Here are the contents in English:

  1. Visual Fine-Tuning: This is a new feature that lets models excel at visual search, object detection, autonomous driving, smart city management, and medical image analysis by formatting image datasets and uploading them to the platform.

  2. Prompt Optimization: A new feature that allows users to describe what they want the model to achieve, and Playground will automatically generate prompts, effective functions, and structured outputs to quickly turn ideas into working prototypes.

These new features aim to boost developer efficiency and interest in using OpenAI API. However, some developers remain curious about other platforms like Claude, which seem to optimize results faster and more cost-effectively with pseudo-code prompts.

Translation

OpenAI在最近的开发者大会上宣布了几项新的功能，其中包括：

  1. 视觉微调：这是一个让模型更擅长视觉搜索、物体检测、自动驾驶、智慧城市管理和医学影像分析等领域的功能。通过准备格式化的图像数据集并上传到平台，开发者可以显著提高模型的视觉任务性能。

  2. 提示词优化：这是一个新功能，允许用户描述想通过模型实现的功能，然后Playground会自动生成提示词以及有效的函数和结构化输出，以快速将想法转化为实用的原型。

这些新功能旨在提高开发者利用OpenAI API的效率和兴趣。然而，有些开发者仍然表现出对其他平台如Claude的好奇，因为它似乎能通过伪代码提示优化结果，速度更快、成本更低。这也许是因为OpenAI最近的研发进展似乎停滞了一段时间。

所以，这次发布会上新功能能否留住或者吸引更多的开发者还需要观察和等待。

Reference:

https://openai.com/devday/; https://www.youtube.com/watch?v=4LvYft8vb2I

<
Previous Post
Liquid-40B (MIT)
>
Next Post
OpenAI Canvas