GPT4Point++: Advancing Unified Point-Language Understanding and Generation.
GPT4Point++:推進統一點雲-語言理解與生成
IEEE Trans Pattern Anal Mach Intell 2025-08-11
GPT4Point 和 GPT4Point++ 是專門用來處理 3D 點雲和文字的多模態 AI 模型,能更準確地辨識、描述和回答 3D 物件相關問題。GPT4Point 用兩階段訓練,GPT4Point++ 則直接端到端訓練,效果更好。作者也推出了大規模 3D 點雲配對資料集 Capverse。實驗證明這些模型在 3D 理解和生成上表現很強。
相關文章PubMedDOI推理
Assessment of Differential Diagnoses for Oculoplastics Cases Produced by Large Language Models.
大型語言模型產生之眼整形病例鑑別診斷的評估
Ophthalmic Plast Reconstr Surg 2025-08-11
A Comparative Study on the Use of DeepSeek-R1 and ChatGPT-4.5 in Different Aspects of Plastic Surgery.
DeepSeek-R1 與 ChatGPT-4.5 在整形外科不同層面應用之比較研究
Aesthetic Plast Surg 2025-08-11
GPT-4 vs. Radiologists: who advances mediastinal tumor classification better across report quality levels? A cohort study.
GPT-4 vs. 放射科醫師:誰能在不同報告品質層級下更好地推進縱膈腫瘤分類?一項世代研究
Int J Surg 2025-08-11
Evaluation of the performance of large language models in endoscopic lumbar surgery: a comparative analysis.
大型語言模型在內視鏡腰椎手術中的表現評估:比較性分析
Ann Med Surg (Lond) 2025-08-11
這項研究比較三款AI在回答內視鏡腰椎手術常見問題的表現。專業醫師認為 Claude 3.5 Sonnet 最專業、內容最優,但一般民眾覺得 ChatGPT o1-preview 最容易懂、最滿意。結果顯示,雖然 Claude 3.5 Sonnet 專業度高,ChatGPT o1-preview 更適合病人衛教,凸顯AI在醫療溝通上的優缺點。
相關文章PubMedDOI推理
Regulating the unseen hand: AI, authorship, and trust in medical science.
調控無形之手:AI、作者身份與醫學科學中的信任
Ann Med Surg (Lond) 2025-08-11
Emerging Brain-to-Content Technologies from Generative AI and Deep Representation Learning.
生成式 AI 與深度表徵學習推動的腦-內容新興技術
IEEE Signal Process Mag 2025-08-11