Papers by Philip Lippmann
Zero-Shot Contextual Embeddings via Offline Synthetic Corpus Generation (2025.findings-emnlp)
Copied to clipboard
| Challenge: | Context-aware embedding methods boost retrieval accuracy by conditioning on corpus statistics extracted from neighboring documents. |
| Approach: | They propose a zero-shot contextual adaptation framework that replaces real corpus access with a one-time offline synthesis of a compact proxy. |
| Outcome: | The proposed framework replaces real corpus access with offline synthesis of a compact proxy. |
Context-Informed Machine Translation of Manga using Multimodal Large Language Models (2025.coling-main)
Copied to clipboard
| Challenge: | Automated manga translation is a promising potential solution, but it is underdeveloped due to the need to incorporate visual elements into the translation process to resolve ambiguities. |
| Approach: | They propose a method that leverages the vision component of multimodal large language models to improve translation quality and evaluate the impact of translation unit size, context length, and propose 'token efficient' approach for manga translation. |
| Outcome: | The proposed method achieves state-of-the-art results for Japanese-English translation and sets a new standard for Japanese and Polish translation. |