Startup entrepreneurs assume what risks when pitching?

Last week I signed four NDAs.

Four!

NDA stands for non-disclosure agreement. A legal document given to someone receiving confidential information. By signing, the person pledges not to share the information for a certain time. If they do, they may be in breach of contract and face legal action.

Companies use NDAs to protect trade secrets and confidential internal information from employees and contractors. Appropriate. If you manage a huge, successful firm, you don't want your employees selling their information to your competitors. To be true, business NDAs don't always prevent corporate espionage, but they usually make employees and contractors think twice before sharing.

I understand employee and contractor NDAs, but I wasn't asked to sign one. I counsel entrepreneurs, thus the NDAs I signed last week were from startups that wanted my feedback on their concepts.

I’m not a startup investor. I give startup guidance online. Despite that, four entrepreneurs thought their company ideas were so important they wanted me to sign a generically written legal form they probably acquired from a shady, spam-filled legal templates website before we could chat.

False. One company tried to get me to sign their NDA a few days after our conversation. I gently rejected, but their tenacity encouraged me. I considered sending retroactive NDAs to everyone I've ever talked to about one of my startups in case they establish a successful company based on something I said.

Two of the other three NDAs were from nearly identical companies. Good thing I didn't sign an NDA for the first one, else they may have sued me for talking to the second one as though I control the firms people pitch me.

I wasn't talking to the fourth NDA company. Instead, I received an unsolicited email from someone who wanted comments on their fundraising pitch deck but required me to sign an NDA before sending it.

That's right, before I could read a random Internet stranger's unsolicited pitch deck, I had to sign his NDA, potentially limiting my ability to discuss what was in it.

You should understand. Advisors, mentors, investors, etc. talk to hundreds of businesses each year. They cannot manage all the companies they deal with, thus they cannot risk legal trouble by talking to someone. Well, if I signed NDAs for all the startups I spoke with, half of the 300+ articles I've written on Medium over the past several years could get me sued into the next century because I've undoubtedly addressed topics in my articles that I discussed with them.

The four NDAs I received last week are part of a recent trend of entrepreneurs sending out NDAs before meetings, despite the practical and legal issues. They act like asking someone to sign away their right to talk about all they see and hear in a day is as straightforward as asking for a glass of water.

Given this inflow of NDAs, I wanted to briefly remind entrepreneurs reading this blog about the merits and cons of requesting investors (or others in the startup ecosystem) to sign your NDA.

Benefits of having investors sign your NDA include:

None. Zero. Nothing.

Disadvantages of requesting investor NDAs:

• You'll come off as an amateur who has no idea what it takes to launch a successful firm.

• Investors won't trust you with their money since you appear to be a complete amateur.

• Printing NDAs will be a waste of paper because no genuine entrepreneur will ever sign one.

I apologize for missing any cons. Please leave your remarks.

And tips for anyone trying.

We see the same story every few years.

BREAKING NEWS: [Platform X] launched a social network. With Facebook's reputation down, the new startup bets millions will switch.

Despite the excitement surrounding each new platform (Diaspora, Ello, Path, MeWe, Minds, Vero, etc.), no major exodus occurred.

Snapchat and TikTok attracted teens with fresh experiences (ephemeral messaging and rapid-fire videos). These features aren't Facebook, even if Facebook replicated them.

Facebook's core is simple: you publish items (typically text/images) and your friends (generally people you know IRL) can discuss them.

It's cool. Sometimes I don't want to, but sh*t. I like it.

Because, well, I like many folks I've met. I enjoy keeping in touch with them and their banter.

I dislike Facebook's corporation. I've been cautiously optimistic whenever a Facebook-killer surfaced.

None succeeded.

Why? Two causes, I think:

People couldn't switch quickly enough, which is reason #1

Your buddies make a social network social.

Facebook started in self-contained communities (college campuses) then grew outward. But a new platform can't.

If we're expected to leave Facebook, we want to know that most of our friends will too.

Most Facebook-killers had bottlenecks. You have to waitlist or jump through hoops (e.g. setting up a server).

Same outcome. Upload. Chirp.

After a week or two of silence, individuals returned to Facebook.

Reason #2: The fundamental experience was different.

Even when many of our friends joined in the first few weeks, it wasn't the same.

There were missing features or a different UX.

Want to reply with a meme? No photos in comments yet. (Trying!)

Want to tag a friend? Nope, sorry. 2019!

Want your friends to see your post? You must post to all your friends' servers. Good luck!

It's difficult to introduce a platform with 100% of the same features as one that's been there for 20 years, yet customers want a core experience.

If you can't, they'll depart.

The causes that led to the causes

Having worked on software teams for 14+ years, I'm not surprised by these challenges. They are a natural development of a few tech sector meta-problems:

Lean startup methodology

Silicon Valley worships lean startup. It's a way of developing software that involves testing a stripped-down version with a limited number of people before selecting what to build.

Billion people use Facebook's functions. They aren't tested. It must work right away*

*This may seem weird to software people, but it's how non-software works! You can't sell a car without wheels.

2. Creativity

Startup entrepreneurs build new things, not copies. I understand. Reinventing the wheel is boring.

We know what works. Different experiences raise adoption friction. Once millions have transferred, more features (and a friendlier UX) can be implemented.

3. Cost scaling

True. Building a product that can sustain hundreds of millions of users in weeks is expensive and complex.

Your lifeboats must have the same capacity as the ship you're evacuating. It's required.

4. Pure ideologies

People who work on Facebook-alternatives are (understandably) critical of Facebook.

They build an open-source, fully-distributed, data-portable, interface-customizable, offline-capable, censorship-proof platform.

Prioritizing these aims can prevent replicating the straightforward experience users expect. Github, not Facebook, is for techies only.

What about the business plan, though?

Facebook-killer attempts have followed three models.

1. Utilize VC funding to increase your user base, then monetize them later. (If you do this, you won't kill Facebook; instead, Facebook will become you.)

2. Users must pay to utilize it. (This causes a huge bottleneck and slows the required quick expansion, preventing it from seeming like a true social network.)

3. Make it a volunteer-run, open-source endeavor that is free. (This typically denotes that something is cumbersome, difficult to operate, and is only for techies.)

Wikipedia is a fourth way.

Wikipedia is one of the most popular websites and a charity. No ads. Donations support them.

A Facebook-killer managed by a good team may gather millions (from affluent contributors and the crowd) for their initial phase of development. Then it might sustain on regular donations, ethical transactions (e.g. fees on commerce, business sites, etc.), and government grants/subsidies (since it would essentially be a public utility).

When you're not aiming to make investors rich, it's remarkable how little money you need.

If you want to build a Facebook competitor, follow these tips:

1. Drop the lean startup philosophy. Wait until you have a finished product before launching. Build it, thoroughly test it for bugs, and then release it.

2. Delay innovating. Wait till millions of people have switched before introducing your great new features. Make it nearly identical for now.

3. Spend money climbing. Make sure that guests can arrive as soon as they are invited. Never keep them waiting. Make things easy for them.

4. Make it accessible to all. Even if doing so renders it less philosophically pure, it shouldn't require technical expertise to utilize.

5. Constitute a nonprofit. Additionally, develop community ownership structures. Profit maximization is not the only strategy for preserving valued assets.

Last thoughts

Nobody has killed Facebook, but Facebook is killing itself.

The startup is burying the newsfeed to become a TikTok clone. Meta itself seems to be ditching the platform for the metaverse.

I wish I was happy, but I'm not. I miss (understandably) removed friends' postings and remarks. It could be a ghost town in a few years. My dance moves aren't TikTok-worthy.

Who will lead? It's time to develop a social network for the people.

Greetings if you're working on it. I'm not a company founder, but I like to help hard-working folks.

Discover How We Used Facebook Ads to Grow a New Mobile App from $0 to $15K MRR in Just 6 Months and Our Strategy to Hit $100K a Month.

Our client introduced a mobile app for Poshmark resellers in December and wanted as many to experience it and subscribe to the monthly plan.

An Error We Committed

We initiated a Facebook ad campaign with a "awareness" goal, not "installs." This sent them to a landing page that linked to the iPhone App Store and Android Play Store. Smart, right?

We got some installs, but we couldn't tell how many came from the ad versus organic/other channels because the objective we chose only reported landing page clicks, not app installs.

We didn't know which interest groups/audiences had the best cost per install (CPI) to optimize and scale our budget.

After spending $700 without adequate data (installs and trials report), we stopped the campaign and worked with our client's app developer to set up app events tracking.

This allowed us to create an installs campaign and track installs, trials, and purchases (in some cases).

Finding a Successful Audience

Once we knew what ad sets brought in what installs at what cost, we began optimizing and testing other interest groups and audiences, growing the profitable low CPI ones and eliminating the high CPI ones.

We did all our audience testing using an ABO campaign (Ad Set Budget Optimization), spending $10 to $30 on each ad set for three days and optimizing afterward. All ad sets under $30 were moved to a CBO campaign (Campaign Budget Optimization).

We let Facebook's AI decide how much to spend on each ad set, usually the one most likely to convert at the lowest cost.

If the CBO campaign maintains a nice CPI, we keep increasing the budget by $50 every few days or duplicating it sometimes in order to double the budget. This is how we've scaled to $400/day profitably.

Finding Successful Creatives

Per campaign, we tested 2-6 images/videos. Same ad copy and CTA. There was no clear winner because some images did better with some interest groups.

The image above with mail packages, for example, got us a cheap CPI of $9.71 from our Goodwill Stores interest group but, a high $48 CPI from our lookalike audience. Once we had statistically significant data, we turned off the high-cost ad.

New marketers who are just discovering A/B testing may assume it's black and white — winner and loser. However, Facebook ads' machine learning and reporting has gotten so sophisticated that it's hard to call a creative a flat-out loser, but rather a 'bad fit' for some audiences, and perfect for others.

You can see how each creative performs across age groups and optimize.

How Many Installs Did It Take Us to Earn $15K Per Month?

Six months after paying $25K, we got 1,940 app installs, 681 free trials, and 522 $30 monthly subscriptions. 522 * $30 gives us $15,660 in monthly recurring revenue (MRR).

Next, what? $100K per month

The conversation above is with the app's owner. We got on a 30-minute call where I shared how I plan to get the app to be making $100K a month like I’ve done for other businesses.

Reverse Engineering $100K

Formula:

For $100K/month, we need 3,334 people to pay $30/month. 522 people pay that. We need 2,812 more paid users.

522 paid users from 1,940 installs is a 27% conversion rate. To hit $100K/month, we need 10,415 more installs. Assuming...

With a $400 daily ad spend, we average 40 installs per day. This means that if everything stays the same, it would take us 260 days (around 9 months) to get to $100K a month (MRR).

Conclusion

You must market your goods to reach your income objective (without waiting forever). Paid ads is the way to go if you hate knocking on doors or irritating friends and family (who aren’t scalable anyways).

You must also test and optimize different angles, audiences, interest groups, and creatives.

Trigger warning — Andrew Tate is running a genius marketing campaign

Andrew Tate is a 2022 internet celebrity.

Kickboxing world champion became rich playboy with controversial views on gender roles.

Andrew's get-rich-quick scheme isn't new. His social media popularity is impressive.

He’s currently running one of the most genius marketing campaigns in history.

He pulls society's pendulum away from diversity and inclusion and toward diversion and exclusion. He's unstoppable.

Here’s everything you need to know about Andrew Tate. And how he’s playing chess while the world plays checkers.

Cobra Tate is the name he goes by.

American-born, English-raised entrepreneur Andrew Tate lives in Romania.

Romania? Says Andrew,

“I prefer a country in which corruption is available to everyone.”

Andrew was a professional kickboxer with the ring moniker Cobra before starting Hustlers University.

Before that, he liked chess and worshipped his father.

Emory Andrew Tate III is named after his grandmaster chess player father.

Emory was the first black-American chess champion. He was military, martial arts-trained, and multilingual. A superhuman.

He lived in his car to make ends meet.

Andrew and Tristan relocated to England with their mother when their parents split.

It was there that Andrew began his climb toward becoming one of the internet’s greatest villains.

Andrew fell in love with kickboxing.

Andrew spent his 20s as a professional kickboxer and reality TV star, featuring on Big Brother UK and The Ultimate Traveller.

These 3 incidents, along with a chip on his shoulder, foreshadowed Andrews' social media breakthrough.

• Chess

• Combat sports

• Reality television

A dangerous trio.

Andrew started making money online after quitting kickboxing in 2017 due to an eye issue.

Andrew didn't suddenly become popular.

Andrew's web work started going viral in 2022.

Due to his contentious views on patriarchy and gender norms, he's labeled the King of Toxic Masculinity. His most contentious views (trigger warning):

• “Women are intrinsically lazy.”

• “Female promiscuity is disgusting.”

• “Women shouldn’t drive cars or fly planes.”

• “A lot of the world’s problems would be solved if women had their body count tattooed on their foreheads.”

Andrew's two main beliefs are:

1. “These are my personal opinions based on my experiences.”

2. “I believe men are better at some things and women are better at some things. We are not equal.”

Andrew intentionally offends.

Andrew's thoughts began circulating online in 2022.

In July 2022, he was one of the most Googled humans, surpassing:

• Joe Biden

• Donald Trump

• Kim Kardashian

Andrews' rise is a mystery since no one can censure or suppress him. This is largely because Andrew nor his team post his clips.

But more on that later.

Andrew's path to wealth.

Andrew Tate is a self-made millionaire. His morality is uncertain.

Andrew and Tristan needed money soon after retiring from kickboxing.

“I owed some money to some dangerous people. I had $70K and needed $100K to stay alive.”

Andrews lost $20K on roulette at a local casino.

Andrew had one week to make $50,000, so he started planning. Andrew locked himself in a chamber like Thomas Edison to solve an energy dilemma.

He listed his assets.

• Physical strength (but couldn’t fight)

• a BMW (worth around $20K)

• Intelligence (but no outlet)

A lightbulb.

He had an epiphany after viewing a webcam ad. He sought aid from women, ironically. His 5 international girlfriends are assets.

Then, a lightbulb.

Andrew and Tristan messaged and flew 7 women to a posh restaurant. Selling desperation masked as opportunity, Andrew pitched his master plan:

A webcam business — with a 50/50 revenue split.

5 women left.

2 stayed.

Andrew Tate, a broke kickboxer, became Top G, Cobra Tate.

The business model was simple — yet sad.

Andrew's girlfriends moved in with him and spoke online for 15+ hours a day. Andrew handled ads and equipment as the women posed.

Andrew eventually took over their keyboards, believing he knew what men wanted more than women.

Andrew detailed on the Full Send Podcast how he emotionally manipulated men for millions. They sold houses, automobiles, and life savings to fuel their companionship addiction.

When asked if he felt bad, Andrew said,

“F*ck no.“

Andrew and Tristan wiped off debts, hired workers, and diversified.

Tristan supervised OnlyFans models.

Andrew bought Romanian casinos and MMA league RXF (Real Xtreme Fighting).

Pandemic struck suddenly.

Andrew couldn't run his 2 businesses without a plan. Another easy moneymaker.

He banked on Hustlers University.

The actual cause of Andrew's ubiquity.

On a Your Mom’s House episode Andrew's 4 main revenue sources:

1. Hustler’s University

2. Owning casinos in Romania

3. Owning 10% of the Romanian MMA league “RXF”

4. “The War Room” — a society of rich and powerful men

When the pandemic hit, 3/4 became inoperable.

So he expanded Hustlers University.

But what is Hustler’s University?

Andrew says Hustlers University teaches 18 wealth-building tactics online. Examples:

• Real estate

• Copywriting

• Amazon FBA

• Dropshipping

• Flipping Cryptos

How to swiftly become wealthy.

Lessons are imprecise, rudimentary, and macro-focused, say reviews. Invest wisely, etc. Everything is free online.

You pay for community. One unique income stream.

The only money-making mechanism that keeps the course from being a scam.

The truth is, many of Andrew’s students are actually making money. Maybe not from the free YouTube knowledge Andrew and his professors teach in the course, but through Hustler’s University’s affiliate program.

Affiliates earn 10% commission for each new student = $5.

Students can earn $10 for each new referral in the first two months.

Andrew earns $50 per membership per month.

This affiliate program isn’t anything special — in fact, it’s on the lower end of affiliate payouts. Normally, it wouldn’t be very lucrative.

But it has one secret weapon— Andrew and his viral opinions.

Andrew is viral. Andrew went on a media tour in January 2022 after appearing on Your Mom's House.

• The Full Send Podcast

• The Fresh & Fit Podcast

• “BFFs” by Barstool Sports

And many, many more…

He chatted with Twitch streamers. Hustlers University wanted more controversy (and clips).

Here’s the strategy behind Hustler’s University that has (allegedly) earned students upwards of $10K per month:

1. Make a social media profile with Andrew Tates' name and photo.

2. Post any of the online videos of Andrews that have gone viral.

3. Include a referral link in your bio.

Effectively simple.

Andrew's controversy attracts additional students. More student clips circulate as more join. Andrew's students earn more and promote the product as he goes viral.

A brilliant plan that's functioning.

At the beginning of his media tour, Hustler’s University had 5,000 students. 6 months in, and he now has over 100,000.

One income stream generates $5 million every month.

Andrew's approach is not new.

But it is different.

In the early 2010s, Tai Lopez dominated the internet.

His viral video showed his house.

“Here in my garage. Just bought this new Lamborghini.”

Tais' marketing focused on intellect, not strength, power, and wealth to attract women.

How reading quicker leads to financial freedom in 67 steps.

Years later, it was revealed that Tai Lopez rented the mansion and Lamborghini as a marketing ploy to build social proof. Meanwhile, he was living in his friend’s trailer.

Faked success is an old tactic.

Andrew is doing something similar. But with one major distinction.

Andrew outsources his virality — making him nearly impossible to cancel.

In 2022, authorities searched Andrews' estate over human trafficking suspicions. Investigation continues despite withdrawn charges.

Andrew's divisive nature would normally get him fired. Andrew's enterprises and celebrity don't rely on social media.

He doesn't promote or pay for ads. Instead, he encourages his students and anyone wishing to get rich quick to advertise his work.

Because everything goes through his affiliate program. Old saying:

“All publicity is good publicity.”

Final thoughts: it’s ok to feel triggered.

Tate is divisive.

His emotionally charged words are human nature. Andrews created the controversy.

It's non-personal.

His opinions are those of one person. Not world nor generational opinion.

Briefly:

• It's easy to understand why Andrews' face is ubiquitous. Money.

• The world wide web is a chessboard. Misdirection is part of it.

• It’s not personal, it’s business.

• Controversy sells

Sometimes understanding the ‘why’, can help you deal with the ‘what.’

Deep learning-based generative art is being researched. As usual, self-learning is better. Some models, like OpenAI's DALL-E 2, require registration and can only be used online, but others can be used locally, which is usually more enjoyable for curious users. I'll demonstrate the Stable Diffusion model's operation on a standard PC.

Let’s get started.

What It Does

Stable Diffusion uses numerous components:

• A generative model trained to produce images is called a diffusion model. The model is incrementally improving the starting data, which is only random noise. The model has an image, and while it is being trained, the reversed process is being used to add noise to the image. Being able to reverse this procedure and create images from noise is where the true magic is (more details and samples can be found in the paper).

• An internal compressed representation of a latent diffusion model, which may be altered to produce the desired images, is used (more details can be found in the paper). The capacity to fine-tune the generation process is essential because producing pictures at random is not very attractive (as we can see, for instance, in Generative Adversarial Networks).

• A neural network model called CLIP (Contrastive Language-Image Pre-training) is used to translate natural language prompts into vector representations. This model, which was trained on 400,000,000 image-text pairs, enables the transformation of a text prompt into a latent space for the diffusion model in the scenario of stable diffusion (more details in that paper).

This figure shows all data flow:

The weights file size for Stable Diffusion model v1 is 4 GB and v2 is 5 GB, making the model quite huge. The v1 model was trained on 256x256 and 512x512 LAION-5B pictures on a 4,000 GPU cluster using over 150.000 NVIDIA A100 GPU hours. The open-source pre-trained model is helpful for us. And we will.

Install

Before utilizing the Python sources for Stable Diffusion v1 on GitHub, we must install Miniconda (assuming Git and Python are already installed):

wget https://repo.anaconda.com/miniconda/Miniconda3-py39_4.12.0-Linux-x86_64.sh
chmod +x Miniconda3-py39_4.12.0-Linux-x86_64.sh
./Miniconda3-py39_4.12.0-Linux-x86_64.sh
conda update -n base -c defaults conda

Install the source and prepare the environment:

git clone https://github.com/CompVis/stable-diffusion
cd stable-diffusion
conda env create -f environment.yaml
conda activate ldm
pip3 install transformers --upgrade

Download the pre-trained model weights next. HiggingFace has the newest checkpoint sd-v14.ckpt (a download is free but registration is required). Put the file in the project folder and have fun:

python3 scripts/txt2img.py --prompt "hello world" --plms --ckpt sd-v1-4.ckpt --skip_grid --n_samples 1

Almost. The installation is complete for happy users of current GPUs with 12 GB or more VRAM. RuntimeError: CUDA out of memory will occur otherwise. Two solutions exist.

Running the optimized version

Try optimizing first. After cloning the repository and enabling the environment (as previously), we can run the command:

python3 optimizedSD/optimized_txt2img.py --prompt "hello world" --ckpt sd-v1-4.ckpt --skip_grid --n_samples 1

Stable Diffusion worked on my visual card with 8 GB RAM (alas, I did not behave well enough to get NVIDIA A100 for Christmas, so 8 GB GPU is the maximum I have;).

Running Stable Diffusion without GPU

If the GPU does not have enough RAM or is not CUDA-compatible, running the code on a CPU will be 20x slower but better than nothing. This unauthorized CPU-only branch from GitHub is easiest to obtain. We may easily edit the source code to use the latest version. It's strange that a pull request for that was made six months ago and still hasn't been approved, as the changes are simple. Readers can finish in 5 minutes:

• Replace if attr.device!= torch.device(cuda) with if attr.device!= torch.device(cuda) and torch.cuda.is available at line 20 of ldm/models/diffusion/ddim.py ().

• Replace if attr.device!= torch.device(cuda) with if attr.device!= torch.device(cuda) and torch.cuda.is available in line 20 of ldm/models/diffusion/plms.py ().

• Replace device=cuda in lines 38, 55, 83, and 142 of ldm/modules/encoders/modules.py with device=cuda if torch.cuda.is available(), otherwise cpu.

• Replace model.cuda() in scripts/txt2img.py line 28 and scripts/img2img.py line 43 with if torch.cuda.is available(): model.cuda ().

Run the script again.

Testing

Test the model. Text-to-image is the first choice. Test the command line example again:

python3 scripts/txt2img.py --prompt "hello world" --plms --ckpt sd-v1-4.ckpt --skip_grid --n_samples 1

The slow generation takes 10 seconds on a GPU and 10 minutes on a CPU. Final image:

Hello world is dull and abstract. Try a brush-wielding hamster. Why? Because we can, and it's not as insane as Napoleon's cat. Another image:

Generating an image from a text prompt and another image is interesting. I made this picture in two minutes using the image editor (sorry, drawing wasn't my strong suit):

I can create an image from this drawing:

python3 scripts/img2img.py --prompt "A bird is sitting on a tree branch" --ckpt sd-v1-4.ckpt --init-img bird.png --strength 0.8

It was far better than my initial drawing:

I hope readers understand and experiment.

Stable Diffusion UI

Developers love the command line, but regular users may struggle. Stable Diffusion UI projects simplify image generation and installation. Simple usage:

• Unpack the ZIP after downloading it from https://github.com/cmdr2/stable-diffusion-ui/releases. Linux and Windows are compatible with Stable Diffusion UI (sorry for Mac users, but those machines are not well-suitable for heavy machine learning tasks anyway;).

• Start the script.

Done. The web browser UI makes configuring various Stable Diffusion features (upscaling, filtering, etc.) easy:

V2.1 of Stable Diffusion

I noticed the notification about releasing version 2.1 while writing this essay, and it was intriguing to test it. First, compare version 2 to version 1:

• alternative text encoding. The Contrastive LanguageImage Pre-training (CLIP) deep learning model, which was trained on a significant number of text-image pairs, is used in Stable Diffusion 1. The open-source CLIP implementation used in Stable Diffusion 2 is called OpenCLIP. It is difficult to determine whether there have been any technical advancements or if legal concerns were the main focus. However, because the training datasets for the two text encoders were different, the output results from V1 and V2 will differ for the identical text prompts.

• a new depth model that may be used to the output of image-to-image generation.

• a revolutionary upscaling technique that can quadruple the resolution of an image.

• Generally higher resolution Stable Diffusion 2 has the ability to produce both 512x512 and 768x768 pictures.

The Hugging Face website offers a free online demo of Stable Diffusion 2.1 for code testing. The process is the same as for version 1.4. Download a fresh version and activate the environment:

conda deactivate  
conda env remove -n ldm  # Use this if version 1 was previously installed
git clone https://github.com/Stability-AI/stablediffusion
cd stablediffusion
conda env create -f environment.yaml
conda activate ldm

Hugging Face offers a new weights ckpt file.

The Out of memory error prevented me from running this version on my 8 GB GPU. Version 2.1 fails on CPUs with the slow conv2d cpu not implemented for Half error (according to this GitHub issue, the CPU support for this algorithm and data type will not be added). The model can be modified from half to full precision (float16 instead of float32), however it doesn't make sense since v1 runs up to 10 minutes on the CPU and v2.1 should be much slower. The online demo results are visible. The same hamster painting with a brush prompt yielded this result:

It looks different from v1, but it functions and has a higher resolution.

The superresolution.py script can run the 4x Stable Diffusion upscaler locally (the x4-upscaler-ema.ckpt weights file should be in the same folder):

python3 scripts/gradio/superresolution.py configs/stable-diffusion/x4-upscaling.yaml x4-upscaler-ema.ckpt

This code allows the web browser UI to select the image to upscale:

The copy-paste strategy may explain why the upscaler needs a text prompt (and the Hugging Face code snippet does not have any text input as well). I got a GPU out of memory error again, although CUDA can be disabled like v1. However, processing an image for more than two hours is unlikely:

Stable Diffusion Limitations

When we use the model, it's fun to see what it can and can't do. Generative models produce abstract visuals but not photorealistic ones. This fundamentally limits The generative neural network was trained on text and image pairs, but humans have a lot of background knowledge about the world. The neural network model knows nothing. If someone asks me to draw a Chinese text, I can draw something that looks like Chinese but is actually gibberish because I never learnt it. Generative AI does too! Humans can learn new languages, but the Stable Diffusion AI model includes only language and image decoder brain components. For instance, the Stable Diffusion model will pull NO WAR banner-bearers like this:

V1:

V2.1:

The shot shows text, although the model never learned to read or write. The model's string tokenizer automatically converts letters to lowercase before generating the image, so typing NO WAR banner or no war banner is the same.

I can also ask the model to draw a gorgeous woman:

V1:

V2.1:

The first image is gorgeous but physically incorrect. A second one is better, although it has an Uncanny valley feel. BTW, v2 has a lifehack to add a negative prompt and define what we don't want on the image. Readers might try adding horrible anatomy to the gorgeous woman request.

If we ask for a cartoon attractive woman, the results are nice, but accuracy doesn't matter:

V1:

V2.1:

Another example: I ordered a model to sketch a mouse, which looks beautiful but has too many legs, ears, and fingers:

V1:

V2.1: improved but not perfect.

V1 produces a fun cartoon flying mouse if I want something more abstract:

I tried multiple times with V2.1 but only received this:

The image is OK, but the first version is closer to the request.

Stable Diffusion struggles to draw letters, fingers, etc. However, abstract images yield interesting outcomes. A rural landscape with a modern metropolis in the background turned out well:

V1:

V2.1:

Generative models help make paintings too (at least, abstract ones). I searched Google Image Search for modern art painting to see works by real artists, and this was the first image:

I typed "abstract oil painting of people dancing" and got this:

V1:

V2.1:

It's a different style, but I don't think the AI-generated graphics are worse than the human-drawn ones.

The AI model cannot think like humans. It thinks nothing. A stable diffusion model is a billion-parameter matrix trained on millions of text-image pairs. I input "robot is creating a picture with a pen" to create an image for this post. Humans understand requests immediately. I tried Stable Diffusion multiple times and got this:

This great artwork has a pen, robot, and sketch, however it was not asked. Maybe it was because the tokenizer deleted is and a words from a statement, but I tried other requests such robot painting picture with pen without success. It's harder to prompt a model than a person.

I hope Stable Diffusion's general effects are evident. Despite its limitations, it can produce beautiful photographs in some settings. Readers who want to use Stable Diffusion results should be warned. Source code examination demonstrates that Stable Diffusion images feature a concealed watermark (text StableDiffusionV1 and SDV2) encoded using the invisible-watermark Python package. It's not a secret, because the official Stable Diffusion repository's test watermark.py file contains a decoding snippet. The put watermark line in the txt2img.py source code can be removed if desired. I didn't discover this watermark on photographs made by the online Hugging Face demo. Maybe I did something incorrectly (but maybe they are just not using the txt2img script on their backend at all).

Conclusion

The Stable Diffusion model was fascinating. As I mentioned before, trying something yourself is always better than taking someone else's word, so I encourage readers to do the same (including this article as well;).

Is Generative AI a game-changer? My humble experience tells me:

• I think that place has a lot of potential. For designers and artists, generative AI can be a truly useful and innovative tool. Unfortunately, it can also pose a threat to some of them since if users can enter a text field to obtain a picture or a website logo in a matter of clicks, why would they pay more to a different party? Is it possible right now? unquestionably not yet. Images still have a very poor quality and are erroneous in minute details. And after viewing the image of the stunning woman above, models and fashion photographers may also unwind because it is highly unlikely that AI will replace them in the upcoming years.

• Today, generative AI is still in its infancy. Even 768x768 images are considered to be of a high resolution when using neural networks, which are computationally highly expensive. There isn't an AI model that can generate high-resolution photographs natively without upscaling or other methods, at least not as of the time this article was written, but it will happen eventually.

• It is still a challenge to accurately represent knowledge in neural networks (information like how many legs a cat has or the year Napoleon was born). Consequently, AI models struggle to create photorealistic photos, at least where little details are important (on the other side, when I searched Google for modern art paintings, the results are often even worse;).

• When compared to the carefully chosen images from official web pages or YouTube reviews, the average output quality of a Stable Diffusion generation process is actually less attractive because to its high degree of randomness. When using the same technique on their own, consumers will theoretically only view those images as 1% of the results.

Anyway, it's exciting to witness this area's advancement, especially because the project is open source. Google's Imagen and DALL-E 2 can also produce remarkable findings. It will be interesting to see how they progress.

Alex Mashinsky, CEO of Celsius, speaks at Web Summit 2021 in Lisbon. 

Celsius Network filed for Chapter 11 bankruptcy a month after freezing customer withdrawals, joining other crypto casualties.

Celsius took the step to stabilize its business and restructure for all stakeholders. The filing was done in the Southern District of New York.

The company, which amassed more than $20 billion by offering 18% interest on cryptocurrency deposits, paused withdrawals and other functions in mid-June, citing "extreme market conditions."

As the Fed raises interest rates aggressively, it hurts risk sentiment and squeezes funding costs. Voyager Digital Ltd. filed for Chapter 11 bankruptcy this month, and Three Arrows Capital has called in liquidators.

Celsius called the pause "difficult but necessary." Without the halt, "the acceleration of withdrawals would have allowed certain customers to be paid in full while leaving others to wait for Celsius to harvest value from illiquid or longer-term asset deployment activities," it said.

Celsius declined to comment. CEO Alex Mashinsky said the move will strengthen the company's future.

The company wants to keep operating. It's not requesting permission to allow customer withdrawals right now; Chapter 11 will handle customer claims. The filing estimates assets and liabilities between $1 billion and $10 billion.

Celsius is advised by Kirkland & Ellis, Centerview Partners, and Alvarez & Marsal.

Yield-promises

Celsius promised 18% returns on crypto loans. It lent those coins to institutional investors and participated in decentralized-finance apps.

When TerraUSD (UST) and Luna collapsed in May, Celsius pulled its funds from Terra's Anchor Protocol, which offered 20% returns on UST deposits. Recently, another large holding, staked ETH, or stETH, which is tied to Ether, became illiquid and discounted to Ether.

The lender is one of many crypto companies hurt by risky bets in the bear market. Also, Babel halted withdrawals. Voyager Digital filed for bankruptcy, and crypto hedge fund Three Arrows Capital filed for Chapter 15 bankruptcy.

According to blockchain data and tracker Zapper, Celsius repaid all of its debt in Aave, Compound, and MakerDAO last month.

Celsius charged Symbolic Capital Partners Ltd. 2,000 Ether as collateral for a cash loan on June 13. According to company filings, Symbolic was charged 2,545.25 Ether on June 11.

In July 6 filings, it said it reshuffled its board, appointing two new members and firing others.