- [Inference-Time Intervention: Eliciting Truthful Answers from a Language Model](https://mlsanity.com/paper/2306.03341/) (2023)
- [RealToxicityPrompts: Evaluating Neural Toxic Degeneration in Language Models](https://mlsanity.com/paper/2009.11462/) (2020)
- [Steering Llama 2 via Contrastive Activation Addition](https://mlsanity.com/paper/2312.06681/) (2023)
- [Neural Text Generation with Unlikelihood Training](https://mlsanity.com/paper/1908.04319/) (2019)
- [MAUVE: Measuring the Gap Between Neural Text and Human Text using Divergence Frontiers](https://mlsanity.com/paper/2102.01454/) (2021)
- [AxBench: Steering LLMs? Even Simple Baselines Outperform Sparse Autoencoders](https://mlsanity.com/paper/2501.17148/) (2025)
- [ReFT: Representation Finetuning for Language Models](https://mlsanity.com/paper/2404.03592/) (2024)
- [A Survey of Evaluation Metrics Used for NLG Systems](https://mlsanity.com/paper/2008.12009/) (2020)
- [Language Model Evaluation Beyond Perplexity](https://mlsanity.com/paper/2106.00085/) (2021)
- [Controlling Language and Diffusion Models by Transporting Activations](https://mlsanity.com/paper/2410.23054/) (2024)
- [On the Blind Spots of Model-Based Evaluation Metrics for Text Generation](https://mlsanity.com/paper/2212.10020/) (2022)
- [Whispering Experts: Neural Interventions for Toxicity Mitigation in Language Models](https://mlsanity.com/paper/2407.12824/) (2024)
- [Beyond Linear Steering: Unified Multi-Attribute Control for Language Models](https://mlsanity.com/paper/2505.24535/) (2025)
- [Perplexity Cannot Always Tell Right from Wrong](https://mlsanity.com/paper/2601.22950/) (2026)
- [Enhancing Instruction Following of LLMs via Activation Steering with Dynamic Rejection](https://mlsanity.com/paper/2603.06745/) (2026)
- [Beyond Multiple Choice: Evaluating Steering Vectors for Summarization](https://mlsanity.com/paper/2505.24859/) (2025)
- [Steering Language Models in Multi-Token Generation: A Case Study on Tense and Aspect](https://mlsanity.com/paper/2509.12065/) (2025)

## Built on by (most cited first, 1 of 1)

- [One Token Can Be Enough: Bridging Prompting and Activation Steering with Prefix Steering](https://mlsanity.com/paper/2610.04967/) (2026)

## Abstract (arXiv)

Controlling the output of Large Language Models (LLMs) is a central challenge for their reliable deployment, yet a clear understanding of the involved trade-offs remains elusive. Current approaches to conditioning are often evaluated with a narrow focus on their effectiveness at injecting or removing a target concept, neglecting generation quality. We systematically investigate a range of conditioning methods in both injection and removal scenarios. We find that efficient steering methods frequently achieve conditioning at a steep cost to fluency. Furthermore, we identify a critical yet previously overlooked interaction with the training paradigm: activation steering methods are far less effective on instruction-tuned models than on their base counterparts. Simple prompting and full-fledged supervised fine-tuning, on the other hand, are viable options for concept injection, but are not as good at concept removal. Finally, cheaply computed textual metrics highly correlate to costly LLM-as-judge scores, and provide insights on the behavior of conditioning methods.