MasterScrat

11 ms·

91.
by MasterScrat 3y ago
And also they pay for your health insurance, which in Switzerland is far from negligible
92.
by MasterScrat 3y ago
> They know that money is coming right back to them. Oh I assumed they were directly delivering them truckful of A100s? do they actually still bother with cash? ;-)
93.
by MasterScrat 3y ago
I’ll take this opportunity to mention our research scaling StyleGAN 2 to larger datasets (using LAION) on food images, leveraging free TPU compute through the TRC program. We trained for 36 days on a v4-8 on 558k images. https://
94.
Prototyping XGen-Image-1 – 860M text-to-image diffusion model from Salesforce(blog.salesforceairesearch.com)
2 points by MasterScrat 3y ago|0 comments
95.
Ask HN: Who Died?
1 points by MasterScrat 3y ago|3 comments
96.
by MasterScrat 3y ago
We've added Paypal this way on our service (dreamlook.ai, "Stable Diffusion finetuning as a Service") and it was incredibly smooth. The setup took maybe 15min, and an hour later we had already received our first PayPal paymen
97.
by MasterScrat 3y ago
Yes, it's now been released
98.
SD-XL 1.0 Model Card(huggingface.co)
5 points by MasterScrat 3y ago|1 comments
99.
by MasterScrat 3y ago
For 1.0? Where do you apply? Or are you talking about 0.9?
100.
by MasterScrat 3y ago
It’ll be "released" once the model weights show up on the repo or in HuggingFace… for now it’s "announced" It should appear here at some point, currently only the VAE was added: https://huggingface.co/sta
101.
GPT-4: 8 x 220B experts trained with different data/task distributions(twitter.com)
47 points by MasterScrat 3y ago|5 comments
102.
by MasterScrat 3y ago
We trained a large language model and had it deployed for multiple month last year (Cedille.ai, largest open source LLM for French at the time). We had multiple companies experimenting with it to use it in production. There were multiple oc
103.
by MasterScrat 3y ago
I integrated it yesterday (dreamlook.ai, Stable Diffusion finetuning-as-a-service) and it was impressively smooth. The setup took maybe 15min, and an hour later we had already received our first PayPal payments.
104.
by MasterScrat 3y ago
I've used Wordpress a long time ago and hated it. I know multiple people who work with WordPress professionally and hate it. Has there been recent improvements? I remember a terrible mess of tangled PHP that would easily break when per
105.
Ask HN: Simple, low-maintenance startup blog platform?
3 points by MasterScrat 3y ago|5 comments
106.
by MasterScrat 3y ago
> Your comment essentially says - this is not a high quality submission because readers might not actually read it I'd argue that on average, most readers won't have a good enough understanding, or read the paper far enough, to
107.
by MasterScrat 3y ago
FYI we're building a service to make this process even simpler and faster: dreamlook.ai Upload your pictures, we train the model in a few minutes, then you can download your trained checkpoint. $1/model, first one for free. For ap
108.
by MasterScrat 3y ago
I've been using GPT4 very intensively for the past week. I've mostly been writing frontend code and writing Python CLI utilities. - First, there's a MAJOR difference between GPT-3.5 and GPT-4. From a branding point of view it
109.
by MasterScrat 3y ago
No, you can see the time breakdown in the table under the "coffee break" quote: it is the time for the 3-step RLHF process only. Training a 1.3B parameter model from scratch is still a very large undertaking.
110.
by MasterScrat 3y ago
Note that by "train" here they mean "finetune that network a bit", not train from scratch.
111.
by MasterScrat 3y ago
I'm looking for a new platform to write technical blog posts, is Substack the right place? I'm basically looking for the "Medium from when Medium was good"
112.
by MasterScrat 4y ago
Getting LoRa/Dreambooth to work reliably without human intervention is HARD, and Dreambooth still works better than LoRa in most situations. Sure it’s "easy" to do it by hand, on a few persons, tweaking the parameters. Doing
113.
by MasterScrat 4y ago
One of the core component of such products is Dreambooth, which allows you to finetune text to image models (typically Stable Diffusion) to be able to create images of a specific person, object or style. We are building a platform to speed
114.
by MasterScrat 4y ago
How reliable is this publication?
115.
by MasterScrat 4y ago
If you want to learn about which GPUs make sense to buy or rent for various ML workload, i highly recommend this Tim Dettmers article: https://timdettmers.com/2023/01/30/which-gpu-for-deep-learni...
116.
by MasterScrat 4y ago
What's a CA?
117.
by MasterScrat 4y ago
Now that we have some distance from the whole forced-at-home-during-pandemic episode, what are people's opinions on the topic? I do feel seeing people face to face a couple of times per week does help teams function better. Random wate
118.
by MasterScrat 4y ago
yeah - would be good to update the title given scale of recent layoffs?
119.
by MasterScrat 4y ago
I am getting the message "You are over the daily limit for sending Tweets". I have not tweeted anything in the last 24h.
120.
by MasterScrat 4y ago
That and the fact they provide "zombie mode" as a feature

More ›