TD-interpreter: enhancing the understanding of timing diagrams with visual-language learning

Counts differ because each database indexes a different set of publications. We treat OpenAlex as the canonical count; Google Scholar is not shown (no API, and crawling it violates its ToS).

We introduce TD-Interpreter, a specialised machine learning (ML) workflow that assists engineers in understanding complex timing diagrams (TDs), originating from a third party, during their design and verification process. The outcome of this workflow is a visual question-answer environment which allows engineers to input a set of TDs and ask design and verification queries regarding these TDs. To address limited training data availability, we developed a synthetic data generation pipeline that aligns visual information with its textual interpretation. We then implemented TD-Interpreter using a lightweight 7B Multimodal Large Language Model (MLLM), guided by both supervised fine-tuning (SFT) and reinforcement learning (RL). We also developed extensive TD-reasoning benchmarks from simple retrieval tasks to complex reasoning problems in the real world. Our experimental results show that TD-Interpreter substantially outperformed both the untuned base model, and commercial MLLMs (GPT-5.2), on the evaluated benchmarks, demonstrating the effectiveness of the proposed workflow.