REVIEW 7 cited by
IterGen: Iterative Semantic-aware Structured LLM Generation with Backtracking
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
Large Language Models (LLMs) are widely used for tasks such as natural language and code generation, but their outputs often suffer from issues like hallucination, toxicity, and incorrect results. Current libraries for structured LLM generation rely on left-to-right decoding without support for backtracking, limiting the ability to correct or refine outputs mid-generation. To address this, we introduce IterGen, a user-friendly library for iterative, grammar-guided LLM generation that enables users to move both forward and backward within the generated output based on grammar symbols. By leveraging a symbol-to-position mapping and maintaining the key-value (KV) cache state, IterGen ensures efficient and structured generation while allowing for corrections during the process. We demonstrate IterGen's effectiveness in two important applications: reducing privacy leakage in LLM outputs and improving the accuracy of LLM-generated SQL and Vega-Lite queries. Our code and additional resources are available at https://structuredllm.com.
Forward citations
Cited by 7 Pith papers
-
Visualizing Graph-to-Answer Mechanism Recovery in Materials-Science Hypothesis Generation
Under graph corruption, mechanism recovery in Graph-PRefLexOR-8B concentrates in late synthesis and answer-start layers 30 and 36, not in transition layers 7 to 10.
-
The Format Tax
Structured-output instructions alone impose a large accuracy tax on open-weight LLMs; decoupling freeform reasoning from formatting recovers most of it, while recent closed models largely avoid the tax.
-
Componentization: Decomposing Monolithic LLM Responses into Manipulable Semantic Units
Generative model outputs can be decomposed into typed, linkable components that users edit, toggle, and regenerate before recomposition, as implemented in the MAODchat prototype.
-
CRANE: Reasoning with constrained LLM generation
CRANE switches between unconstrained and constrained LLM decoding, increasing functional accuracy on GSM-symbolic and FOLIO while keeping outputs syntactically valid.
-
FASTGEN: Fast and Cost-Effective Synthetic Tabular Data Generation with LLMs
FASTGEN uses an LLM to infer per-field distributions and generate reusable Python sampling scripts, cutting token cost by 60x at 10,000 records while approximately matching direct-generation quality on several metrics.
-
Exploring the Landscape of Text-to-SQL with Large Language Models: Progresses, Challenges and Opportunities
A systematic review organizing LLM-based text-to-SQL methods into pre-processing, in-context learning, fine-tuning, and post-processing paradigms, with a catalog of datasets, metrics, challenges, and future directions.
-
Translating Federated Learning Algorithms in Python into CSP Processes Using ChatGPT
A ChatGPT-based process, with manual error fixing, translated two PTB-FLA Python algorithms into CSP# models that pass PAT verification.
Discussion (0). Continue with ORCID to comment.