Post #6 623 Mar 2, 2023, 00:17 UTC 巴黎理工大学的 Marc Lelarge 教授准备的手把手深度学习教程,有不少是用 Julia 写的。 https://dataflowr.github.io/#ml #mooc 👍 1
Post #5 698 Mar 1, 2023, 12:31 UTC Unum 使用四块 RTX 3090 显卡就能训练一个比 OpenAI 还要好的 Vision-Language transformer 模型 -- UForm。要知道 OpenAI 用了 1024块 A100 显卡才训练出来。UForm 不仅各方面表现优于 OpenAI 的 CLIP,而且推算速度也大大优于 CLIP。对了,Unum 是一家在亚美尼亚的初创公司,目前只有13个人。https://www.unum.cloud/blog/2023-02-20-efficient-multimodality#ml www.unum.cloud Beating OpenAI CLIP with 100x less data and compute | Unum Blog Demonstrating how to train state-of-the-art Vision-Language transformers for semantic search using 100x less data and compute than OpenAI CLIP.
Post #4 399 Mar 1, 2023, 10:05 UTC 一键安装 LLaMA 之后,在一台 M1 Mac Mini 上跑起了 7B 的模型,速度还OK。大概吃了4G 内存。这台是 16G 内存,8核的 M1 CPU。
Post #3 382 Mar 1, 2023, 08:25 UTC System Design Interview 作者 Alex Xu 的 newsletterhttps://blog.bytebytego.com/#system Bytebytego ByteByteGo Newsletter | Alex Xu | Substack Explain complex systems with simple terms, from the authors of the best-selling system design book series. Join over 1,000,000 friendly readers. Click to read ByteByteGo Newsletter, a Substack publication.
Post #2 401 Feb 28, 2023, 18:00 UTC Pandas 2.0 将逐步适用 arrow 取代目前的 Numpy 来存储数据。读写性能及处理速度将大为提升。https://datapythonista.me/blog/pandas-20-and-the-arrow-revolution-part-i#python #data datapythonista blog pandas 2.0 and the Arrow revolution (part I) Introduction At the time of writing this post, we are in the process of releasing pandas 2.0. The project has a large number of users,...