码奋

码奋

TensorFlow -图像分类

This code imports the necessary libraries for TensorFlow and Keras, loads the Fashion MNIST dataset, and creates a neural network model to classify clothing images. The model, consisting of a Flatten layer and two Dense layers, is compiled and trained for five epochs. It evaluates test images and prints predictions. Additionally, it visualizes the results, showing images with their predicted labels, correct predictions in blue and incorrect in red. Finally, it demonstrates predicting a single image.

python-pyecharts 数据可视化学习 (数据来源:丁香园)

This post details the creation of COVID-19 visualization tools using Python, including nation-wide and Hubei province-specific infection maps, and a growth trend chart. Data is scraped from DXY, an online health community, and visualizations are generated using pyecharts with JSON data. The code snippets illustrate scraping methods, JSON file handling, and chart customization techniques for clearly representing the epidemic’s data.

深入浅出,解析ChatGPT背后的工作原理

ChatGPT,OpenAI发布的新语言模型,在准确度与连贯性上超越了前作GPT-3。它通过结合监督学习与独特的RLHF(人类反馈强化学习)方法来优化,克服了一致性问题,更贴近人类的语言使用方式。ChatGPT的培训包括三个阶段:监督调优、模拟人类偏好和近端策略优化,有效提升了聊天机器人处理自然语言任务的能力。