At the moment I have been pretty satisfied with the LLMs that Kagi is offering in their standard their for my low requirement queries. Supposedly most of the apis they use are too alternative servers that host instances of the models, some claiming they don’t use your data.
I am not aware to what extend open source is serviced in those models, but I have little interest in Lumo for the time being. Can someone explain to me why open source matters for LLMs?
How is the heat production and ticking noise on the Belkin? I have been pretty content with a qi1 anker magnetic puck charger, but I read the qi2 from them gets pretty hot.
Currently I ran invidious in docker. The solution is … just okay, performance isn’t great. But at least you only stream whatever you pick for watching.
I only follow around 15 channels. For simplicity 10 video are added daily to my subscription feed.
Are there privacy concerns with this solution. This together with ad blocking were my main drivers for invidious.
I have been waiting to get approved for a peertube account on tilvids.com, but after 7 days i am still waiting. Did you manage to get it in dev mode already? I have the video for you still, if you have a suggestion for another platform i can upload it to.
The custom maps community for wahoo is very strong. They update maps, and using OSM data, more frequently then wahoo. You can download maps here and then you push it to your device with an app called Elemantary. Find the read me in that onedrive folder.
Getting the bolt into dev mode such that it can connect via usb is a bit fiddly, you have to push power and up/down buttons at once. But if you are persistent you will get it.
Now having used Ridewithgps for planning I find it superior to Komoot. I find the yearly plan a bit expensive as I don’t use it often, but subbing to the monthly plan on occasion is a fair price imo. The free feature list also pretty rich already. So consider your use cases.
Maybe download something like Organic Maps on iPhone for offline maps as a back up solution.
It does continue your workout when you turn off and on the device, within a certain time window. I thought it was a couple of hours. Just be wary not to stop the activity after pausing.
I do find the bolt loses the cycling category when doing this, it defaults to ‘other’. It is not a mega concern, all metrics are still present.
Imported chinese EVs are still undercutting the European manufacturers, as much they have been lobbying to have increased import fees on the cheaper non EU EVs. The narrative is that they are heavily subsidised to the extend they cannot compete anymore, in so crippling overal in-continent EV development. Resulting in a dependence to the East.
If I am not mistaken such import fee has just passed in Brussel, in order to protect the European market.
In a weird way, humanity has been longing for greener automobile solutions, and now that it’s there, Europe wants to slow it down.
I have a Mac mini m1 and run several containers to manage media, arrs and vpn. I also have a Debian vm for homeassitant. I had used UTM prior but now moved the vm to VMware fusion.
Main issue I run into is not enough ram as the base model only has 8gb. And I do not have redundancy optimised when I reboot. I have it locked with a password so it doesn’t auto run the containers or the vm.
I double the mac for light browsing too. I considered getting a dedicated Linux mini pc for the services next to a desktop Mac. But my usage is too sparse at both ends of the spectrum.
I tried the hass route instead, but can’t get it to work due to dependency issues once I try to install the supervisor package. Even though the aarch64 OS agent seems to install without issue. I’m tired of getting it to work.
I have managed to install HASS on a Mac mini m1 through a Debian vm in UTM, that is serving its introduction purposes right now. Likely I will end up getting a home assistant Green at some point, but I don’t find the price that appealing for what it is. Or I need to shell out even more for a n100 mini pc.
At the moment I have been pretty satisfied with the LLMs that Kagi is offering in their standard their for my low requirement queries. Supposedly most of the apis they use are too alternative servers that host instances of the models, some claiming they don’t use your data. I am not aware to what extend open source is serviced in those models, but I have little interest in Lumo for the time being. Can someone explain to me why open source matters for LLMs?