requirements.txt (or “requirements in setup() call”) is not a valid way to manage dependencies — and it hasn’t been for the past 5+ years. If you are still using requirements.txt it shows you need professional help. luckily, i’m a professional.

Let’s go over some bad / good / example practices for living your best python life in 2024.

A vector database indexes and stores vector embeddings for fast retrieval and similarity search, with capabilities like CRUD operations, metadata filtering, horizontal scaling, and serverless.

Touchscreen devices, ubiquitous in humans' day-to-day life, offer a promising avenue for animal enrichment. With advanced cognitive abilities, keen visual perception, and adeptness to engage with capacitive screens using dexterous tongues, parrots are uniquely positioned to benefit from this technology. Additionally, pet parrots often lack appropriate stimuli, supporting the need for inexpensive solutions using off-the-shelf devices. However, the current human-centric interaction design standards of tablet applications do not optimally cater to the tactile affordances and ergonomic needs of parrots. To address this, we conducted a study with 20 pet parrots, examining their tactile interactions with touchscreens and evaluating the applicability of existing HCI interaction models. Our research highlights key ergonomic characteristics unique to parrots, which include pronounced multi-tap behavior, a critical size threshold for touch targets, and greater effectiveness of larger targets over closer proximity. Based on these insights, we propose guidelines for tablet-based enrichment systems for companion parrots.

The Village Ukraine has put together a list of films to introduce foreign viewers to contemporary Ukrainian cinema. Our intention is to draw our readers’ attention to films that they can legally stream from abroad, rather than compiling a definitive “best of” list.

Yes, writing is hard. But if you can first grasp the origins and qualities of bad writing, you may learn to diagnose and cure problems in your own prose (keeping things simple helps a lot). Similarly heartening is the observation that most first drafts are second-rate, so becoming a skilled rewriter is the thing. These five works are excellent sources of insight and inspiration.

BIMI is a way to verify information about your brand. Like DMARC, DKIM, and SPF—three methods for verifying sender information—BIMI is a text record that lives on your servers. In fact, it works right alongside SPF, DMARC, and DKIM to signal to email clients that you are you. As such, BIMI aids in deliverability, too.

spaCy is a modern Python library for industrial-strength Natural Language Processing. In this free and interactive online course, you'll learn how to use spaCy to build advanced natural language understanding systems, using both rule-based and machine learning approaches.

There are many tools that we can use for SSO, such as Authelia, Authentik, or Keycloak.

Although some of the aforementioned SSO tools may be easier to set up, I decided to go with Keycloak. Keycloak is an enterprise-level tool that is supported by Redhat. Using Keycloak will give us a lot of flexibility, and ticks the boxes for acceptable memory usage, theme-ability, and multi-factor authentication support.

Russian president Vladimir Putin wants you to believe that NATO is responsible for his February 24 invasion of Ukraine—that rounds of NATO enlargement made Russia insecure, forcing Putin to lash out. This argument has two key flaws. First, NATO has been a variable and not a constant source of tension between Russia and the West. Moscow has in the past acknowledged Ukraine’s right to join NATO; the Kremlin’s complaints about the alliance spike in a clear pattern after democratic breakthroughs in the post-Soviet space. This highlights a second flaw: Since Putin fears democracy and the threat that it poses to his regime, and not expanded NATO membership, taking the latter off the table will not quell his insecurity. His declared goal of the invasion, the “denazification” of Ukraine, is a code for his real aim: antidemocratic regime change.

Abstract We introduce MusicLM, a model generating high-fidelity music from text descriptions such as "a calming violin melody backed by a distorted guitar riff". MusicLM casts the process of conditional music generation as a hierarchical sequence-to-sequence modeling task, and it generates music at 24 kHz that remains consistent over several minutes. Our experiments show that MusicLM outperforms previous systems both in audio quality and adherence to the text description. Moreover, we demonstrate that MusicLM can be conditioned on both text and a melody in that it can transform whistled and hummed melodies according to the style described in a text caption. To support future research, we publicly release MusicCaps, a dataset composed of 5.5k music-text pairs, with rich text descriptions provided by human experts.