Appreciate the thoughtful response! I’m actually right in the ICP you described — I’ve run my own VMs in the past and recently switched to Cloud Run to simplify ops and take advantage of scale-to-zero. In my case, I was running a few inference jobs and expected a ~$100 bill. But due to the instance-based behavior, it stayed up the whole time, and I ended up with a $1,000 charge for relatively little usage.
I’m fairly experienced with GCP, but even then, the billing model here caught me off guard. When you’re dealing with machines that can run up to $64K/month, small missteps get expensive quickly. Predictability is key, and I’d love to see more safeguards or clearer cost modeling tooling around these types of workloads.
Hey! It's awesome to read other people's solutions to this.
I've been working on solving this for the past 2 years or so and I went through much of the same struggles in the beginning until we came up with a solution which is fairly complex, to get LLM's to output data in a way we can use.
The big problem is that 95% accuracy is not good enough for calendars. People lose confidence after 1 failed attempt. Trying to get LLM's to output JSON can have a 1 in 1000 invalid JSON problem which is unrecoverable. What I wound up doing is training models for the tasks with tremendous amounts of data. I did not use OpenAI's models as they were not right for the job. Would love feedback.
This is not a simple "do this", except to say you need to see a doctor. You are not alone! Many suffer with anxiety for their whole life. What helps is therapy to learn to train your brain that many are anxious too and that you are normal. The ones who don't have chronic anxiety are people you can learn from. When you figure it out you'll have superpowers that most would kill for.
You may choose medication to help you with that initial training, you may have a completely different diagnosis.
No one here should diagnose you, because a doctor wouldn't diagnose over the internet.
A helpful book:
"Brain Lock" by Jeffrey Schwartz. It's about OCD but in my experience all these things are connected.
Hey there! Nice work, looks great!
I've built a few of these and the biggest concern with those motors (I've tried them before) is that they aren't made to handle the torque of even small hills. They'll burn out on mile 25. It's not the wattage that's the problem, they just start to slip and burn out unless you gear them down a lot, but then you lose a lot of their speed. Maybe I've just always bought cheap stuff. Also worried about the tension on the chain, is it adjustable?
I also had trouble with XT90 in the past. I would melt them. Proper soldiering iron fixed that.
Nice build! It's super unique!
I looked up the names for the parts of the fork: It's the part of the fork called the dropout that fails. Here's a picture of a dropout itself: https://i.stack.imgur.com/VZ8bO.jpg, not mine.
When it fails it may look exactly the same (you won't see anything wrong), if you take calipers to it, you'll notice the gap has widened. So measure it now and remeasure every once in a while. If it's changing you've got a problem.
OR you may be able to see it is visibly slightly wider, like more bowl shaped. It doesn't take much for the wheel to come off, it's very tight tolerances. I've never had one snap yet, but I've heard of other's whose dropouts have snapped.
You will notice it is starting to fail when you brake and the wheel feels like it is shifting ever so slightly. It might make a tiny little clink sound. Sometimes it will make the wheel clink around in the dropout when you go over bumps, since the tolerances are SO tight in that area.
It's not just stretching the C towards a straight line that happens, but also twisting at that area or flattening from side to side movement.
Ever since they failed on me a few times I've been riding less because it's a problem I have not solved yet. I reinforced them with 1/4 inch steel. Still it moved independently.
One other solution I've heard about to get bigger tires, they are often called "fat tires". That dampens the stress on the bike a lot. It also makes for a much smoother ride. But I've not done that yet since fat tires need space in the fork. I think a stronger fork and fat tires may totally solve the problem. Fat tires would be nice because it'll turn your bike into a nice smooth ride, and you know how rough the ride can be at that speed for that long.
Hi Jacques! A little late here. It's cool to see someone else spot welding! For the DC area, it's totally faster to commute and I took my daughter everywhere! Now we're in Philly and I use it for long ranges. I've built a bunch of these in various forms, my favorite is a big wheel scooter. It's my main commuter. Look up rivet nuts for securing to the frame! Was a big life saver.
One thing: After years of riding various bikes I've built, please for the love of everything, be careful at high speeds because I've discovered the weakest point of the bike is the C shaped thing at the end of the fork for the wheels. At super high speeds the bumps are rougher on the bike and will eventually pull it apart, even with mid motor power. It may not even come apart all the way, but spread open slightly. Maybe millimeters.
I've tried several bikes, offroad and everything. It won't happen now, but like at odometer 6000 or something. Every single time it fails there. I've had 3 bikes, various scooters, all fail that way. It means your wheel may eventually, suddenly come off at a high speed. I think the answer is a custom fork, but haven't gone there yet.
I'm jakecodes on twitter if you want to trade knowledge.
I always wanted to make an electric skateboard. After a few successful iterations, I decided to make a really fast one that could replace a car. I built my own lion battery out of ~500 18650s. It could go about 30 miles on a charge. It could go about 50mph. It was a very wide body with huge air filled tires. It could pull 10 of my friends up the steepest hill. It was very powerful because I installed an electric motorcycle engine on it. My one extra stupid move was I didn't account for the battery dying at high speeds. Normally I'd stop well before an empty battery but I only had 3 LEDs of primitive battery indication. And that only measured with voltage. Voltage fluctuates wildly in motorized vehicles. It died while only going about 20mph but up a steep hill. The sudden loss of forward momentum flung me off the board into the pavement where I flipped a few times and scrapped across the pavement for a while. Was a painful lesson.
I’m fairly experienced with GCP, but even then, the billing model here caught me off guard. When you’re dealing with machines that can run up to $64K/month, small missteps get expensive quickly. Predictability is key, and I’d love to see more safeguards or clearer cost modeling tooling around these types of workloads.