# Pointer-Augmented Autoregressive Generation of Patent Claims with Joint Topology and Content Decoding

arXiv: https://arxiv.org/abs/2607.24040 · HTML: https://mlsanity.com/paper/2607.24040/

## Key facts

- Cited by 1 papers in the map (data through 14 Oct 2026).
- Builds on 21 of its 36 references that are in the map.
- First posted on arXiv 27 Jul 2026.

## Builds on (most cited first, 21 of 21)

- [Training language models to follow instructions with human feedback](https://mlsanity.com/paper/2203.02155/) (2022)
- [RoBERTa: A Robustly Optimized BERT Pretraining Approach](https://mlsanity.com/paper/1907.11692/) (2019)
- [Direct Preference Optimization: Your Language Model is Secretly a Reward Model](https://mlsanity.com/paper/2305.18290/) (2023)
- [BERTScore: Evaluating Text Generation with BERT](https://mlsanity.com/paper/1904.09675/) (2019)
- [A General Theoretical Paradigm to Understand Learning from Human Preferences](https://mlsanity.com/paper/2310.12036/) (2023)
- [KTO: Model Alignment as Prospect Theoretic Optimization](https://mlsanity.com/paper/2402.01306/) (2024)
- [Is DPO Superior to PPO for LLM Alignment? A Comprehensive Study](https://mlsanity.com/paper/2404.10719/) (2024)
- [Grammar-Constrained Decoding for Structured NLP Tasks without Finetuning](https://mlsanity.com/paper/2305.13971/) (2023)
- [Language to Logical Form with Neural Attention](https://mlsanity.com/paper/1601.01280/) (2016)
- [Pretrained Language Models for Text Generation: A Survey](https://mlsanity.com/paper/2201.05273/) (2022)
- [Coarse-to-Fine Decoding for Neural Semantic Parsing](https://mlsanity.com/paper/1805.04793/) (2018)
- [Abstract Syntax Networks for Code Generation and Semantic Parsing](https://mlsanity.com/paper/1704.07535/) (2017)
- [Patent Claim Generation by Fine-Tuning OpenAI GPT-2](https://mlsanity.com/paper/1907.02052/) (2019)
- [Stack-Pointer Networks for Dependency Parsing](https://mlsanity.com/paper/1805.01087/) (2018)
- [The Harvard USPTO Patent Dataset: A Large-Scale, Well-Structured, and Multi-Purpose Corpus of Patent Applications](https://mlsanity.com/paper/2207.04043/) (2022)
- [Can Large Language Models Generate High-quality Patent Claims?](https://mlsanity.com/paper/2406.19465/) (2024)
- [PatentScore: Multi-dimensional Evaluation of LLM-Generated Patent Claims](https://mlsanity.com/paper/2505.19345/) (2025)
- [Pap2Pat: Benchmarking Outline-Guided Long-Text Patent Generation with Patent-Paper Pairs](https://mlsanity.com/paper/2410.07009/) (2024)
- [Enriching Patent Claim Generation with European Patent Dataset](https://mlsanity.com/paper/2505.12568/) (2025)
- [Large Language Model for Patent Concept Generation](https://mlsanity.com/paper/2409.00092/) (2024)
- [FlowPlan-G2P: A Structured Generation Framework for Transforming Scientific Papers into Patent Descriptions](https://mlsanity.com/paper/2601.02589/) (2026)

## Built on by (most cited first, 1 of 1)

- [Benchmarking Patent Drafting from Inventor-Style Disclosures](https://mlsanity.com/paper/2608.21249/) (2026)

## Abstract (arXiv)

Autoregressive decoders emit flat token sequences and cannot enforce hierarchical constraints across output segments, a limitation that becomes acute in patent claim generation, where a claim set forms a dependency forest whose scope must narrow monotonically with depth. Topology and content are mutually dependent: a dependent claim's wording must reflect its parent's scope, yet the parent must be chosen before that wording exists, so neither post-hoc parsing nor grammar-constrained decoding suffices. We propose SPG (Structure-aware Patent Generation), which predicts topology inside the autoregressive pass. A pointer head selects each dependent claim's parent, and its gradients, together with a depth-adaptive scope regularizer, reshape the shared decoder's representations during training. A second stage then applies a violation-weighted preference objective over self-generated deficient candidates, supplying the negative signal that granted-patent corpora lack. On HUPD-DCG, SPG on Llama-3-8B-Instruct recovers 79.0% of gold parent links, a quantity its training reward never supervises, and raises antecedent consistency from 0.292 to 0.478 over a supervised baseline of equal scale, with expert evaluation corroborating these gains.