akumetsu

I was unemployed for 6 months after my first job after college stopped abruptly due to bankruptcy & corruption of upper management. Ended up going back to university to get a grad degree on top, then spent 5 months unemployed after that. Ended up joining a small startup some friends were building, wasn't much fun. Left after a year, spend another 8 months unemployed then landed my dream job through dumb luck. In all of those periods, I tried to cope via lots of applications or lost myself in spirals of reduced social contacts and lots of video games. Worked on some personal tech projects, dabbled in robotics, tried to help write a book... I am very happy with and grateful for where I am right now. At the same time, reading the posts here brought back strong emotions & memories. My heart goes out to all of you searching, don't let it get to you! You are all valuable human beings, and life always involves some element of luck. Plus all of recruiting is very dehumanising and degrading, no-one should be ghosted, ignored or lied to on this scale. Things get better eventually and it's ok to feel down and powerless, just remember you are not alone in this!

"We find that Claude has remarkable resource efficiency – the degree to which the quality of the translation model depends on a language pair’s resource level.”

It looks like the main takeaway is that Claude 3 (Opus) is a lot better at low resource language pairs compared with other LLMs and Google Translate. While this is not as important for the typical english usage of Google Translate it promises way better usage of LLMs as simple translators for languages with fewer speakers or just less data available.

Sad to hear, I remember the excitement over the experimental evidence once his particle was detected. I'm always amazed by theoretical predictions that can actually be verified plus it was interrsting to hear about the higgs boson as part of my studies shortly after it was detected. Nowadays it seems many theoretical predications are not even close to being verifiable in the coming years or with the current and planned tech. Unless we are talking about superconductivity at room temperature ofc

On my first day of my first job I was led around the office and introduced to my new colleagues, who were all sitting at their workstations. Then I was given my own and this was running on it. They all claimed this was the normal look and watched me try to figure out what was happening, lovely memories

LinkedIn gave me a random notification a couple days back, suggesting I verify my account via ID & Persona. I declined and got no changes to my account or further notification, maybe they are currently trying to push this feature on people?

Maybe try their form for "compromised account"?

https://www.linkedin.com/help/linkedin/ask/TS-RHA

Nice how 1 hour turned into 70 minutes.

From personal experience: Learning and understanding the fundamentals via courses and personal projects/implementation is different from keeping up to date with blogs, newsletters etc. I suggest dedicating the full hour to 1 thing, but alternating between courses (i.e coursera), reading (i.e the batch) and building/coding.

Location: Frankfurt, Germany

Remote: Possible

Willing to relocate: Yes

Technologies: Python, AI/DS packages (scipy, pandas, numpy), AI/DS frameworks (PyTorch, Keras, TensorFlow), Git, Docker, Linux, MongoDB, SQL, some Azure/Airflow,/AWS

Résumé/CV: https://drive.google.com/file/d/1QgZVakiLBFiaSqVeS5GonUAh7AY...

LinkedIn: https://www.linkedin.com/in/malte-cordts-328273193/

Email:cordtsmalte[at]gmail[dot]com

Worked a year in industry after my physics BSc as IT Consultant/Forensics with Big Data in an AI team before the company exploded early in the pandemic. Went back to university and recently obtained my physics MSc, worked in 2 startups in the meantime. I am looking for a full time position

The roles you mentioned are not listed on the website.

I am interested, how do I apply?

Cool to see this but I am always surprised how often LLM output is used to train other LLMs. They used gpt3.5 turbo and gpt4 for multiple tasks. Even simple translation of englisch benchmarks and writing German poems to train on, in order to create an LLM that works better in German? AFAIK, this still goes against OpenAI TOS and also the basic idea that training on AI output leads to worse results in general. Was there some major shift in this over the years, or has it simply become the default approach due to it being easy to do? Also nice to finally see something from Hessian.AI, as a local, I heard them talk big more than once but never saw results. I wonder what Aleph Alpha thinks about this, since they want to make "AI made in Europe to challenge OpenAI"

Been a fan of The Batch by Andrew Ng for years

I used to buy&sell mtg cards and some collections, depending on the time you want to invest there are different options: 1. Take everything to a local shop that buys cards, most big cities have one and they buy everything at once, but pay under market value. 2. Ebay, sell it as a single collection 3. cardmarket.com is very good, helps you grade the cards and shows accurate market values, might take time to set everything up and might result in lots of different sales, so lots of Briefen