REVIEW 28 cited by
Controllable Text Generation for Large Language Models: A Survey
Not yet reviewed by Pith; the record is open.
This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.
SPECIMEN: schema-true, not a live event
T0 review · schema-true
One-sentence machine reading of the paper's core claim.
pith:XXXXXXXX · record.json · timestamp
read the original abstract
In Natural Language Processing (NLP), Large Language Models (LLMs) have demonstrated high text generation quality. However, in real-world applications, LLMs must meet increasingly complex requirements. Beyond avoiding misleading or inappropriate content, LLMs are also expected to cater to specific user needs, such as imitating particular writing styles or generating text with poetic richness. These varied demands have driven the development of Controllable Text Generation (CTG) techniques, which ensure that outputs adhere to predefined control conditions--such as safety, sentiment, thematic consistency, and linguistic style--while maintaining high standards of helpfulness, fluency, and diversity. This paper systematically reviews the latest advancements in CTG for LLMs, offering a comprehensive definition of its core concepts and clarifying the requirements for control conditions and text quality. We categorize CTG tasks into two primary types: content control and attribute control. The key methods are discussed, including model retraining, fine-tuning, reinforcement learning, prompt engineering, latent space manipulation, and decoding-time intervention. We analyze each method's characteristics, advantages, and limitations, providing nuanced insights for achieving generation control. Additionally, we review CTG evaluation methods, summarize its applications across domains, and address key challenges in current research, including reduced fluency and practicality. We also propose several appeals, such as placing greater emphasis on real-world applications in future research. This paper aims to offer valuable guidance to researchers and developers in the field. Our reference list and Chinese version are open-sourced at https://github.com/IAAR-Shanghai/CTGSurvey.
Forward citations
Cited by 28 Pith papers
-
IFHierBench: Hierarchical Instruction Following for Large Language Models
IFHierBench, a 600-prompt benchmark with deterministic per-scope checkers, finds strong LLMs pass only ~54% of hierarchical instruction-following prompts and fail more as nesting deepens.
-
Aligning Language Models with Selective Prediction
RLSR aligns LLMs via a lifted AURC reward and batch ranking inside GRPO, producing better risk-coverage curves than accuracy- or calibration-based RL on in- and out-of-domain tasks.
-
OptiVerse: A Comprehensive Benchmark towards Optimization Problem Solving
OptiVerse is a new benchmark spanning neglected optimization domains that shows LLMs suffer sharp accuracy drops on hard problems due to modeling and logic errors, with a Dual-View Auditor Agent proposed to improve pe...
-
From Recall to Forgetting: Benchmarking Long-Term Memory for Personalized Agents
Memora benchmark and FAMA metric show that LLMs and memory agents frequently reuse invalid memories and struggle to reconcile evolving information in long-term interactions.
-
BiST: A Gold Standard Bangla-English Bilingual Corpus for Sentence Structure and Tense Classification with Inter-Annotator Agreement
BiST is a curated Bangla-English corpus of 30,534 sentences with annotations for syntactic structure and tense, achieving Fleiss Kappa scores of 0.82 and 0.88.
-
Multi-Objective Exploration and Preference Optimization via Mutual Information
Maximizing joint conditional mutual information I(Y; C_Z, W, Z | X) decomposes multi-objective LLM alignment into preference-specific DPO terms plus an I(Y;W|X) exploration term that reduces reward-distribution overlap.
-
Do Sparse Autoencoders Learn Meaningful Concept Hierarchies?
Sparse autoencoders provide a basis for sensible concept hierarchies on visual data but are undermined by hard and soft feature absorption.
-
Answer Engineering: Local Trajectory Editing for Protocol-Constrained Decision Making in Large Language Models
Answer Engineering uses local trajectory editing during autoregressive generation to raise protocol compliance on a clinical SSNHL benchmark from 25.1% to 83.5% and balanced accuracy from 42.0% to 80.7%.
-
Meta-Aligner: Bidirectional Preference-Policy Optimization for Multi-Objective LLMs Alignment
Meta-Aligner introduces a meta-learner network that produces dynamic preference weights to enable bidirectional optimization between preferences and LLM policy responses for multi-objective alignment.
-
Dual-Cluster Memory Agent: Resolving Multi-Paradigm Ambiguity in Optimization Problem Solving
DCM-Agent improves LLM performance on multi-paradigm optimization problems by 11-21% via dual-cluster memory construction and dynamic inference guidance.
-
Universally Empowering Zeroth-Order Optimization via Adaptive Layer-wise Sampling
AdaLeZO uses a non-stationary multi-armed bandit to adaptively allocate perturbation budget across layers in zeroth-order optimization and applies inverse probability weighting to reduce variance while preserving unbi...
-
Escaping the BLEU Trap: A Signal-Grounded Framework with Decoupled Semantic Guidance for EEG-to-Text Decoding
SemKey predicts four semantic attributes from EEG and conditions a frozen LLM on them, beating prior decoders on new semantic-alignment metrics while leaving true word-level accuracy low (2.7% content recall).
-
Escaping the BLEU Trap: A Signal-Grounded Framework with Decoupled Semantic Guidance for EEG-to-Text Decoding
SemKey decouples semantic objectives to ground EEG-to-text generation in neural signals, eliminating hallucinations on noise and improving results on retrieval accuracy and Fréchet distance metrics.
-
A Mixture of Linear Corrections Generates Secure Code
An inference-time mixture of linear correction vectors, derived from linear probes on LLM hidden states, improves the security and functionality of code generated by Qwen2.5-Coder and CodeLlama models.
-
Towards Fast Domain Adaptation and Fine-Grained User Simulation for Evaluating Conversational Recommender Systems
AdaptSim is an adaptive user simulator for CRS evaluation that combines automatic prompt generation, open actions, controlled text generation, and BFS-based pairwise comparison to produce realistic dialogues and asses...
-
Benchmarking EngGPT2-16B-A3B against Comparable Italian and International Open-source LLMs
EngGPT2MoE-16B-A3B matches or exceeds other Italian open-source LLMs on most international benchmarks while remaining competitive on ITALIC, though it trails some top international models.
-
Dual-Cluster Memory Agent: Resolving Multi-Paradigm Ambiguity in Optimization Problem Solving
DCM-Agent improves LLM optimization solving by 11–21% on seven benchmarks via dual-cluster memory of Approaches, Checklists, and Pitfalls plus adaptive path switching.
-
When Models Meet Users: An Empirical Study of Perceptions of General LLMs and Multimodal LLMs on Hugging Face
Across 662 annotated Hugging Face threads, gated access (dominated by Llama), multimodal generation quality, and deployment/invocation complexity are the most prominent user concerns.
-
A Survey on Evaluating Quality and Trustworthiness in LLM-Generated Data
A metric-oriented survey that classifies intrinsic quality and trustworthiness metrics for LLM-generated data across six modalities and documents systematic evaluation gaps in the current literature.
-
How Instruction-Tuning Imparts Length Control: A Cross-Lingual Mechanistic Analysis
Instruction-tuned Llama 3.1 controls word count far better than the base model, and attribution scores point to later layers, but the scoring rule mishandles outputs that are too short.
-
Multi-Objective Exploration and Preference Optimization via Mutual Information
MI-EPO maximizes joint conditional mutual information among responses, feedback, and preference vectors, using probabilistic routing to improve alignment and controllability in multi-objective LLM optimization.
-
Position: AI Safety Requires Effective Controllability
Position paper claiming that AI safety requires explicit runtime controllability and introducing ControlBench to demonstrate gaps in existing alignment methods.
-
When Models Meet Users: An Empirical Study of Perceptions of General LLMs and Multimodal LLMs on Hugging Face
Hugging Face discussions show that access barriers, output quality, and setup complexity are the main user concerns for both general and multimodal LLMs.
-
A Survey on Generative Model Unlearning: Fundamentals, Taxonomy, Evaluation, and Future Direction
A survey and framework that categorizes generative model unlearning by point-wise versus concept-wise objectives, parameter-based versus non-parametric methods, and completeness/utility/efficiency evaluation.
-
MedReadCtrl: Personalizing medical text generation with readability-controlled instruction learning
MedReadCtrl instruction-tunes LLaMA3 to control readability at 12 grade levels, reporting lower readability errors than GPT-4 and higher content scores on unseen clinical simplification.
-
Benchmarking EngGPT2-16B-A3B against Comparable Italian and International Open-source LLMs
EngGPT2MoE-16B-A3B matches or beats other Italian models on most international benchmarks but trails top international models such as GPT-5 nano and Qwen3-8B.
-
Measuring and Mitigating Toxicity in Large Language Models: A Comprehensive Replication Study
DExperts blocks explicit toxicity at 100% but drops to 98.5% on implicit hate speech while increasing generation latency by roughly 10x.
-
Measuring and Mitigating Toxicity in Large Language Models: A Comprehensive Replication Study
DExperts reaches 100% safety on explicit toxicity benchmarks but only 98.5% on implicit hate speech from ToxiGen while imposing a 10x latency increase on GPT-2.
Discussion (0). Sign in to comment.