# Inference-Time Policy Adapters (IPA): Tailoring Extreme-Scale LMs without Fine-tuning

arXiv: https://arxiv.org/abs/2305.15065 · HTML: https://mlsanity.com/paper/2305.15065/

## Key facts

- Cited by 33 papers in the map (data through 14 Oct 2026).
- Builds on 88 of its 100 references that are in the map.
- First posted on arXiv 24 May 2023.

## Builds on (most cited first, 30 of 88)

- [Language Models are Few-Shot Learners](https://mlsanity.com/paper/2005.14165/) (2020)
- [GPT-4 Technical Report](https://mlsanity.com/paper/2303.08774/) (2023)
- [Training language models to follow instructions with human feedback](https://mlsanity.com/paper/2203.02155/) (2022)
- [Proximal Policy Optimization Algorithms](https://mlsanity.com/paper/1707.06347/) (2017)
- [Exploring the Limits of Transfer Learning with a Unified Text-to-Text Transformer](https://mlsanity.com/paper/1910.10683/) (2019)
- [LoRA: Low-Rank Adaptation of Large Language Models](https://mlsanity.com/paper/2106.09685/) (2021)
- [Prefix-Tuning: Optimizing Continuous Prompts for Generation](https://mlsanity.com/paper/2101.00190/) (2021)
- [Self-Refine: Iterative Refinement with Self-Feedback](https://mlsanity.com/paper/2303.17651/) (2023)
- [The Curious Case of Neural Text Degeneration](https://mlsanity.com/paper/1904.09751/) (2019)
- [Learning to summarize from human feedback](https://mlsanity.com/paper/2009.01325/) (2020)
- [The Lottery Ticket Hypothesis: Finding Sparse, Trainable Neural Networks](https://mlsanity.com/paper/1803.03635/) (2018)
- [Don't Stop Pretraining: Adapt Language Models to Domains and Tasks](https://mlsanity.com/paper/2004.10964/) (2020)
- [Google's Neural Machine Translation System: Bridging the Gap between Human and Machine Translation](https://mlsanity.com/paper/1609.08144/) (2016)
- [Rethinking the Role of Demonstrations: What Makes In-Context Learning Work?](https://mlsanity.com/paper/2202.12837/) (2022)
- [Calibrate Before Use: Improving Few-Shot Performance of Language Models](https://mlsanity.com/paper/2102.09690/) (2021)
- [Don't Give Me the Details, Just the Summary! Topic-Aware Convolutional Neural Networks for Extreme Summarization](https://mlsanity.com/paper/1808.08745/) (2018)
- [RealToxicityPrompts: Evaluating Neural Toxic Degeneration in Language Models](https://mlsanity.com/paper/2009.11462/) (2020)
- [Fantastically Ordered Prompts and Where to Find Them: Overcoming Few-Shot Prompt Order Sensitivity](https://mlsanity.com/paper/2104.08786/) (2021)
- [Diffusion-LM Improves Controllable Text Generation](https://mlsanity.com/paper/2205.14217/) (2022)
- [Neural Network Acceptability Judgments](https://mlsanity.com/paper/1805.12471/) (2018)
- [Analyzing Multi-Head Self-Attention: Specialized Heads Do the Heavy Lifting, the Rest Can Be Pruned](https://mlsanity.com/paper/1905.09418/) (2019)
- [DialoGPT: Large-Scale Generative Pre-training for Conversational Response Generation](https://mlsanity.com/paper/1911.00536/) (2019)
- [Are Sixteen Heads Really Better than One?](https://mlsanity.com/paper/1905.10650/) (2019)
- [Sequence-Level Knowledge Distillation](https://mlsanity.com/paper/1606.07947/) (2016)
- [Plug and Play Language Models: A Simple Approach to Controlled Text Generation](https://mlsanity.com/paper/1912.02164/) (2019)
- [CTRL: A Conditional Transformer Language Model for Controllable Generation](https://mlsanity.com/paper/1909.05858/) (2019)
- [Intrinsic Dimensionality Explains the Effectiveness of Language Model Fine-Tuning](https://mlsanity.com/paper/2012.13255/) (2020)
- [Linear Mode Connectivity and the Lottery Ticket Hypothesis](https://mlsanity.com/paper/1912.05671/) (2019)
- [Contrastive Decoding: Open-ended Text Generation as Optimization](https://mlsanity.com/paper/2210.15097/) (2022)
- [Recipes for building an open-domain chatbot](https://mlsanity.com/paper/2004.13637/) (2020)

## Built on by (most cited first, 30 of 33)

- [Parameter-Efficient Fine-Tuning for Large Models: A Comprehensive Survey](https://mlsanity.com/paper/2403.14608/) (2024)
- [From Decoding to Meta-Generation: Inference-time Algorithms for Large Language Models](https://mlsanity.com/paper/2406.16838/) (2024)
- [Weak-to-Strong Jailbreaking on Large Language Models](https://mlsanity.com/paper/2401.17256/) (2024)
- [GuardReasoner: Towards Reasoning-based LLM Safeguards](https://mlsanity.com/paper/2501.18492/) (2025)
- [Aligner: Efficient Alignment by Learning to Correct](https://mlsanity.com/paper/2402.02416/) (2024)
- [Value Kaleidoscope: Engaging AI with Pluralistic Human Values, Rights, and Duties](https://mlsanity.com/paper/2309.00779/) (2023)
- [Tuning Language Models by Proxy](https://mlsanity.com/paper/2401.08565/) (2024)
- [Decoding-time Realignment of Language Models](https://mlsanity.com/paper/2402.02992/) (2024)
- [Safety in Large Reasoning Models: A Survey](https://mlsanity.com/paper/2504.17704/) (2025)
- [PAL: Pluralistic Alignment Framework for Learning from Heterogeneous Preferences](https://mlsanity.com/paper/2406.08469/) (2024)
- [Parameter-Efficient Fine-Tuning in Large Models: A Survey of Methodologies](https://mlsanity.com/paper/2410.19878/) (2024)
- [What is the Role of Small Models in the LLM Era: A Survey](https://mlsanity.com/paper/2409.06857/) (2024)
- [Tailoring Self-Rationalizers with Multi-Reward Distillation](https://mlsanity.com/paper/2311.02805/) (2023)
- [Parameter-Efficient Detoxification with Contrastive Decoding](https://mlsanity.com/paper/2401.06947/) (2024)
- [StyleRemix: Interpretable Authorship Obfuscation via Distillation and Perturbation of Style Elements](https://mlsanity.com/paper/2408.15666/) (2024)
- [Selective Prompt Anchoring for Code Generation](https://mlsanity.com/paper/2408.09121/) (2024)
- [HiddenGuard: Fine-Grained Safe Generation with Specialized Representation Router](https://mlsanity.com/paper/2410.02684/) (2024)
- [BBox-Adapter: Lightweight Adapting for Black-Box Large Language Models](https://mlsanity.com/paper/2402.08219/) (2024)
- [Let Me Teach You: Pedagogical Foundations of Feedback for Language Models](https://mlsanity.com/paper/2307.00279/) (2023)
- [Controllable Text Generation in the Instruction-Tuning Era](https://mlsanity.com/paper/2405.01490/) (2024)
- [PortLLM: Personalizing Evolving Large Language Models with Training-Free and Portable Model Patches](https://mlsanity.com/paper/2410.10870/) (2024)
- [Sysformer: Safeguarding Frozen Large Language Models with Adaptive System Prompts](https://mlsanity.com/paper/2506.15751/) (2025)
- ["It's trained by non-disabled people": Evaluating How Image Quality Affects Product Captioning with Vision-Language Models](https://mlsanity.com/paper/2511.08917/) (2025)
- [PANDA: Preference Adaptation for Enhancing Domain-Specific Abilities of LLMs](https://mlsanity.com/paper/2402.12835/) (2024)
- [Adaptive and Resource-efficient Agentic AI Systems for Mobile and Embedded Devices: A Survey](https://mlsanity.com/paper/2510.00078/) (2025)
- [SDA: Steering-Driven Distribution Alignment for Open LLMs without Fine-Tuning](https://mlsanity.com/paper/2511.16324/) (2025)
- [Contrastive Perplexity for Controlled Generation: An Application in Detoxifying Large Language Models](https://mlsanity.com/paper/2401.08491/) (2024)
- [Transferable Model-agnostic Vision-Language Model Adaptation for Efficient Weak-to-Strong Generalization](https://mlsanity.com/paper/2508.08604/) (2025)
- [Kad: A Framework for Proxy-based Test-time Alignment with Knapsack Approximation Deferral](https://mlsanity.com/paper/2510.27017/) (2025)
- [Advanced Black-Box Tuning of Large Language Models with Limited API Calls](https://mlsanity.com/paper/2511.10210/) (2025)

## Abstract (arXiv)

While extreme-scale language models have demonstrated exceptional performance on a variety of language tasks, the degree of control over these language models through pure prompting can often be limited. Directly fine-tuning such language models can be effective for tailoring them, but it can be either extremely costly (e.g., GPT-3) or not even feasible for the broader community (e.g., GPT-4). We propose Inference-time Policy Adapters (IPA), which efficiently tailors a language model such as GPT-3 without fine-tuning it. IPA guides a large base model during decoding time through a lightweight policy adapter trained to optimize an arbitrary user objective with reinforcement learning. On five challenging text generation tasks, such as toxicity reduction and lexically constrained generation, IPA consistently brings significant improvements over off-the-shelf language models. It outperforms competitive baseline methods, sometimes even including expensive fine-tuning. In particular, tailoring GPT-2 with IPA can outperform GPT-3, while tailoring GPT-3 with IPA brings a major performance boost over GPT-3 (and sometimes even over GPT-4). Our promising results highlight the potential of IPA as a lightweight alternative to tailoring extreme-scale language models.