[Feb 18] - Interesting things I learnt this week

A few interesting things I learnt about this week are

1. Coffee Stains in LaTeX - This package provides an essential feature that LaTeX has been missing for too long: It adds coffee stains to your documents. A lot of time can be saved by printing stains directly on the page rather than adding them manually.

My Take: There is too much fun to be had with LaTeX. I guess the rest of the world missed this wonderful piece of tech completely. These small and sometimes pointless tools are what really make any technology platform stand out among others, not because of the tech but the fact that its users are emotionally invested to build these kind of tech nuggets

2. Meshtastic - Meshtastic is a project that enables you to use inexpensive LoRa radios as a long range off-grid communication platform in areas without existing or reliable communications infrastructure. This project is 100% community driven and open source. LoRa uses license-free sub-gigahertz radio frequency bands EU868 (863–870/873 MHz) in Europe; AU915/AS923-1 (915–928 MHz) in South America; US915 (902–928 MHz) in North America; IN865 (865–867 MHz) in India; and AS923 (915–928 MHz) in Asia

My Take: This is an interesting piece of technology. I am absolutely amazed by it and the fact that it's available and approved for use in India, I just can't wait to try it after finding out about it. I have a friend who is almost 30 km away from me. He too is interested, I just hope we both can somehow communicate over it, not that phone calls are too much work. Let me know if you are interested too.

3. Stable Audio - Team from Stability AI has introduced Stable Audio which is based on latent diffusion, with its latent defined by a fully-convolutional variational autoencoder. It is conditioned on text prompts as well as timing embeddings, allowing for fine control over both the content and length of the generated music and sounds. Stable Audio is capable of rendering stereo signals of up to 95 sec at 44.1kHz in 8 sec on an A100 GPU. Despite its compute efficiency and fast inference, it is one of the best in two public text-to-music and -audio benchmarks and, differently from state-of-the-art models, can generate music with structure and stereo sounds.

My Take: Stability AI team has done it with image generation and now with sound generation. I think they are doing a phenomenal job with it. One of the leading GenAI tools, not just in open source. I tried it a little bit and it does seem very awe inspiring.


4. WhiteRabbitNeo - WhiteRabbitNeo is a model series that can be used for offensive and defensive cybersecurity. This 13B model is getting released as a public preview of its capabilities, and also to assess the societal impact of such an AI.

My Take: This is not a new model per say, but an optimisation over the Llama models. I guess this one makes the model a little more unsafe and around the restrictions that might have been around spitting things which enabled offensive cybersecurity. Might end up being a replacement for Script Kiddies of sorts but I don't know much about it, but I am a little scared with its prospects.


Popular posts from this blog

My learnings at Google

Top reasons why Tata Elxsi should be blacklisted

[March 24] Interesting Things I Learnt This Week