MK2 at PBIG Competition: A Prompt Generation Solution
Fuente:
arXiv
Gespeichert in:
| Hauptverfasser: | , , , , |
|---|---|
| Format: | Preprint |
| Veröffentlicht: |
2025
|
| Schlagworte: | |
| Online-Zugang: | |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| _version_ | 1866912476223766528 |
|---|---|
| author | Xu, Yuzheng Hirasawa, Tosho Kawano, Seiya Kato, Shota Kozuno, Tadashi |
| author_facet | Xu, Yuzheng Hirasawa, Tosho Kawano, Seiya Kato, Shota Kozuno, Tadashi |
| contents | The Patent-Based Idea Generation task asks systems to turn real patents into product ideas viable within three years. We propose MK2, a prompt-centric pipeline: Gemini 2.5 drafts and iteratively edits a prompt, grafting useful fragments from weaker outputs; GPT-4.1 then uses this prompt to create one idea per patent, and an Elo loop judged by Qwen3-8B selects the best prompt-all without extra training data. Across three domains, two evaluator types, and six criteria, MK2 topped the automatic leaderboard and won 25 of 36 tests. Only the materials-chemistry track lagged, indicating the need for deeper domain grounding; yet, the results show that lightweight prompt engineering has already delivered competitive, commercially relevant ideation from patents. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2507_08335 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | MK2 at PBIG Competition: A Prompt Generation Solution Xu, Yuzheng Hirasawa, Tosho Kawano, Seiya Kato, Shota Kozuno, Tadashi Computation and Language The Patent-Based Idea Generation task asks systems to turn real patents into product ideas viable within three years. We propose MK2, a prompt-centric pipeline: Gemini 2.5 drafts and iteratively edits a prompt, grafting useful fragments from weaker outputs; GPT-4.1 then uses this prompt to create one idea per patent, and an Elo loop judged by Qwen3-8B selects the best prompt-all without extra training data. Across three domains, two evaluator types, and six criteria, MK2 topped the automatic leaderboard and won 25 of 36 tests. Only the materials-chemistry track lagged, indicating the need for deeper domain grounding; yet, the results show that lightweight prompt engineering has already delivered competitive, commercially relevant ideation from patents. |
| title | MK2 at PBIG Competition: A Prompt Generation Solution |
| topic | Computation and Language |
| url | https://arxiv.org/abs/2507.08335 |