I have! I recently compared Gemma 1b to ModernBERT Large for a binary classification task and ModernBERT was the clear winner. It learned faster and performed the task better by a significant margin by the end of training. It seems the bidirectional encoder only architecture works really well for classification tasks, and I think it is related to being bidirectional whereas decoder only models like Gemma (or Qwen) can only “look backwards”. I used a mixture of FFT and LoRA as well as a mixture of CE Loss and SupCon Loss.
Which to be fair isn’t that great in either country regardless if you work in tech or not. In major cities if you rent or have a mortgage (taken out recently) it’s basically poverty level.
Porsche did something similar for the 4 cylinder 718s. There is a microphone, processor and amplifier above the exhaust that feeds a speaker unit in the cabin to “enhance” the sound that is muffled due to the turbo.
It is good for aggressive driving but just creates a bad drone when highway cruising. Good have been implemented much better. Lots of people pull the fuse to disable it.
Omarchy!! shipping a window manager with defaults and or a terminal with a config - what an unspeakable sin.
Linux should be hard and shitty and it should break all the time!
What is this newfound obsession with distros that just works and have some great setups and defaults.
Where do we end up if you can just close your laptop lid or copy paste with the same key or or or if even… gasp… the theme is automatically applied across all apps?
Having literal billionaire best buds backing the project probably helps one gain traction, no? Toss in a dash of weird culture war bullshit, and… well. Distro popularity isn’t exactly a meritocratic system.
Respectfully, in the best case you don't really know what you are talking about. These three cities certainly cannot be categorised as tax havens in any sense of the word, and the local legislation is objectively different - as evident to anyone comparing it.
This reads like an AI writing to a point an author wants to make without there being evidence.
It is completely incoherent. Apparently we just need markdown and git, but also a knowledge graph and pgvector which accounts for most of the performance.
We don’t need semantic search, because we use… hybrid search (semantic search plus bm25)???
Really bad look for an AI consulting company this.
The way this argues against its own premise is really ChatGPT like.
This happens when you ask it writing about something that isn’t actually true.
Which again is funny because as an AI consulting company you should have expertise to know what you want to weite about
But good news, this company also has a free ebook. I am sure it is fantastic.
Big topic early 2020s
reply