bratao

Unfortunately it is not Opensource yet - https://github.com/fireducks-dev/fireducks/issues/22

SPANN is also implemented in the open-source Vespa.ai

Thanks for sharing it. This is something that I noticed. Office Working and University are very good opportunities to become more social and have a dosis of interaction. Choosing their own hobbies while sounds easy, is something that we keeping postponing.

Feels like we are about to get a new wave of large layoffs. Things like Intel last week and Berkshire Hathaway sells off... Brace for impact!

Found on https://twitter.com/aidan_mclau/status/1818071890755469365

My feeling is that the market was down and falling, but AI gave the tech and startup market a lifeline with investments starting to happen again. However, the question is whether the AI market really is big or just a few companies will win and the rest really were in a bubble.

We have been using https://github.com/rushter/selectolax as a faster alternative to BeautifulSoup with html5lib because many malformed webpages in the wild don't work with lxml.

The next versions of CPython are expanding in two very important areas for speed. The free-threading (NoGIL) project will allow true multithreading support and the Faster CPython is trying to speedup Python and are developing a JIT. Exciting times!

When I see things like Bitcoin price and GameStop stock I guess if we still have too much money on the economy.

This is sad. I hope that this is not the beginning of a (much larger) escalation cycle. Multiple reports says that when those drones get near, Iran will launch multiple Ballistic and Cruise missile that are much faster, with the objective of saturating Israel Defenses.

I´m an outsider, but most demand in the last months have been GPUs and AI accelerators. And Intel foundries are not making them yet, only the ARC that are not very strong.

+1 endorsement for me. I run a kubernetes cluster on a set of dedicated servers for some years now with Rancher and love it. All my team are very productive with it!!

The BlackRock ETF don´t is completely against the Bitcoin principles? It is a megacorp centralizing it.

And Latam has a huge fleet of Airbus! It is worrying to see so many incidents with Boeing.

The title is wrong. This is a newer version of Code Llama announcement(I didn't found the model yet). Not a Llama 3!

> The Tesla v12 software update is expected to introduce what CEO Elon Musk has been calling “end-to-end neural nets”. The biggest difference with previous FSD updates is that the vehicle’s controls would now be handled by neural nets rather than being coded by programmers

This sounds very scary to me. Neural Networks have great results on many classification tasks, however they "hallucinate" on completely random and/or easy inputs in an unpredictable way. I already delivered hundreds of classifiers to production in my career and I always had to safeguard inputs for cases I knew the answers or it was wrong. Production ML is always hybrid with hand-made rules. I hope this is just a marketing-term and not something forced to please Musk.

It just compares to itself and AMD. It is missing the comparison to GPUs

The hardware looks amazing. But I imagine from my experience with Asus that the software is subpar.

2.7B size with better performance than Mistral 7B is impressive!!

Just one observation. This model seems "less" multilingual than LLAMA and Mistral. I always ask in Portuguese "What is the capital of Bahia?"/ "Qual a capital da Bahia?" and DeciLM answered wrong and in English ( it said it was Brasilia, the right answer is Salvador). The other models replied correctly.

And this uses Hyena, that can be considered a "previous generation" of Mamba. I think that this anwsers the question about the scalability of SSM and the transformer finally found an opponent.

I don´t think so. There are few algorithms that cover the majority of use. For example in NLP a fast LLM (such as LLAMA) training/prediction and support for Hugging face transformers (BERT and such) would cover 95%+ of NLP use-cases that need a GPU.

Do anyone know what Q* is?

I fully agree that the work load should reach an optimal work/life balance. But how this is possible if someone is taking two extra full time jobs on top of it?

I'm curious as to how this would work in practice. For instance, if an individual is able to complete their sprint while juggling two other full-time jobs, does this imply they should have taken on more tasks?

This approach seems to only be effective if one team member significantly outperforms the rest. If a company's sprint is easily manageable, does this suggest that the team size should be reduced, potentially leading to layoffs?

For software development, quantifying output or defining a finite set of tasks is almost impossible. Unlike traditional industries where tasks can be standardized and expected output can be clearly defined, software development is a different beast altogether.

Speaking from personal experience, my task list is filled with an infinite number of features to implement and bugs to fix. It's a dynamic landscape that changes with every new technology, user feedback, or market demand. It's not something that can be completed within a set timeframe, rather it's a continuous process of improvement and adaptation.

OpenAI (https://fortune.com/2023/05/05/openai-ceo-sam-altman-remote-...) is a example of a high performing company where remote work is an exception. That's why I say that there is no cake recipe that guarantees that remote or in-person work is better than the other. Each company has its own culture and employees have to look for the culture that makes the most sense for them.

I feel like everyone is pointing out and screaming to AMD what they should do to get money, and they are still struggling to find it out.

The GIL only protect the Interpreter. The only thing it may do is to make it infrequently. There are MANY threading bugs in actual Python code.

In my opinion, Python needs to have an urgent answer to the use of multiple cores. AMD just launched a CPU with 96 cores. Today the use on multiple cores is through the use multiprocess, which has many limitations. I understand that the multiple interpreters could come with something like Goroutines, but I still like the real multi thread option more.