Thanks!
Cost is not the bottleneck in this case... The problem is that I am rather ignorant about the options and their benefits/limitations. Moving the images the first time was painfully slow because of those same rate limits, and I expect the next migration to be the same, so I want to make a better choice next time and would rather find a solution with the current provider 😅
Thanks for the heads up. I am still trying to resolve this without a migration... I will try again to get a response from them as they have not replied in a week.
This is the current status:
The instance is serving the images via object storage. Specifically, I am making use of Contabo to save and serve the images.
I now know that the default limits are 250 requests / second and 80 Mbit/s: https://help.contabo.com/en/support/solutions/articles/103000275478-what-limits-are-there-on-object-storage-
It appears to me like when the requests are exceeded, the "Too many requests" error is triggered and it takes a few seconds before the requests are accepted again. This can happen if few users access the front page at once as this will fetch all of the thumbnails and icons on the page.
I have been in touch with Contabo's customer support via e-mail. But they mis-understood my original e-mails and thought I was speaking about increasing the maximum number of images that can be stored (3 million by default). I have clarified that I want to increase the rate limit and have been waiting for their response for a few days now.
The other solution would be to move the images to a different object storage provider. The migration is also limited to the 250 requests/s and 80 Mbit/s, so it will require turning off the images for 4 - 7 days while all the images are moved... Since I am not familiar with the policies of other object storage providers I would also need to do research to avoid falling into the same trap.
So, I am hoping that Contabo's support will get back to me soon and allow me to increase the rate limits, as this would be the most straight forward approach.
And you are doing a great job at that! 😄
Very interesting article, thanks for sharing. I agree that it is a good one to pin!!
This error is a rate limit from the object storage provider. I did not know of this limit when I chose them, and I still have not found a way to change the limit. I will send them an e-mail. If the limit can't be increased, one option is to pick another object storage provider, but the migration takes days.
Check in your settings whether you have disabled the visibility of bot responses. This can happen if bots replied to you and your settings are set to not see them.
I did not know of the term "open washing" before reading this article. Unfortunately it does seem like the pending EU legislation on AI has created a strong incentive for companies to do their best to dilute the term and benefit from the regulations.
There are some paragraphs in the article that illustrate the point nicely:
In 2024, the AI landscape will be shaken up by the EU's AI Act, the world's first comprehensive AI law, with a projected impact on science and society comparable to GDPR. Fostering open source driven innovation is one of the aims of this legislation. This means it will be putting legal weight on the term “open source”, creating only stronger incentives for lobbying operations driven by corporate interests to water down its definition.
[.....] Under the latest version of the Act, providers of AI models “under a free and open licence” are exempted from the requirement to “draw up and keep up-to-date the technical documentation of the model, including its training and testing process and the results of its evaluation, which shall contain, at a minimum, the elements set out in Annex IXa” (Article 52c:1a). Instead, they would face a much vaguer requirement to “draw up and make publicly available a sufficiently detailed summary about the content used for training of the general-purpose AI model according to a template provided by the AI Office” (Article 52c:1d).
If this exemption or one like it stays in place, it will have two important effects: (i) attaining open source status becomes highly attractive to any generative AI provider, as it provides a way to escape some of the most onerous requirements of technical documentation and the attendant scientific and legal scrutiny; (ii) an as-yet unspecified template (and the AI Office managing it) will become the focus of intense lobbying efforts from multiple stakeholders (e.g., [12]). Figuring out what constitutes a “sufficiently detailed summary” will literally become a million dollar question.
Thank you for pointing out Grayjay, I had not heard of it. I will look into it.
Cross-posting to the OpenSource community as I think this topic will also be of interest here.
This is an analysis of how "open" different open source AI systems are. I am also posting the two figures from the paper that summarize this information below.
ABSTRACT
The past year has seen a steep rise in generative AI systems that claim to be open. But how open are they really? The question of what counts as open source in generative AI is poised to take on particular importance in light of the upcoming EU AI Act that regulates open source systems differently, creating an urgent need for practical openness assessment. Here we use an evidence-based framework that distinguishes 14 dimensions of openness, from training datasets to scientific and technical documentation and from licensing to access methods. Surveying over 45 generative AI systems (both text and text-to-image), we find that while the term open source is widely used, many models are ‘open weight’ at best and many providers seek to evade scientific, legal and regulatory scrutiny by withholding information on training and fine-tuning data. We argue that openness in generative AI is necessarily composite (consisting of multiple elements) and gradient (coming in degrees), and point out the risk of relying on single features like access or licensing to declare models open or not. Evidence-based openness assessment can help foster a generative AI landscape in which models can be effectively regulated, model providers can be held accountable, scientists can scrutinise generative AI, and end users can make informed decisions.
Figure 2 (click to enlarge): Openness of 40 text generators described as open, with OpenAI’s ChatGPT (bottom) as closed reference point. Every cell records a three-level openness judgement (✓ open, ∼ partial or ✗ closed). The table is sorted by cumulative openness, where ✓ is 1, ∼ is 0.5 and ✗ is 0 points. RL may refer to RLHF or other forms of fine-tuning aimed at fostering instruction-following behaviour. For the latest updates see: https://opening-up-chatgpt.github.io
Figure 3 (click to enlarge): Overview of 6 text-to-image systems described as open, with OpenAI's DALL-E as a reference point. Every cell records a three-level openness judgement (✓ open, ∼ partial or ✗ closed). The table is sorted by cumulative openness, where ✓ is 1, ∼ is 0.5 and ✗ is 0 points.
There is also a related Nature news article: Not all ‘open source’ AI models are actually open: here’s a ranking
PDF Link: https://dl.acm.org/doi/pdf/10.1145/3630106.3659005
Jajaja, sí, soy Mexicano 😁
🥳 Muchas gracias!
Amazing work! Thanks a lot!! Took me a few days to get to it but I have upgraded now and it looks great 😄
Awesome! The one with the sustained source loop is my favorite:
Also, the one that shoots out flames paints a picture similar to how a synchrotron behaves, shooting out X-rays into the beamlines as the electron bunches move around.
Upon looking into it closer, the synchrotron is a bit of a mixture of those two concepts - the source loop (booster ring) that is fed by the linear accelerator, and then the larger loop (storage room) that feeds X-rays the beamlines. Of course, many details differ, but still it is interesting to notice the similarities !
Good news! Just got a reply from them and they have increased the connection limit. They did not specify what the new number is, but hopefully it is high enough to not be an issue for the foreseeable future.
So, if you do run into other similar reports after this comment I would appreciate it if you tag me again.