CRANE: Reasoning with constrained LLM generation

By Debangshu Banerjee, Tarun Suresh, Shubham Ugare, Sasa Misailovic, Gagandeep Singh. 2025

Subfields: Large Language Models, Prompting & Reasoning, Natural Language Generation

Citations in the corpus

• CoRNStack: High-Quality Contrastive Data for Better Code Retrieval and Reranking (2024, unknown): “The rapid advancement of software development has led to an increased reliance on automated tools for code generation (Chen et al., 2021; Li et al., 2022b; Nijkamp et al.; Ugare et al., 2024b; Banerjee et al., 2025; Ugare et al., 2024a; Suresh et al., 2025).”
• Is The Watermarking Of LLM-Generated Code Robust? (2024, unknown): “…et al., 2021; OpenAI, 2023; Suresh et al., 2024b; Reddy et al., 2025; Wei et al., 2025b;a;d;c), especially when combined with constrained decoding algorithms (Ugare et al., 2024c;a; Banerjee et al., 2025; Suresh et al., 2025), which improve the syntactic and semantic correctness of the LLM output.”
• CodeARC: Benchmarking Reasoning Capabilities of LLM Agents for Inductive Program Synthesis (2025, unknown): “…al., 2021; Liu et al., 2023b), APPS (Hendrycks et al., 2021), and others (Li et al., 2022; Liu et al., 2023c; Li et al., 2023; Jain et al., 2024a; Banerjee et al., 2025; Ni et al., 2024b; Liu et al., 2023d; Du et al., 2023; Suresh et al., 2025b; Zhuo et al., 2024; Yin et al., 2022; Lai et al.,…”