bearjaws

51 ms·

841.
by bearjaws 3y ago
No mention of how many cybertruck pre-order cancellations in the shareholder PDF... I personally know two people who got tired of waiting and now own Rivians (and then cancelled).
842.
by bearjaws 3y ago
'Member when DevOps was about improving DX? I member. Since every organization lost sight of the fact that DevOps should be serving the Developers as a customer - DevEx was a low priority and DevOps became about infrastructure purism.
843.
by bearjaws 3y ago
I used inc file, it was pricey but it worked really well.
844.
by bearjaws 3y ago
Except you need to build 'boring' 35-50 floor towers that are pre-designed and already have built versions of themselves. That way you can just buy the design and have a few people with prior experience building them to help out.
845.
by bearjaws 3y ago
I have a suspicion that Tesla FSD / Vision is "cooking the books" so to speak. The fact they have an Elon mode, they probably focus on his day to day trips heavily. All the influencers and YouTubers who use it experience mont
846.
by bearjaws 3y ago
I view C Corp the same way as asking the question "how do I make my infrastructure scale to 1 million users (I have 0 currently)". Make a successful LLC or S Corp, then worry about the structure. Unless you plan to give out equity
847.
by bearjaws 3y ago
Just make sure to host it on some other hosting provider. If you're on AWS, use GCP or Azure, etc.. A previous DevOps team I worked with brilliantly centralized everything into K8s, including the status page software. Then pushed a K8s
848.
by bearjaws 3y ago
Would fine tuning be able to fix this kind of problem in a model like LLaVA? I am not sure if it even has the right foundation data set to even understand the myriad of gauges that exist in the world, so I doubt fine tuning will fix anythin
849.
by bearjaws 3y ago
Backlog size is inversely proportional to how strong the product team is in your organization. More "No we aren't doing that idea" vs yes men. If your backlog is massive, people aren't having hard conversations about wha
850.
by bearjaws 3y ago
This is why we need to focus on smaller, more efficient models. A lot of the 13 and 30bn parameter models are actually quite good, maybe not GPT4 but it can run on my M1 MBP without an issue, and a bit faster than GPT4 anyway. I don't
851.
by bearjaws 3y ago
Record time compared to the crypto space, "all we need is infinite data storage with instant recall. Then we can store everything on the chain, and verify its authenticity!"
852.
by bearjaws 3y ago
Yeah, I saw a TikTok of a person installing a electronic kill switch in the car so that the fuel pump wouldn't run. It was a 2020 Lexus SUV, and I was baffled since it was in Toronto. Didn't realize car theft has become that bad i
853.
by bearjaws 3y ago
Not really surprising, even if you sell your car honestly it can end up in the Middle East. > Texas Plumber suing dealership after his car was used by terrorists in Syria, with his companies logo and phone number still on it https:&#x2F
854.
by bearjaws 3y ago
Anything Brother makes. My father has a 10 year old black and white laser printer from them and it hasn't had an issue. I have the HL-L3280CDW and its been solid.
855.
by bearjaws 3y ago
Feels like a fluff piece. 100 microwatts of power at 15x15x15mm means its applications are very limited. Given that its nuclear I assume it will have diminishing power generation over time. A regular RTG like the ones used in voyager produc
856.
by bearjaws 3y ago
Yeah sorry Massachusetts General Hospital, we always used the abbreviation when talking about them.
857.
by bearjaws 3y ago
Seen this several times in my career. It's not all VC hockey stick growth, many companies are just profitable, they have 10-20 employees and the owner lives a great life style. Generally the pay is pretty good and they focus on custome
858.
by bearjaws 3y ago
Theres a field of auto-immune disorders, where patients suffer from their own immune system. I believe they would disagree with you.
859.
by bearjaws 3y ago
I actually worked with MGH on their first proton treatment software for non cyberknife proton treatment. Later scaled it into AWS so their dosimetrists could iterate on treatment plans much faster. The initial treatments were incredibly suc
860.
by bearjaws 3y ago
So in that same 30%, Spotify pays: 1. Engineering staff 2. Hosting costs (probably tens of millions of dollars annually) 3. All other liabilities Apple for that 30% 1. Charges a credit card 2. Pushes app updates 3. "Verifies" app
861.
by bearjaws 3y ago
> 3. Doesn’t Sootify also take a 30% cut to allow artists on their platform? Is Apple paying for their engineering staff, data centers and streaming bandwidth? No. Once someone purchases Spotify, Apple just pushes app updated, a microsco
862.
by bearjaws 3y ago
Those who fail to learn from history are doomed to repeat it. Anti-trust was started for a reason, we know how this ends long term, little competition and increased prices for everyone.
863.
by bearjaws 3y ago
I think were at the point in capitalism (late stage corporatism?) that employees are more or less going to capture 'excess productivity' for themselves by doing chores and running errands. Same goes for the office, they will go ar
864.
by bearjaws 3y ago
Watch out for the memory bandwidth limit on M3, you have to buy the top of the line 16c CPU to get the same memory speed as an M2 Pro.
865.
by bearjaws 3y ago
Shout out to OP for ModelFusion. Probably one of the better ways to use LLMs instead of just integrating the OAI API and reinventing the rest of the wheel. Handles lots of corner cases and lets you swap in other models easily. https:/
866.
by bearjaws 3y ago
That's a lot of words for "we had to do this because of licensing".
867.
by bearjaws 3y ago
For inference and fine tuning yes, not training.
868.
by bearjaws 3y ago
Its far worse than that. Apple clearly abuses its market place "railroads" to control shipments of apps "oil" from its competitors, mainly Kindle and Spotify. Of course while offering the same apps at a much higher margi
869.
by bearjaws 3y ago
I doubt many people are using local setups for serious work. Even fine tuning Mixtral is 4xH100 for 4 days. Which is a ~$200k server currently. To fully train, not just fine tune a small model, say Llama 2 7b you need over 128GiB of vram, s
870.
by bearjaws 3y ago
I think were at the point in capitalism (late stage corporatism?) that employees are more or less going to capture 'excess productivity' for themselves by doing chores and running errands. Same goes for the office, they will go ar

More ›