{"id": "njw-2f5a81", "key": "brainet-2f5a81", "title": "上海人工智能实验室InternLM3：80亿模型用4万亿词元训练，成本砍掉75%", "author": "脑机网编辑部", "publisher": "脑机网", "date": "2026-11-14", "url": "https://aiai.cafe/a/njw-2f5a81", "section": "AI大模型", "gbt7714": "脑机网编辑部. 上海人工智能实验室InternLM3：80亿模型用4万亿词元训练，成本砍掉75%[EB/OL]. (脑机网)[2026-11-14]. 北京: 脑机网, https://aiai.cafe/a/njw-2f5a81. 2026-11-15.", "apa": "脑机网编辑部. (2026, November 14). 上海人工智能实验室InternLM3：80亿模型用4万亿词元训练，成本砍掉75%. 脑机网. https://aiai.cafe/a/njw-2f5a81", "mla": "\"上海人工智能实验室InternLM3：80亿模型用4万亿词元训练，成本砍掉75%.\" 脑机网编辑部, 脑机网, 14 November 2026, aiai.cafe/a/njw-2f5a81.", "chicago": "\"上海人工智能实验室InternLM3：80亿模型用4万亿词元训练，成本砍掉75%.\" 脑机网编辑部, 脑机网. November 14, 2026.", "bibtex": "@misc{brainet-2f5a81,\n  author       = {{脑机网编辑部}},\n  title        = {{上海人工智能实验室InternLM3：80亿模型用4万亿词元训练，成本砍掉75\\%}},\n  year         = {2026},\n  month        = {November},\n  publisher    = {{脑机网}},\n  howpublished = {\\url{https://aiai.cafe/a/njw-2f5a81}},\n  note         = {Accessed 2026-11-15},\n  language     = {chinese}\n}\n", "ris": "TY  - GEN\nTI  - 上海人工智能实验室InternLM3：80亿模型用4万亿词元训练，成本砍掉75%\nAU  - 脑机网编辑部\nPY  - 2026\nDA  - 2026/11/14\nPB  - 脑机网\nUR  - https://aiai.cafe/a/njw-2f5a81\nLA  - zh\nER  - \n", "exports": {"bibtex": "https://aiai.cafe/a/njw-2f5a81.bib", "ris": "https://aiai.cafe/a/njw-2f5a81.ris", "markdown": "https://aiai.cafe/a/njw-2f5a81.md", "json": "/api/cite/njw-2f5a81"}}