# Efficient (Soft) Q-Learning for Text Generation with Limited Good Data

arXiv: https://arxiv.org/abs/2106.07704 · HTML: https://mlsanity.com/paper/2106.07704/

## Key facts

- Cited by 37 papers in the map (data through 14 Oct 2026).
- Builds on 54 of its 73 references that are in the map.
- First posted on arXiv 14 Jun 2021.

## Builds on (most cited first, 30 of 54)

- [Attention Is All You Need](https://mlsanity.com/paper/1706.03762/) (2017)
- [Language Models are Few-Shot Learners](https://mlsanity.com/paper/2005.14165/) (2020)
- [BART: Denoising Sequence-to-Sequence Pre-training for Natural Language Generation, Translation, and Comprehension](https://mlsanity.com/paper/1910.13461/) (2019)
- [Prefix-Tuning: Optimizing Continuous Prompts for Generation](https://mlsanity.com/paper/2101.00190/) (2021)
- [The Power of Scale for Parameter-Efficient Prompt Tuning](https://mlsanity.com/paper/2104.08691/) (2021)
- [Playing Atari with Deep Reinforcement Learning](https://mlsanity.com/paper/1312.5602/) (2013)
- [The Curious Case of Neural Text Degeneration](https://mlsanity.com/paper/1904.09751/) (2019)
- [A Broad-Coverage Challenge Corpus for Sentence Understanding through Inference](https://mlsanity.com/paper/1704.05426/) (2017)
- [A large annotated corpus for learning natural language inference](https://mlsanity.com/paper/1508.05326/) (2015)
- [Google's Neural Machine Translation System: Bridging the Gap between Human and Machine Translation](https://mlsanity.com/paper/1609.08144/) (2016)
- [PEGASUS: Pre-training with Extracted Gap-sentences for Abstractive Summarization](https://mlsanity.com/paper/1912.08777/) (2019)
- [Beyond Accuracy: Behavioral Testing of NLP models with CheckList](https://mlsanity.com/paper/2005.04118/) (2020)
- [Stabilizing Off-Policy Q-Learning via Bootstrapping Error Reduction](https://mlsanity.com/paper/1906.00949/) (2019)
- [Adversarial NLI: A New Benchmark for Natural Language Understanding](https://mlsanity.com/paper/1910.14599/) (2019)
- [Plug and Play Language Models: A Simple Approach to Controlled Text Generation](https://mlsanity.com/paper/1912.02164/) (2019)
- [Universal Adversarial Triggers for Attacking and Analyzing NLP](https://mlsanity.com/paper/1908.07125/) (2019)
- [Is BERT Really Robust? A Strong Baseline for Natural Language Attack on Text Classification and Entailment](https://mlsanity.com/paper/1907.11932/) (2019)
- [Reinforcement Learning with Deep Energy-Based Policies](https://mlsanity.com/paper/1702.08165/) (2017)
- [Sequence Level Training with Recurrent Neural Networks](https://mlsanity.com/paper/1511.06732/) (2015)
- [Self-critical Sequence Training for Image Captioning](https://mlsanity.com/paper/1612.00563/) (2016)
- [GeDi: Generative Discriminator Guided Sequence Generation](https://mlsanity.com/paper/2009.06367/) (2020)
- [A Deep Reinforced Model for Abstractive Summarization](https://mlsanity.com/paper/1705.04304/) (2017)
- [Benchmarking Zero-shot Text Classification: Datasets, Evaluation and Entailment Approach](https://mlsanity.com/paper/1909.00161/) (2019)
- [Deep Reinforcement Learning for Dialogue Generation](https://mlsanity.com/paper/1606.01541/) (2016)
- [Toward Controlled Generation of Text](https://mlsanity.com/paper/1703.00955/) (2017)
- [The E2E Dataset: New Challenges For End-to-End Generation](https://mlsanity.com/paper/1706.09254/) (2017)
- [The GEM Benchmark: Natural Language Generation, its Evaluation and Metrics](https://mlsanity.com/paper/2102.01672/) (2021)
- [Bridging the Gap Between Value and Policy Based Reinforcement Learning](https://mlsanity.com/paper/1702.08892/) (2017)
- [Maximum Entropy RL (Provably) Solves Some Robust RL Problems](https://mlsanity.com/paper/2103.06257/) (2021)
- [Equivalence Between Policy Gradients and Soft Q-Learning](https://mlsanity.com/paper/1704.06440/) (2017)

## Built on by (most cited first, 30 of 37)

- [Amortizing intractable inference in large language models](https://mlsanity.com/paper/2310.04363/) (2023)
- [Probabilistic Inference in Language Models via Twisted Sequential Monte Carlo](https://mlsanity.com/paper/2404.17546/) (2024)
- [Flocks of Stochastic Parrots: Differentially Private Prompt Learning for Large Language Models](https://mlsanity.com/paper/2305.15594/) (2023)
- [A Dense Reward View on Aligning Text-to-Image Diffusion with Preference](https://mlsanity.com/paper/2402.08265/) (2024)
- [Offline Reinforcement Learning for LLM Multi-Step Reasoning](https://mlsanity.com/paper/2412.16145/) (2024)
- [Value Augmented Sampling for Language Model Alignment and Personalization](https://mlsanity.com/paper/2405.06639/) (2024)
- [When LLM Meets DRL: Advancing Jailbreaking Efficiency via DRL-guided Search](https://mlsanity.com/paper/2406.08705/) (2024)
- [PRewrite: Prompt Rewriting with Reinforcement Learning](https://mlsanity.com/paper/2401.08189/) (2024)
- [Alignment of Diffusion Models: Fundamentals, Challenges, and Future](https://mlsanity.com/paper/2409.07253/) (2024)
- [Inference-Time Policy Adapters (IPA): Tailoring Extreme-Scale LMs without Fine-tuning](https://mlsanity.com/paper/2305.15065/) (2023)
- [On the Privacy Risk of In-context Learning](https://mlsanity.com/paper/2411.10512/) (2024)
- [Preference-grounded Token-level Guidance for Language Model Fine-tuning](https://mlsanity.com/paper/2306.00398/) (2023)
- [RL-JACK: Reinforcement Learning-powered Black-box Jailbreaking Attack against LLMs](https://mlsanity.com/paper/2406.08725/) (2024)
- [InPO: Inversion Preference Optimization with Reparametrized DDIM for Efficient Diffusion Model Alignment](https://mlsanity.com/paper/2503.18454/) (2025)
- [Leftover Lunch: Advantage-based Offline Reinforcement Learning for Language Models](https://mlsanity.com/paper/2305.14718/) (2023)
- [MORL-Prompt: An Empirical Analysis of Multi-Objective Reinforcement Learning for Discrete Prompt Optimization](https://mlsanity.com/paper/2402.11711/) (2024)
- [GROVE: A Retrieval-augmented Complex Story Generation Framework with A Forest of Evidence](https://mlsanity.com/paper/2310.05388/) (2023)
- [Can a Bayesian Oracle Prevent Harm from an Agent?](https://mlsanity.com/paper/2408.05284/) (2024)
- [Learning to Plan for Language Modeling from Unlabeled Data](https://mlsanity.com/paper/2404.00614/) (2024)
- [Segmenting Text and Learning Their Rewards for Improved RLHF in Language Model](https://mlsanity.com/paper/2501.02790/) (2025)
- [Reinforcement Learning for Generative AI: A Survey](https://mlsanity.com/paper/2308.14328/) (2023)
- [Self-Guided Process Reward Optimization with Redefined Step-wise Advantage for Process Reinforcement Learning](https://mlsanity.com/paper/2507.01551/) (2025)
- [BRiTE: Bootstrapping Reinforced Thinking Process to Enhance Language Model Reasoning](https://mlsanity.com/paper/2501.18858/) (2025)
- [Contrastive Learning with Counterfactual Explanations for Radiology Report Generation](https://mlsanity.com/paper/2407.14474/) (2024)
- [M2Lingual: Enhancing Multilingual, Multi-Turn Instruction Alignment in Large Language Models](https://mlsanity.com/paper/2406.16783/) (2024)
- [On the Power of (Approximate) Reward Models for Inference-Time Scaling](https://mlsanity.com/paper/2602.01381/) (2026)
- [PIEClass: Weakly-Supervised Text Classification with Prompting and Noise-Robust Iterative Ensemble Training](https://mlsanity.com/paper/2305.13723/) (2023)
- [Anchoring Values in Temporal and Group Dimensions for Flow Matching Model Alignment](https://mlsanity.com/paper/2512.12387/) (2025)
- [Supervised Optimism Correction: Be Confident When LLMs Are Sure](https://mlsanity.com/paper/2504.07527/) (2025)
- [How to Auto-optimize Prompts for Domain Tasks? Adaptive Prompting and Reasoning through Evolutionary Domain Knowledge Adaptation](https://mlsanity.com/paper/2510.21148/) (2025)

## Abstract (arXiv)

Maximum likelihood estimation (MLE) is the predominant algorithm for training text generation models. This paradigm relies on direct supervision examples, which is not applicable to many emerging applications, such as generating adversarial attacks or generating prompts to control language models. Reinforcement learning (RL) on the other hand offers a more flexible solution by allowing users to plug in arbitrary task metrics as reward. Yet previous RL algorithms for text generation, such as policy gradient (on-policy RL) and Q-learning (off-policy RL), are often notoriously inefficient or unstable to train due to the large sequence space and the sparse reward received only at the end of sequences. In this paper, we introduce a new RL formulation for text generation from the soft Q-learning (SQL) perspective. It enables us to draw from the latest RL advances, such as path consistency learning, to combine the best of on-/off-policy updates, and learn effectively from sparse reward. We apply the approach to a wide range of novel text generation tasks, including learning from noisy/negative examples, adversarial attacks, and prompt generation. Experiments show our approach consistently outperforms both task-specialized algorithms and the previous RL methods.