444 karma · joined August 14, 2019
https://github.com/albumentations-team/AlbumentationsX
Stars help demonstrate real-world adoption when we apply for grants that fund open-source maintenance, testing, documentation, security improvements, and new features.
Based on my BMI, I am overweight. Based on my body fat percentage (measured by DEXA scan), I am athletic.
Conclusions from BMI are not exactly "Garbage in - garbage out," but I suspect that if the body fat percentage were used by scientists, conclusions would be more accurate and insightful.
- 12800+ stars
- 1M downloads last month, 37k per day
- Paper about the project: 1500+ citations
- Used in 18k other repositories and 317 packages.
=> People use it.
But!
- 365 open issues
- 25 pull request
that hang for years
Only one sponsor.
And this is fine. People use the result of our work, but we do not feel that we are entitled / deserve / [some other vomit words] of more support.
In the beginning, we decided that we would do it:
- only for fan
- when and how we want it
- if some user is unhappy with our commitment or decisions
=> feel free to fork.
But!
We do enjoy when people thank us, create pull requests (we do review and merge them, although it could take time), or create feature requests or bug reports.
I can see open source as a great pet project that you do for fun and to improve your skills, but unless it is an OpenCore business or another setup where maintainers are financially compensated, all whiners and complainers can go and fuck themselves.
I would not recommend maintainers of the open source software even notice them.
If working on OSS is not fun - do not do it. Life is too short for unnecessary stress.
Waiting for someone will match GPT-4 and especially GPT4-V quality and latency.
And this would be the start of the battle, which could be bad for combatants but great for the rest of us.
Fingers crossed, it will happen soon. Not sure that early 2024 ill be that day, but let's hope.
There are many knobs one can tweak to increase the quality and abstinence of proper length (not too small and not too long is the most impactful and most straightforward)
One day, and for most people, two days, is not enough to pass the threshold to consider donation successful.
And for donation, it is actually not one threshold, but many. As you may see in the image with compensation per visit, there is a thing called vile.
Ideally, one vial is one pregnancy. In practice, I was told that due to possible complications, it requires 9-12 vials to get a woman pregnant.
=> They do not allow donations more often than two days between visits, and if your physical and mental health is not ideal (level of stress affects the number of vials per visit as well), it may take 3-4 days.
In SF Bay Area - sperm donation.
I do not personally know any SWEs who do this, but a couple of my friends (and me) who work on their startup ideas do.
Link to an example of how much you will get paid.
Here is the math, assuming you are in great health:
one 15 minute visit => 6+ vial Yield => $175
You cannot do it every day. There should be a 2-5 days abstinence period.
=> best case scenario => 3 times a week.
=> 4 weeks * 3 times per week * $175 per donation = $2100
As you can see in that link, there is a $1000 bonus if you make 12 monthly visits.
=> $3300 per month.
This is good for ethical reasons as well -> you help women with an ultra-important personal problem - how to become a mother.
In a time when the population is aging, you help to mitigate this problem with your own hands.
If everything is so good, what is the catch?
Such programs are picky about who they accept, claiming the acceptance rate is 1%.
I do not know what is the quality of the initial pool of candidates, so hard to tell what percent would be for the pool of SWEs that need side income.
P.S. I am quite serious about all of it, as I am part of this program as well, and the link with how much you get paid is for November in the company that I am donating to.
P.P.S. I wish I had known about such an option when I was a graduate student making $1600 / month for teaching. An extra $3k would make a huge difference in the quality of living and how much I could send home.
It also shows that GPT-4V created a new angle in web scraping.
I guess, this or similar code would be leveraged in many projects like:
1. Scrape XXX websites, say LinkedIn or Twitter use all types of methods in the DOM to prevent it, but fighting working well GPT-4V + OCR would be ultra hard.
2. Give me an analysis of what these XXX companies are doing. And this could be done for competitors, to understand the landscape of some industry, or even plainly to get news.
Large-scale scrapping, not depending on the source code of the pages is a powerful infrastructural change.
Flying on a plane is faster, but if we cannot have comfortable trains in the US, let's have quiet, comfortable airships.
As GPT is used in all text-related activities to give feedback to the writer, GPT-4V in the future will be applied to all websites, posters, book covers, graphs, plots, design of hardware devices, movies, computer games, etc.
I suspect that teams at Figma, Canva, Adobe, and others are already working on it.
It looks like GPT-4V is not just another model but a change in infrastructure, unlocking new possibilities and new "blue oceans." as any good new multi-modality.
It feels so great to live in the future.
As other people mentioned in the thread I did not really need them while in University, but after graduation, being in SF Bay Area in tech, meeting women became challenging.
In this geography (SF Bay Area) and subculture (tech) the number of women, especially good looking, who are single is objectively small. You rarely meet them in real life.
Dating apps solve this issue - open an app, match, chat, meet.
But in other geographies and subcultures, meeting in person would be more efficient and enjoyable.
But even if students stay away form it, dating apps industry waits for disruption. And I hope, this time executive teams in Tinder, Bumble, etc will not ignore this signal and invest into creating a better experience for all sides.
At least my trips to mountains would be more safe as I will be always reach for help when needed.
Being able to call and chat with your friends and family, while you are climbing big wall and sleep on the portable ledges would be really cool as well.
Example:
> The U.S. government has requested that Max Polyakov, a wealthy Ukrainian tech entrepreneur, sell his stake in the rocket company Firefly Aerospace Inc., citing national security concerns.
https://www.bloomberg.com/news/articles/2021-12-29/firefly-o...
Book: "How Heavens Went On Sale" by Ashlee Vans talks in details about the raise of commercial space in recent years and ways government deals with the corresponding issues.
Unlike many other health related problems this could be easily addressed within one trip to the computer store.
I do not see how it can negatively affect me as data is anonymized.
At the same time if it will help to develop better drugs -> this is good thing for the humanity.
If someone will make money in the process - 23andMe and drugmakers - this is great as it will incentivise other companies invest to similar research projects which, again, looks as a good outcome for the humanity.
If information about my DNA will make a drug being more beneficial to people who share DNA with me - my relatives and I will benefit as well.
win-win everywhere
You are exposed to such a great and infinite pool of potential partners that you would not be able to get in real life, especially in SF Bay Area. Which changes the problem from - how to meet someone, to a better problem - how to filter out those that you definitely do not want to meet.
Even more, it is not replacing your real life meetups but compliments them.
My issue with dating apps is that the number of matches is large and all of them are very similar, hence the challenge is how to filter properly.
As many people here, I love 10 year old version of OKCupid where you needed to fill a lot of information, answer questions and invest in the profile in general.
Even if we forget about this extra information, fact that person invested in filling the profile told me that she was serious about meeting someone which is a strong signal in itself.
I believe that online dating industry is a great achievement of the humanity as it drastically expanded our choice of partners, but existing implementation is waiting for the disruption.
Personally, I would love if dating apps added:
1. Tags for conversations, similar to labels in gmail.
2. Ability to search - when you have hundreds of conversations, it is hard to find one that you are interested in.
3. Option to add more information to profile, say I loved questions at OKCupid.
4. Id and age verification. Feels like every second lady lies about her age. (One my female friend who is 49, uses 36 in online dating. I do not think that such a difference is that common, but still)
5. Somehow force people to use photos that are less than a year old. Very unpleasant come to a date and meet a girl that is not even close to be as fit as she is in her pictures. They are also very surprised when you do not go on a second date and, I suspect, complain to their girlfriends that "online dating is not working for them". I do not blame them - misleading with photos is so common -> she needs to mislead to be in that competitive market (I was told that men lie about their height as well). I believe this could be addressed or at least mitigated on the platform level. (Being fair - having video chat before the date helps to address this issue to some extent)
6. Showing stats. Communication, flirting, building comfort in online dating is a special skill, quite different from doing the same in real life:
[6a] It is happening in asynchronous way
[6b] Information channels are limited - you do not have voice, intonations, appearance, body language, face expressions, in the beginning it is just text
[6c] As conversation may be stalled or end at any moment both sides increase odds by having many conversations happening at the same time.
these limitations change everything and force to learn new communication skills from scratch.
If we were shown metrics like: yours and average in your geography and age range stats for open rate, reply rate, length of the conversation, number of matches, number and percent of conversations that lead to a meeting (Hinge these days asks if you met in person.), etc it would help people see what their limitations are and focus on addressing them - same as in sales as sales and online dating are similar in many ways.
But, I would say that the main issue with CLIP is not performance, but that textual input is limited to 77 characters.
This is a severe limitations, if Meta or other company collected the dataset that allowed model with 1024 characters instead it would enrich the word of open source models much more than +2% accuracy.
My hope is that next person or company who works on that will invest into longer context size for text input :fingers_crossed:
We, I mean humanity, need an example that L4 self driving is possible.
When SpaceX showed that private company can access space and build a product in the area world changed.
Startup founders, investors, government changed their opinion and many space companies were created and trying to do things that they would not even consider before SpaceX: Astra, Rocket Labs, Planet Labs, Firefly, etc
(Great book on the topic "When Heavens went on sale": https://amzn.to/46NRx9l)
If we get one example of self driving as a product on the roads, it may have the same effect. We are going toward self driving winter and unless Cruise, Waymo or similar companies will do the impossible new iteration will happen in 10+ years.
It is mentioned in the text that in Q3 GM spent 700M on Cruise.
Not that many companies in the world: 1. have similar level of talent as engineers at Cruise 2. can spent similar amount of money 3. are willing to spend them
And we need all three.
There are so many corner cases that without large scale as close to the real life deployment as possible...
I feel very sad.
P.S. Worked for 3 years in a self driving company, till our division was sold :(
Rays of support to all immigrants who came to SF Bay Area and are trying to make it a new home.
I like this part in the text.
I lift weights regularly for many years, can pass requirements for 2-3rd class sportsman in powerlifting, weighted pull ups, weighted dips.
( requirements for different sexes and weight categories: https://www.sportscategory.info/powerlifting )
I do not use steroids, but plenty of my friends do.
Some take them to perform well in competitions. Unlike athletes mentioned in the article they work out more than average weight lifter.
Steroids do not give muscles for free, they just allow faster progress, and faster recovery. Natural athlete will not be able to handle training plan that person on steroids can. For example, Arnold Schwarzenegger talks in his book "Total Recall" on how he trained twice a day, every day with heavy weights.
Other people that I talked to were not satisfied with their appearance, I would say they were insecure, lifting weights did not give much results and they use steroids to address the issue. Many of them are ectomorphs. When I ask them - do you want to stop and become natural athlete, they say no.
They also mention is that high testosterone gives them more confidence, less stress, high sex drive. They are not ready to give it away and get back to the insecure, stressful life, much smaller attention from women even if steroids have negative side effects.
I guess they have some mental deviations (I wanted to say issues, but sounds too strong) and insecurities. How they are perceived by others is a big chunk of their identity. Would not be surprised if they were prone to some psychological problems without steroid use.
1) > The study encompassed 118 previous and current male steroid users and 97 weightlifters who confirmed they had never used steroids.
It is a very small number
2) Even if results of the study are statistically significant mental issues and steroid use are only correlation and assuming that steroids lead to problems with your head is an overkill, which authors of the paper clearly state, but readers may not see.
P.S. Steroids have so many negative side effects that it is a mystery for me, why one decides to use them. Although I do not feel that have problems with my body and appearance overall.
Investment department and cloud compute departments are very different entities and if the investment to Antropic was not a part of the Google cloud expansion strategy moving to a new cloud is such a pain, that noone will do it.
Opposite example is Kaggle. Google bought it as a marketing tool for Google Cloud => all the years since acquisition Kaggle team worked on adding differnt Google Cloud features to the platform
The game is unplayable, same issue as for folks in that thread:
https://www.reddit.com/r/Starfield/comments/16bm1f7/audio_cu...
GPU: GTX 2080 Ti
CPU: threadripper 3950x
RAM: 64Gb
Game installed on SSD
Which I use before uploading photos to pubic services.
Exif has surprising amount of infomration
But, it is the best that we have and on average it works.
Rephrasing Churchill:
"No one pretends that career ladder is perfect or all-wise. Indeed it has been said that career ladder is the worst form of building hierarchy in the company except for all those other forms that have been tried from time to time.…"
What are other, presumably better, ways to build a hierarchy in the company that has 1000+ people?
People whose value for the company is not for the next level get promoted due to luck or their ability to hack the system. Those who perform for the next level are not promoted.
The system feels broken. But! It is the best working process in big tech. And, on average, this bureaucracy helps to limit politics.
Basically perf review process feels horrible, till you work in a company that does not have it.
Limitations discussed in the text like manager could not explain why XXX exceeds on her role feels strange.
In every big company there is a career path document that shows what types of tasks person is expected to do at each level. You perform well on your level - (ME) meets expectations, you perform on the higher level - (EE) exceeds or strongly exceeds (SE).
You perform for 1 year+ on the higher level (typically EE + SE) + apply for promo => you get promo.
It is more or less standard across big tech.
Again, the system could be hacked, but on average it works.
P.S. Part about "you deserve" in the title and across the text talks about unhealthy entitlement. I would imagine that if engineer come to me and instead of showing evidence of how they consistently perform for the next level will talk about what they deserve it would not end with a promotion.
P.P.S. Read the text only till "This post is for paying subscribers only"
[1] The compilers don’t produce great instructions;
[2] The drivers crash frequently: ML workloads feel experimental;
[3] Software adoption is getting there, but kernels are less optimized within frameworks, in particular because of the fracture between ROCm and CUDA. When you are a developer and you need to write code twice, one version won’t be as good, and it is the one with less adoption;
[4] StackOverflow mindshare is lesser. Debugging problems is thus harder, as fewer people have encountered them.
---
were crucial when we had enough supply of NVidia GPUs, but if demand described in https://gpus.llm-utils.org/nvidia-h100-gpus-supply-and-deman...
is real (450,000+ H100)
Software bottlenecks most likely will be addressed sometime soon
Slowly but consistently, thanks to careful testing, there is some flow of "plant medicine" application from "traditional medicine" to medicine.
Not sure that homeopathy, acupuncture, etc will be able to survive careful scientific verification to become part of the modern medicine.