This is the story of how I, a PhD student, owe
@OpenAI
$5,000 dollars and might lose access to my main tool of work.
This is a lesson for others building AI apps, but one that has not been resolved yet.
It all starts with a viral app in 2022 and a GPT-4 key in 2023.
終於搞好用 Google SEO 官方資料來餵 Google NotebookLM 了,也是走了好多坑呀。
- 文件上傳不能太長
- 一次只能 20 個文件
- 引用來源是 google doc,可是文件上傳雲端必須先手動開啟然後儲存成 google doc 形式(後來用 Google driver api 解決)
看起來整體效果還不錯,我餵 json
▎CoDoC:醫療領域專用的 AI 工具
Google Deepmind 是非常有指標性的 AI 公司,所以其研究是非常值得參考的。
Google DeepMind 和 Google Research 在《自然醫學》雜誌上發表了聯合論文,提出了CoDoC(基於互補性的臨床工作流程推遲)。
🤔 CoDoC 是什麼?
想像一下,AI
The craziness of the last few days inspired me to share how to raise from top-tier investors including Sam Altman and keep control.
Being a founder is hard. Rewind can help make it a bit easier. Starting today, founders get 1 year for free:
A very useful trick with OpenAI Functions to coerce the model to always return structured JSON:
Create a function called `print` that is called with the typed parameters of the actual <things> you want to be generated and describe it such that it is always executed. Then grab
Anthropic releases Claude 2
blog:
Claude 2 has improved performance, longer responses, and can be accessed via API as well as a new public-facing beta website, . We have heard from our users that Claude is easy to converse with,
【TryOnDiffusion:快速試衣】科普時間
TryOnDiffusion 是 Google 近期發佈的試衣模型,用來模擬穿搭真是太合適了。
需要用到使用者的一張全身照以及穿著該衣服的模特兒照片,然後透過 Diffusion(擴散)原理來進行合成,等等....擴散是什麼?
首先,我們假設這個 AI 模型是個超級強大的畫家。
We've just launched fine-tuning for GPT-3.5 Turbo! Fine-tuning lets you train the model on your company's data and run it at scale. Early tests have shown that fine-tuned GPT-3.5 Turbo can match or exceed GPT-4 on narrow tasks: