After 10 years of working in scale-ups, I've embraced a few concepts for scaling Tech and Product teams.

First, cross-functionalize teams. Product Managers represent the business, Product Designers the consumer, and Engineers build.

I organize teams of 5-10 individuals, following AWS's two pizza teams guidelines, with a Product Trio guiding each.

If more individuals are needed to reach a goal, I group teams under a Product Trio.

With Engineering being the biggest group, Staff/Principal Engineers often support the Trio on cross-team technical decisions.

Product Managers, Engineering Managers, or Engineers in the team may manage projects (depending on the project or aim), but the trio is collectively responsible for the team's output and outcome.

Once the Product Trio model is created, roles, duties, team ceremonies, and cooperation models must be clarified.

Keep reporting lines by discipline. Line managers are accountable for each individual's advancement, thus it's crucial that they know the work in detail.

Cross-team collaboration becomes more important after 3 teams (15-30 people). Teams can easily diverge in how they write code, run ceremonies, and build products.

Establishing groups of people that are cross-team, but grouped by discipline and skills, sharing and agreeing on working practices becomes critical.

The “Spotify Guild” model has been where I’ve taken a lot of my inspiration from.

Last, establish a taxonomy for communication channels.

In Slack, I create one channel per team and one per guild (and one for me to have discussions with the team leads).

These are just some of the basic principles I follow to organize teams.

A book I particularly like about team types and how they interact with each other is https://teamtopologies.com/.

Y Combinators Winter 2022 Demo Day continues the trend of more startups engaging in accelerator Demo Days. Our team evaluated almost 400 projects in Y Combinator's ninth year.

After Winter 2021 Demo Day, we noticed a hurry pushing shorter rounds, inflated valuations, and larger batches.

Despite the batch size, this event's behavior showed a return to normalcy. Our observations show that investors evaluate and fund businesses more carefully. Unlike previous years, more YC businesses gave investors with data rooms and thorough pitch decks in addition to valuation data before Demo Day.

Demo Day pitches were virtual and fast-paced, limiting unplanned meetings. Investors had more time and information to do their due research before meeting founders. Our staff has more time to study diverse areas and engage with interesting entrepreneurs and founders.

This was one of the most regionally diversified YC cohorts to date. This year's Winter Demo Day startups showed some interesting tendencies.

Trends and Industries to Watch Before Demo Day

Demo day events at any accelerator show how investment competition is influencing startups. As startups swiftly become scale-ups and big success stories in fintech, e-commerce, healthcare, and other competitive industries, entrepreneurs and early-stage investors feel pressure to scale quickly and turn a notion into actual innovation.

Too much eagerness can lead founders to focus on market growth and team experience instead of solid concepts, technical expertise, and market validation. Last year, YC Winter Demo Day funding cycles ended too quickly and valuations were unrealistically high.

Scrum Ventures observed a longer funding cycle this year compared to last year's Demo Day. While that seems promising, many factors could be contributing to change, including:

• Market patterns are changing and the economy is becoming worse.

• the industries that investors are thinking about.

• Individual differences between each event batch and the particular businesses and entrepreneurs taking part

The Winter 2022 Batch's Trends

Each year, we also wish to examine trends among early-stage firms and YC event participants. More international startups than ever were anticipated to present at Demo Day.

Less than 50% of demo day startups were from the U.S. For the S21 batch, firms from outside the US were most likely in Latin America or Europe, however this year's batch saw a large surge in startups situated in Asia and Africa.

YC Startup Directory

163 out of 399 startups were B2B software and services companies. Financial, healthcare, and consumer startups were common.

Our team doesn't plan to attend every pitch or speak with every startup's founders or team members. Let's look at cleantech, Web3, and health and wellness startup trends.

Our Opinions Following Conversations with 87 Startups at Demo Day

In the lead-up to Demo Day, we spoke with 87 of the 125 startups going. Compared to B2C enterprises, B2B startups had higher average valuations. A few outliers with high valuations pushed B2B and B2C means above the YC-wide mean and median.

Many of these startups develop business and technology solutions we've previously covered. We've seen API, EdTech, creative platforms, and cybersecurity remain strong and increase each year.

While these persistent tendencies influenced the startups Scrum Ventures looked at and the founders we interacted with on Demo Day, new trends required more research and preparation. Let's examine cleantech, Web3, and health and wellness startups.

Hardware and software that is green

Cleantech enterprises demand varying amounts of funding for hardware and software. Although the same overarching trend is fueling the growth of firms in this category, each subgroup has its own strategy and technique for investigation and identifying successful investments.

Many cleantech startups we spoke to during the YC event are focused on helping industrial operations decrease or recycle carbon emissions.

• Carbon Crusher: Creating carbon negative roads

• Phase Biolabs: Turning carbon emissions into carbon negative products and carbon neutral e-fuels

• Seabound: Capturing carbon dioxide emissions from ships

• Fleetzero: Creating electric cargo ships

• Impossible Mining: Sustainable seabed mining

• Beyond Aero: Creating zero-emission private aircraft

• Verdn: Helping businesses automatically embed environmental pledges for product and service offerings, boost customer engagement

• AeonCharge: Allowing electric vehicle (EV) drivers to more easily locate and pay for EV charging stations

• Phoenix Hydrogen: Offering a hydrogen marketplace and a connected hydrogen hub platform to connect supply and demand for hydrogen fuel and simplify hub planning and partner program expansion

• Aklimate: Allowing businesses to measure and reduce their supply chain’s environmental impact

• Pina Earth: Certifying and tracking the progress of businesses’ forestry projects

• AirMyne: Developing machines that can reverse emissions by removing carbon dioxide from the air

• Unravel Carbon: Software for enterprises to track and reduce their carbon emissions

Web3: NFTs, the metaverse, and cryptocurrency

Web3 technologies handle a wide range of business issues. This category includes companies employing blockchain technology to disrupt entertainment, finance, cybersecurity, and software development.

Many of these startups overlap with YC's FinTech trend. Despite this, B2C and B2B enterprises were evenly represented in Web3. We examined:

• Stablegains: Offering consistent interest on cash balance from the decentralized finance (DeFi) market

• LiquiFi: Simplifying token management with automated vesting contracts, tax reporting, and scheduling. For companies, investors, and finance & accounting

• NFTScoring: An NFT trading platform

• CypherD Wallet: A multichain wallet for crypto and NFTs with a non-custodial crypto debit card that instantly converts coins to USD

• Remi Labs: Allowing businesses to more easily create NFT collections that serve as access to products, memberships, events, and more

• Cashmere: A crypto wallet for Web3 startups to collaboratively manage funds

• Chaingrep: An API that makes blockchain data human-readable and tokens searchable

• Courtyard: A platform for securely storing physical assets and creating 3D representations as NFTs

• Arda: “Banking as a Service for DeFi,” an API that FinTech companies can use to embed DeFi products into their platforms

• earnJARVIS: A premium cryptocurrency management platform, allowing users to create long-term portfolios

• Mysterious: Creating community-specific experiences for Web3 Discords

• Winter: An embeddable widget that allows businesses to sell NFTs to users purchasing with a credit card or bank transaction

• SimpleHash: An API for NFT data that provides compatibility across blockchains, standardized metadata, accurate transaction info, and simple integration

• Lifecast: Tools that address motion sickness issues for 3D VR video

• Gym Class: Virtual reality (VR) multiplayer basketball video game

• WorldQL: An asset API that allows NFT creators to specify multiple in-game interpretations of their assets, increasing their value

• Bonsai Desk: A software development kit (SDK) for 3D analytics

• Campfire: Supporting virtual social experiences for remote teams

• Unai: A virtual headset and Visual World experience

• Vimmerse: Allowing creators to more easily create immersive 3D experiences

Fitness and health

Scrum Ventures encountered fewer health and wellness startup founders than Web3 and Cleantech. The types of challenges these organizations solve are still diverse. Several of these companies are part of a push toward customization in healthcare, an area of biotech set for growth for companies with strong portfolios and experienced leadership.

Here are several startups we considered:

• Syrona Health: Personalized healthcare for women in the workplace

• Anja Health: Personalized umbilical cord blood banking and stem cell preservation

• Alfie: A weight loss program focused on men’s health that coordinates medical care, coaching, and “community-based competition” to help users lose an average of 15% body weight

• Ankr Health: An artificial intelligence (AI)-enabled telehealth platform that provides personalized side effect education for cancer patients and data collection for their care teams

• Koko — A personalized sleep program to improve at-home sleep analysis and training

• Condition-specific telehealth platforms and programs:

• Reviving Mind: Chronic care management covered by insurance and supporting holistic, community-oriented health care

• Equipt Health: At-home delivery of prescription medical equipment to help manage chronic conditions like obstructive sleep apnea

• LunaJoy: Holistic women’s healthcare management for mental health therapy, counseling, and medication

12 Startups from YC's Winter 2022 Demo Day to Watch

Bobidi: 10x faster AI model improvement

Artificial intelligence (AI) models have become a significant tool for firms to improve how well and rapidly they process data. Bobidi helps AI-reliant firms evaluate their models, boosting data insights in less time and reducing data analysis expenditures. The business has created a gamified community that offers a bug bounty for AI, incentivizing community members to test and find weaknesses in clients' AI models.

Magna: DeFi investment management and token vesting

Magna delivers rapid, secure token vesting so consumers may turn DeFi investments into primitives. Carta for Web3 allows enterprises to effortlessly distribute tokens to staff or investors. The Magna team hopes to allow corporations use locked tokens as collateral for loans, facilitate secondary liquidity so investors can sell shares on a public exchange, and power additional DeFi applications.

Perl Street: Funding for infrastructure

This Fintech firm intends to help hardware entrepreneurs get financing by [democratizing] structured finance, unleashing billions for sustainable infrastructure and next-generation hardware solutions. This network has helped hardware entrepreneurs achieve more than $140 million in finance, helping companies working on energy storage devices, EVs, and creating power infrastructure.

CypherD: Multichain cryptocurrency wallet

CypherD seeks to provide a multichain crypto wallet so general customers can explore Web3 products without knowledge hurdles. The startup's beta app lets consumers access crypto from EVM blockchains. The founders have crypto, financial, and startup experience.

Unravel Carbon: Enterprise carbon tracking and offsetting

Unravel Carbon's AI-powered decarbonization technology tracks companies' carbon emissions. Singapore-based startup focuses on Asia. The software can use any company's financial data to trace the supply chain and calculate carbon tracking, which is used to make regulatory disclosures and suggest carbon offsets.

LunaJoy: Precision mental health for women

LunaJoy helped women obtain mental health support throughout life. The platform combines data science to create a tailored experience, allowing women to access psychotherapy, medication management, genetic testing, and health coaching.

Posh: Automated EV battery recycling

Posh attempts to solve one of the EV industry's largest logistical difficulties. Millions of EV batteries will need to be decommissioned in the next decade, and their precious metals and residual capacity will go unused for some time. Posh offers automated, scalable lithium battery disassembly, making EV battery recycling more viable.

Unai: VR headset with 5x higher resolution

Unai stands apart from metaverse companies. Its VR headgear has five times the resolution of existing options and emphasizes human expression and interaction in a remote world. Maxim Perumal's method of latency reduction powers current VR headsets.

Palitronica: Physical infrastructure cybersecurity

Palitronica blends cutting-edge hardware and software to produce networked electronic systems that support crucial physical and supply chain infrastructure. The startup's objective is to build solutions that defend national security and key infrastructure from cybersecurity threats.

Reality Defender: Deepfake detection

Reality Defender alerts firms to bogus users and changed audio, video, and image files. Reality Deference's API and web app score material in real time to prevent fraud, improve content moderation, and detect deception.

Micro Meat: Infrastructure for the manufacture of cell-cultured meat

MicroMeat promotes sustainable meat production. The company has created technologies to scale up bioreactor-grown meat muscle tissue from animal cells. Their goal is to scale up cultured meat manufacturing so cultivated meat products can be brought to market feasibly and swiftly, boosting worldwide meat consumption.

Fleetzero: Electric cargo ships

This startup's battery technology will make cargo ships more sustainable and profitable. Fleetzero's electric cargo ships have five times larger profit margins than fossil fuel ships. Fleetzeros' founder has marine engineering, ship operations, and enterprise sales and business experience.

Know why it's important.

The entrepreneur I was meeting with said, "I should be doing crypto, or maybe AI? Aren't those the hot spots? I should look there for a startup idea.”

I shook my head. Yes, they're exciting, but that doesn't mean they're best for you and your business.

“There are different types of entrepreneurs?” he asked.

I said "obviously." Two types, actually. Knowing what type of entrepreneur you are helps you build the right startup.

The two types of businesspeople

The best way for me to describe the two types of entrepreneurs is to start by telling you exactly the kinds of entrepreneurial opportunities I never get excited about: future opportunities.

In the early 1990s, my older brother showed me the World Wide Web and urged me to use it. Unimpressed, I returned to my Super Nintendo.

My roommate tried to get me to join Facebook as a senior in college. I remember thinking, This is dumb. Who'll use it?

In 2011, my best friend tried to convince me to buy bitcoin and I laughed.

Heck, a couple of years ago I had to buy a new car, and I never even considered buying something that didn’t require fossilized dinosaur bones.

I'm no visionary. I don't anticipate the future. I focus on the present.

This tendency makes me a problem-solving entrepreneur. I identify entrepreneurial opportunities by spotting flaws and/or inefficiencies in the world and devising solutions.

There are other ways to find business opportunities. Visionary entrepreneurs also exist. I don't mean visionary in the hyperbolic sense that implies world-changing impact. I mean visionary as an entrepreneur who identifies future technological shifts that will change how people work and live and create new markets.

Problem-solving and visionary entrepreneurs are equally good. But the two approaches to building companies are very different. Knowing the type of entrepreneur you are will help you build a startup that fits your worldview.

What is the distinction?

Let's use some simple hypotheticals to compare problem-solving and visionary entrepreneurship.

Imagine a city office building without nearby restaurants. Those office workers love to eat. Sometimes they'd rather eat out than pack a lunch. As an entrepreneur, you can solve the lack of nearby restaurants. You'd open a restaurant near that office, say a pizza parlor, and get customers because you solved the lack of nearby restaurants. Problem-solving entrepreneurship.

Imagine a new office building in a developing area with no residents or workers. In this scenario, a large office building is coming. The workers will need to eat then. As a visionary entrepreneur, you're excited about the new market and decide to open a pizzeria near the construction to meet demand.

Both possibilities involve the same product. You opened a pizzeria. How you launched that pizza restaurant and what will affect its success are different.

Why is the distinction important?

Let's say you opened a pizzeria near an office. You'll probably get customers. Because people are nearby and demand isn't being met, someone from a nearby building will stop in within the first few days of your pizzeria's grand opening. This makes solving the problem relatively risk-free. You'll get customers unless you're a fool.

The market you're targeting existed before you entered it, so you're not guaranteed success. This means people in that market solved the lack of nearby restaurants. Those office workers are used to bringing their own lunches. Why should your restaurant change their habits? Even when they eat out, they're used to traveling far. They've likely developed pizza preferences.

To be successful with your problem-solving startup, you must convince consumers to change their behavior, which is difficult.

Unlike opening a pizza restaurant near a construction site. Once the building opens, workers won't have many preferences or standardized food-getting practices. Your pizza restaurant can become the incumbent quickly. You'll be the first restaurant in the area, so you'll gain a devoted following that makes your food a routine.

Great, right? It's easier than changing people's behavior. The benefit comes with a risk. Opening a pizza restaurant near a construction site increases future risk. What if builders run out of money? No one moves in? What if the building's occupants are the National Association of Pizza Haters? Then you've opened a pizza restaurant next to pizza haters.

Which kind of businessperson are you?

This isn't to say one type of entrepreneur is better than another. Each type of entrepreneurship requires different skills.

As my simple examples show, a problem-solving entrepreneur must operate in markets with established behaviors and habits. To be successful, you must be able to teach a market a new way of doing things.

Conversely, the challenge of being a visionary entrepreneur is that you have to be good at predicting the future and getting in front of that future before other people.

Both are difficult in different ways. So, smart entrepreneurs don't just chase opportunities. Smart entrepreneurs pursue opportunities that match their skill sets.

Ethereum is the second-largest cryptocurrency. The Merge, a mid-September event that will convert Ethereum's consensus process from proof-of-work to proof-of-stake if all goes according to plan, will be a game changer.

Why is Ethereum ditching proof-of-work? Because it can. We're talking about a fully functioning, open-source ecosystem with a capacity for evolution that other cryptocurrencies lack, a change that would allow it to scale up its performance from 15 transactions per second to 100,000 as its blockchain is used for more and more things. It would reduce its energy consumption by 99.95%. Vitalik Buterin, the system's founder, would play a less active role due to decentralization, and miners, who validated transactions through proof of work, would be far less important.

Why has this conversion taken so long and been so cautious? Because it involves modifying a core process while it's running to boost its performance. It requires running the new mechanism in test chains on an ever-increasing scale, assessing participant reactions, and checking for issues or restrictions. The last big test was in early June and was successful. All that's left is to converge the mechanism with the Ethereum blockchain to conclude the switch.

What's stopping Bitcoin, the leader in market capitalization and the cryptocurrency that began blockchain's appeal, from doing the same? Satoshi Nakamoto, whoever he or she is, departed from public life long ago, therefore there's no community leadership. Changing it takes a level of consensus that is impossible to achieve without strong leadership, which is why Bitcoin's evolution has been sluggish and conservative, with few modifications.

Secondly, The Merge will balance the consensus mechanism (proof-of-work or proof-of-stake) and the system decentralization or centralization. Proof-of-work prevents double-spending, thus validators must buy hardware. The system works, but it requires a lot of electricity and, as it scales up, tends to re-centralize as validators acquire more hardware and the entire network activity gets focused in a few nodes. Larger operations save more money, which increases profitability and market share. This evolution runs opposed to the concept of decentralization, and some anticipate that any system that uses proof of work as a consensus mechanism will evolve towards centralization, with fewer large firms able to invest in efficient network nodes.

Yet radical bitcoin enthusiasts share an opposite argument. In proof-of-stake, transaction validators put their funds at stake to attest that transactions are valid. The algorithm chooses who validates each transaction, giving more possibilities to nodes that put more coins at stake, which could open the door to centralization and government control.

In both cases, we're talking about long-term changes, but Bitcoin's proof-of-work has been evolving longer and seems to confirm those fears, while proof-of-stake is only employed in coins with a minuscule volume compared to Ethereum and has no predictive value.

As of mid-September, we will have two significant cryptocurrencies, each with a different consensus mechanisms and equally different characteristics: one is intrinsically conservative and used only for economic transactions, while the other has been evolving in open source mode, and can be used for other types of assets, smart contracts, or decentralized finance systems. Some even see it as the foundation of Web3.

Many things could change before September 15, but The Merge is likely to be a turning point. We'll have to follow this closely.

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.

"GTO" (Game Theory Optimal) has been used a lot in poker recently. To clarify its meaning and application, the aim of this article is to define what it is, when to use it when playing, what strategies to apply for how to play GTO poker, for beginner and more advanced players!

Poker GTO

In poker, you can choose between two main winning strategies:

Exploitative play maximizes expected value (EV) by countering opponents' sub-optimal plays and weaker tendencies. Yes, playing this way opens you up to being exploited, but the weaker opponents you're targeting won't change their game to counteract this, allowing you to reap maximum profits over the long run.

GTO (Game-Theory Optimal): You try to play perfect poker, which forces your opponents to make mistakes (which is where almost all of your profit will be derived from). It mixes bluffs or semi-bluffs with value bets, clarifies bet sizes, and more.

GTO vs. Exploitative: Which is Better in Poker?

Before diving into GTO poker strategy, it's important to know which of these two play styles is more profitable for beginners and advanced players. The simple answer is probably both, but usually more exploitable.

Most players don't play GTO poker and can be exploited in their gameplay and strategy, allowing for more profits to be made using an exploitative approach. In fact, it’s only in some of the largest games at the highest stakes that GTO concepts are fully utilized and seen in practice, and even then, exploitative plays are still sometimes used.

Knowing, understanding, and applying GTO poker basics will create a solid foundation for your poker game. It's also important to understand GTO so you can deviate from it to maximize profits.

GTO Poker Strategy

According to Ed Miller's book "Poker's 1%," the most fundamental concept that only elite poker players understand is frequency, which could be in relation to cbets, bluffs, folds, calls, raises, etc.

GTO poker solvers (downloadable online software) give solutions for how to play optimally in any given spot and often recommend using mixed strategies based on select frequencies.

In a river situation, a solver may tell you to call 70% of the time and fold 30%. It may also suggest calling 50% of the time, folding 35% of the time, and raising 15% of the time (with a certain range of hands).

Frequencies are a fundamental and often unrecognized part of poker, but they run through these 5 GTO concepts.

1. Preflop ranges

To compensate for positional disadvantage, out-of-position players must open tighter hand ranges.

Premium starting hands aren't enough, though. Considering GTO poker ranges and principles, you want a good, balanced starting hand range from each position with at least some hands that can make a strong poker hand regardless of the flop texture (low, mid, high, disconnected, etc).

Below is a GTO preflop beginner poker chart for online 6-max play, showing which hand ranges one should open-raise with. Table positions are color-coded (see key below).

NOTE: For GTO play, it's advisable to use a mixed strategy for opening in the small blind, combining open-limps and open-raises for various hands. This cannot be illustrated with the color system used for the chart.

Choosing which hands to play is often a math problem, as discussed below.

Other preflop GTO poker charts include which hands to play after a raise, which to 3bet, etc. Solvers can help you decide which preflop hands to play (call, raise, re-raise, etc.).

2. Pot Odds

Always make +EV decisions that profit you as a poker player. Understanding pot odds (and equity) can help.

Postflop Pot Odds

Let’s say that we have JhTh on a board of 9h8h2s4c (open-ended straight-flush draw).  We have $40 left and $50 in the pot. He has you covered and goes all-in. As calling or folding are our only options, playing GTO involves calculating whether a call is +EV or –EV. (The hand was empty.)

Any remaining heart, Queen, or 7 wins the hand. This means we can improve 15 of 46 unknown cards, or 32.6% of the time.

What if our opponent has a set? The 4h or 2h could give us a flush, but it could also give the villain a boat. If we reduce outs from 15 to 14.5, our equity would be 31.5%.

We must now calculate pot odds.

(bet/(our bet+pot)) = pot odds

= $50 / ($40 + $90)

= $40 / $130

= 30.7%

To make a profitable call, we need at least 30.7% equity. This is a profitable call as we have 31.5% equity (even if villain has a set). Yes, we will lose most of the time, but we will make a small profit in the long run, making a call correct.

Pot odds aren't just for draws, either. If an opponent bets 50% pot, you get 3 to 1 odds on a call, so you must win 25% of the time to be profitable. If your current hand has more than 25% equity against your opponent's perceived range, call.

Preflop Pot Odds

Preflop, you raise to 3bb and the button 3bets to 9bb. You must decide how to act. In situations like these, we can actually use pot odds to assist our decision-making.

This pot is:

(our open+3bet size+small blind+big blind)

(3bb+9bb+0.5bb+1bb)

= 13.5

This means we must call 6bb to win a pot of 13.5bb, which requires 30.7% equity against the 3bettor's range.

Three additional factors must be considered:

Being out of position on our opponent makes it harder to realize our hand's equity, as he can use his position to put us in tough spots. To profitably continue against villain's hand range, we should add 7% to our equity.

Implied Odds / Reverse Implied Odds: The ability to win or lose significantly more post-flop (than pre-flop) based on our remaining stack.

While statistics on 3bet stats can be gained with a large enough sample size (i.e. 8% 3bet stat from button), the numbers don't tell us which 8% of hands villain could be 3betting with. Both polarized and depolarized charts below show 8% of possible hands.

7.4% of hands are depolarized.

Polarized Hand range (7.54%):

Each hand range has different contents. We don't know if he 3bets some hands and calls or folds others.

Using an exploitable strategy can help you play a hand range correctly. The next GTO concept will make things easier.

3. Minimum Defense Frequency:

This concept refers to the % of our range we must continue with (by calling or raising) to avoid being exploited by our opponents. This concept is most often used off-table and is difficult to apply in-game.

These beginner GTO concepts will help your decision-making during a hand, especially against aggressive opponents.

MDF formula:

MDF = POT SIZE/(POT SIZE+BET SIZE)

Here's a poker GTO chart of common bet sizes and minimum defense frequency.

Take the number of hand combos in your starting hand range and use the MDF to determine which hands to continue with. Choose hands with the most playability and equity against your opponent's betting range.

Say you open-raise HJ and BB calls. Qh9h6c flop. Your opponent leads you for a half-pot bet. MDF suggests keeping 67% of our range.

Using the above starting hand chart, we can determine that the HJ opens 254 combos:

We must defend 67% of these hands, or 170 combos, according to MDF. Hands we should keep include:

Flush draws

Open-Ended Straight Draws

Gut-Shot Straight Draws

Overcards

Any Pair or better

So, our flop continuing range could be:

Some highlights:

Fours and fives have little chance of improving on the turn or river.

We only continue with AX hearts (with a flush draw) without a pair or better.

We'll also include 4 AJo combos, all of which have the Ace of hearts, and AcJh, which can block a backdoor nut flush combo.

Let's assume all these hands are called and the turn is blank (2 of spades). Opponent bets full-pot. MDF says we must defend 50% of our flop continuing range, or 85 of 170 combos, to be unexploitable. This strategy includes our best flush draws, straight draws, and made hands.

Here, we keep combining:

Nut flush draws

Pair + flush draws

GS + flush draws

Second Pair, Top Kicker+

One combo of JJ that doesn’t block the flush draw or backdoor flush draw.

On the river, we can fold our missed draws and keep our best made hands. When calling with weaker hands, consider blocker effects and card removal to avoid overcalling and decide which combos to continue.

4. Poker GTO Bet Sizing

To avoid being exploited, balance your bluffs and value bets. Your betting range depends on how much you bet (in relation to the pot). This concept only applies on the river, as draws (bluffs) on the flop and turn still have equity (and are therefore total bluffs).

On the flop, you want a 2:1 bluff-to-value-bet ratio. On the flop, there won't be as many made hands as on the river, and your bluffs will usually contain equity. The turn should have a "bluffing" ratio of 1:1. Use the chart below to determine GTO river bluff frequencies (relative to your bet size):

This chart relates to your opponent's pot odds. If you bet 50% pot, your opponent gets 3:1 odds and must win 25% of the time to call. Poker GTO theory suggests including 25% bluff combinations in your betting range so you're indifferent to your opponent calling or folding.

Best river bluffs don't block hands you want your opponent to have (or not have). For example, betting with missed Ace-high flush draws is often a mistake because you block a missed flush draw you want your opponent to have when bluffing on the river (meaning that it would subsequently be less likely he would have it, if you held two of the flush draw cards). Ace-high usually has some river showdown value.

If you had a 3-flush on the river and wanted to raise, you could bluff raise with AX combos holding the bluff suit Ace. Blocking the nut flush prevents your opponent from using that combo.

5. Bet Sizes and Frequency

GTO beginner strategies aren't just bluffs and value bets. They show how often and how much to bet in certain spots. Top players have benefited greatly from poker solvers, which we'll discuss next.

GTO Poker Software

In recent years, various poker GTO solvers have been released to help beginner, intermediate, and advanced players play balanced/GTO poker in various situations.

PokerSnowie and PioSolver are popular GTO and poker study programs.

While you can't compute players' hand ranges and what hands to bet or check with in real time, studying GTO play strategies with these programs will pay off. It will improve your poker thinking and understanding.

Solvers can help you balance ranges, choose optimal bet sizes, and master cbet frequencies.

GTO Poker Tournament

Late-stage tournaments have shorter stacks than cash games. In order to follow GTO poker guidelines, Nash charts have been created, tweaked, and used for many years (and also when to call, depending on what number of big blinds you have when you find yourself shortstacked).

The charts are for heads-up push/fold. In a multi-player game, the "pusher" chart can only be used if play is folded to you in the small blind. The "caller" chart can only be used if you're in the big blind and assumes a small blind "pusher" (with a much wider range than if a player in another position was open-shoving).

Divide the pusher chart's numbers by 2 to see which hand to use from the Button. Divide the original chart numbers by 4 to find the CO's pushing range. Some of the figures will be impossible to calculate accurately for the CO or positions to the right of the blinds because the chart's highest figure is "20+" big blinds, which is also used for a wide range of hands in the push chart.

Both of the GTO charts below are ideal for heads-up play, but exploitable HU shortstack strategies can lead to more +EV decisions against certain opponents. Following the charts will make your play GTO and unexploitable.

Poker pro Max Silver created the GTO push/fold software SnapShove. (It's accessible online at www.snapshove.com or as iOS or Android apps.)

Players can access GTO shove range examples in the full version. (You can customize the number of big blinds you have, your position, the size of the ante, and many other options.)

In Conclusion

Due to the constantly changing poker landscape, players are always improving their skills. Exploitable strategies often yield higher profit margins than GTO-based approaches, but knowing GTO beginner and advanced concepts can give you an edge for a few reasons.

It creates a solid gameplay base.

Having a baseline makes it easier to exploit certain villains.

You can avoid leveling wars with your opponents by making sound poker decisions based on GTO strategy.

It doesn't require assuming opponents' play styles.

Not results-oriented.

This is just the beginning of GTO and poker theory. Consider investing in the GTO poker solver software listed above to improve your game.