Update: Vercel appears to be aware of the situation and have updated the Status Page.
333 karma · joined May 12, 2022
NYC-based entrepreneur. Engineer. TCK.
- building: [mage.space](https://mage.space) - twitter: [@greghunkins](https://x.com/greghunkins)
Appreciate the insight!
Post: https://news.ycombinator.com/item?id=37990039
The comment is not visible unless I am logged in. Other accounts or anonymous viewership show the post at 0 comments.
In general, these insane speed boosts come at the cost of bleeding edge features.
[1] https://github.com/huggingface/diffusers/blob/28e8d1f6ec82a6...
Hardware in general requires a rigor and knowledge of math and science that just isn't as present or necessary in computer science. However, I would argue that computer science requires more abstract and creative thinking.
To each their own. I moved to Computer Science and A.I. specifically because my brain enjoys abstract logic puzzles more than the required precision of a physical system.
It plugs into the rest of our ecosystem, so you can use the rest of the tools: image-to-image, inpainting, textual inversions, and more. Additionally, as with all else on Mage, it is unlimited (no tokens!).
We're still working on LoRA support, but it will come soon. Additionally, some older models seem to have issues with generating coherent animations. This is brand-new tech, so any and all feedback is very much appreciated.
See the relevant Github issue here: https://github.com/facebookresearch/xformers/issues/795
There is an infuriatingly high amount of obviously fraudulent charges though with a very low Stripe risk score that only custom rules catch.
Thank you to the Stripe team for finally taking this seriously and stepping in to resolve this. My guess is that prior to today, this was never elevated to the proper people for real action.
Until then, if your name is Jake Minas and you're trying to get a subscription: I am sorry.
Congrats on making it to a final round, that's a huge accomplishment. I've been rejected from many companies and interviews, and (spoiler) so has everyone in the industry - the only failure is giving up! You've got this.
Some motivation: https://www.youtube.com/watch?v=u_ktRTWMX3M
My site (mage.space) has hundreds of potential customers waiting for this. It was our most asked for feature after releasing our premium features this January.
Look at funding announcements for early-stage start-ups (Seed, Series A) in your city of choice and send them an internship request.
They're likely already in hiring mode, and the bar to entry is lower at a start-up, especially if you're a hard worker. Afterwards, there's a good chance they will extend a full-time offer.
The safe route is to invest based on brand name (aka previous success): Bay Area, ex-FAANG, Stanford education, etc. But by definition, you are trying to bet on current and future trends using previous data.
Data like this makes me suspicious of the increasingly loud threads about "10 reasons why if you're not building in SF, you're not building!"
Capital invested still wildly favors the Bay Area, but that could be a lagging indicator.
"We present Composable Diffusion (CoDi), a novel generative model capable of generating any combination of output modalities, such as language, image, video, or audio, from any combination of input modalities. Unlike existing generative AI systems, CoDi can generate multiple modalities in parallel and its input is not limited to a subset of modalities like text or image. Despite the absence of training datasets for many combinations of modalities, we propose to align modalities in both the input and output space. This allows CoDi to freely condition on any input combination and generate any group of modalities, even if they are not present in the training data. CoDi employs a novel composable generation strategy which involves building a shared multimodal space by bridging alignment in the diffusion process, enabling the synchronized generation of intertwined modalities, such as temporally aligned video and audio. Highly customizable and flexible, CoDi achieves strong joint-modality generation quality, and outperforms or is on par with the unimodal state-of-the-art for single-modality synthesis."
"2. All persons obtaining a copy or substantial portion of the Software, a modified version of the Software (or substantial portion thereof), or a derivative work based upon this Software (or substantial portion thereof) must not delete, remove, disable, diminish, or circumvent any inference filters or inference filter mechanisms in the Software, or any portion of the Software that implements any such filters or filter mechanisms."