The striking part is that it's so much higher while we're 7 weeks into the year. The other years include all 52 weeks. Also 40 is close to the maximum for previous years on the chart, not the average, which I'd estimate around 25.
zkfcfbzr
Pretty striking - I'd add a title to the top and the source in the lower right. Would make it much more shareable.
Edit: And a note about 2025 only being up to February 17th. Because the graph may outlive the ~~next Delta flight~~ week.
Same, and that was whether I searched with or without a space between 'libre' and 'office'. In fact all ten results on the first page of results were directly related to the software - either links to the site itself, or reviews of the software.
It also somewhat embarrassingly revealed that the page title for us.libreoffice.org is "Home | Your Site Name - your tagline here"...
I tried out the 8B deepseek and found it pretty underwhelming - the responses were borderline unrelated to the prompts at times. The smallest I had any respectable output with was the 12B model - which I was able to run, at a somewhat usable speed even.
Fair, I didn't realize that. My GPU is a 1060 6 GB so I won't be running any significant LLMs on it. This PC is pretty old at this point.
I have 16 GB of RAM and recently tried running local LLM models. Turns out my RAM is a bigger limiting factor than my GPU.
And, yeah, docker's always taking up 3-4 GB.
Expensive or not, we're well past the point where it's optional. Even if 100% of new carbon emissions stopped today, let alone by 2050, we'd need to continue developing carbon capture technologies to take out what we've already put in the atmosphere. Not every part of the fixing process needs to be profitable.
I know someone who works in UHC's appeals department. They do in fact overturn the majority of denials which are appealed. Might just be selection bias, though, with only those who have the least ambiguous situations bothering to appeal.
Is grandpa's fly open?
Hash tables are often used behind the scenes. dicts and sets in python both utilize hash tables internally, for example.
And then, to perfectly demonstrate your point: 90% of this comments section!
Excellent π