Comparing the best AI sales assistant? An AI sales assistant is software that uses machine learning to help you get more done — it lowers the barrier so anyone can produce professional output. Privacy matters too: check whether your data trains the model and whether a no-log or enterprise tier is available. Whether you are a beginner or a pro, the right AI sales assistant slots into your workflow and pays for itself fast. We tested the leading options and ranked them by quality, value, and ease of use.
1 Second Everyday
1 Second Everyday (1SE) is an application developed by Cesar Kuriyama. The application allows the user to record one second of video every day and then chronologically edits (mashes) them together into a single film. It is compatible with iOS and Android. The idea of the application was developed by Kuriyama's 1 Second Everyday — Age 30 video. The application was launched in January 2013. 1 Second Everyday played a part in the plot of Chef and also became the inspiration for the 2014 short animated clip Feast. == Background == === Kuriyama's video === In February 2011, when Cesar Kuriyama turned 30, after saving money, he quit his job in an advertising firm and took a year off to travel. During this time, he started working on a project he called 1 Second Everyday. As part of the project, every day he recorded one second of video – something that was supposed to help him remember that day. He started the project because he was frustrated with his memory. He planned to stockpile the 365 one-second clips into one film to serve as a memento of his year. While working on the project Kuriyama realized that recording one second every day impacted the decisions he made in a positive way. After a year he made a 365-second clip out of his recordings. The video called 1 Second Everyday – Age 30, went viral. According to Kuriyama, he was initially inspired to take a year off from work by a TED talk given by Stefan Sagmeister called "The Power of Time Off." Kuriyama also delivered a TED talk about 1 Second Everyday in 2012 at TED 2012 in Long Beach California. === Kickstarter campaign === After completing his own video, Kuriyama decided to develop an application that would allow the users to record one second every day and compile their own videos. He developed a prototype of the application and then in 2012, he launched a Kickstarter campaign to raise funds for completing the application. The campaign became one of the most backed app campaigns in the history of Kickstarter. It was backed by 11,281 backers who pledged a total of $56,959 on an initial goal of $20,000. Following the completion of the Kickstarter campaign, he partnered with an application design studio in Brooklyn to develop the application. 1 Second Everyday was released two weeks after the completion of its Kickstarter campaign. == Application == The application was released for iOS on 10 January 2013. An Android-compatible version of the application was developed later. Using it, the user can record the videos in the application or they can select one second portions from their libraries. 1 Second Everyday dates every snippet. The user can also set alarms to remember to record their daily video. In order to compile a video, the user selects the seconds they want and the application creates a compilation video. The user can keep multiple timelines. It also allows users to post directly on social networks. The main interface in 1 Second Everyday is a calendar, which shows the user which days have snippets and which they can still fill in. In the beginning, 1 Second Everyday restricted the recording to one second. However, the developers later released Super Seconds, which allowed users to record an additional half a second video. In 2014, 1 Second Everyday Crowds was launched, which is an area in the application featuring compilations of second clips from different users. == In the media == The Kickstarter campaign of 1 Second Everyday was featured in Entrepreneur's 3 Innovative Tech Startups on Kickstarter Right Now in 2012. The application was featured in The New York Times, The Washington Post, Gawker and other media outlets. By the end of the launch day, it was in Top 10 Free Apps on App Store. It was also selected as the App of the Week on GeekWire in 2013. Several other one-second compilation videos were also posted on the Internet after Kuriyama's video gained media attention. Sam Cornwell, an English photographer documented his son Indigo's growth using a montage of one-second iPhone clips. He shot these clips every single day from the moment of birth right up to the baby's first birthday. According to Cornwell, he was inspired by Kuriyama's project. The video of Cornwell's son gained considerable media attention after it was posted on YouTube. Save the Children also made a video commercial based on a similar format that showed a British girl oblivious of the Syrian war end up being a refugee. 1SE was a finalist for the Fast Company Innovation by Design Award in 2015, but lost to Google Maps. In 2015, Google Android created a gallery, Leap Second 2015, with the help of Droga5 and Kuriyama. The gallery showcased how people around the world enjoyed the one extra second of their lives. Through the 1 Second Everyday app available at Google Play, people were able to submit their extra second, which were then vetted and added to the gallery. The viewers were able to view other celebratory seconds from around the world as well as searching for them using different hashtags.
Computer-automated design
Design Automation usually refers to electronic design automation, or Design Automation which is a Product Configurator. Extending Computer-Aided Design (CAD), automated design and Computer-Automated Design (CAutoD) are more concerned with a broader range of applications, such as automotive engineering, civil engineering, composite material design, control engineering, dynamic system identification and optimization, financial systems, industrial equipment, mechatronic systems, steel construction, structural optimisation, and the invention of novel systems. The concept of CAutoD perhaps first appeared in 1963, in the IBM Journal of Research and Development, where a computer program was written. to search for logic circuits having certain constraints on hardware design to evaluate these logics in terms of their discriminating ability over samples of the character set they are expected to recognize. More recently, traditional CAD simulation is seen to be transformed to CAutoD by biologically-inspired machine learning, including heuristic search techniques such as evolutionary computation, and swarm intelligence algorithms. == Guiding designs by performance improvements == To meet the ever-growing demand of quality and competitiveness, iterative physical prototyping is now often replaced by 'digital prototyping' of a 'good design', which aims to meet multiple objectives such as maximised output, energy efficiency, highest speed and cost-effectiveness. The design problem concerns both finding the best design within a known range (i.e., through 'learning' or 'optimisation') and finding a new and better design beyond the existing ones (i.e., through creation and invention). This is equivalent to a search problem in an almost certainly, multidimensional (multivariate), multi-modal space with a single (or weighted) objective or multiple objectives. == Normalized objective function: cost vs. fitness == Using single-objective CAutoD as an example, if the objective function, either as a cost function J ∈ [ 0 , ∞ ) {\displaystyle J\in [0,\infty )} , or inversely, as a fitness function f ∈ ( 0 , 1 ] {\displaystyle f\in (0,1]} , where f = J 1 + J {\displaystyle f={\tfrac {J}{1+J}}} , is differentiable under practical constraints in the multidimensional space, the design problem may be solved analytically. Finding the parameter sets that result in a zero first-order derivative and that satisfy the second-order derivative conditions would reveal all local optima. Then comparing the values of the performance index of all the local optima, together with those of all boundary parameter sets, would lead to the global optimum, whose corresponding 'parameter' set will thus represent the best design. However, in practice, the optimization usually involves multiple objectives and the matters involving derivatives are a lot more complex. == Dealing with practical objectives == In practice, the objective value may be noisy or even non-numerical, and hence its gradient information may be unreliable or unavailable. This is particularly true when the problem is multi-objective. At present, many designs and refinements are mainly made through a manual trial-and-error process with the help of a CAD simulation package. Usually, such a posteriori learning or adjustments need to be repeated many times until a ‘satisfactory’ or ‘optimal’ design emerges. == Exhaustive search == In theory, this adjustment process can be automated by computerised search, such as exhaustive search. As this is an exponential algorithm, it may not deliver solutions in practice within a limited period of time. == Search in polynomial time == One approach to virtual engineering and automated design is evolutionary computation such as evolutionary algorithms. === Evolutionary algorithms === To reduce the search time, the biologically-inspired evolutionary algorithm (EA) can be used instead, which is a (non-deterministic) polynomial algorithm. The EA based multi-objective "search team" can be interfaced with an existing CAD simulation package in a batch mode. The EA encodes the design parameters (encoding being necessary if some parameters are non-numerical) to refine multiple candidates through parallel and interactive search. In the search process, 'selection' is performed using 'survival of the fittest' a posteriori learning. To obtain the next 'generation' of possible solutions, some parameter values are exchanged between two candidates (by an operation called 'crossover') and new values introduced (by an operation called 'mutation'). This way, the evolutionary technique makes use of past trial information in a similarly intelligent manner to the human designer. The EA based optimal designs can start from the designer's existing design database, or from an initial generation of candidate designs obtained randomly. A number of finely evolved top-performing candidates will represent several automatically optimized digital prototypes. There are websites that demonstrate interactive evolutionary algorithms for design. allows you to evolve 3D objects online and have them 3D printed. allows you to do the same for 2D images.
Constructive cooperative coevolution
The constructive cooperative coevolutionary algorithm (also called C3) is a global optimisation algorithm in artificial intelligence based on the multi-start architecture of the greedy randomized adaptive search procedure (GRASP). It incorporates the existing cooperative coevolutionary algorithm (CC). The considered problem is decomposed into subproblems. These subproblems are optimised separately while exchanging information in order to solve the complete problem. An optimisation algorithm, usually but not necessarily an evolutionary algorithm, is embedded in C3 for optimising those subproblems. The nature of the embedded optimisation algorithm determines whether C3's behaviour is deterministic or stochastic. The C3 optimisation algorithm was originally designed for simulation-based optimisation but it can be used for global optimisation problems in general. Its strength over other optimisation algorithms, specifically cooperative coevolution, is that it is better able to handle non-separable optimisation problems. An improved version was proposed later, called the Improved Constructive Cooperative Coevolutionary Differential Evolution (C3iDE), which removes several limitations with the previous version. A novel element of C3iDE is the advanced initialisation of the subpopulations. C3iDE initially optimises the subpopulations in a partially co-adaptive fashion. During the initial optimisation of a subpopulation, only a subset of the other subcomponents is considered for the co-adaptation. This subset increases stepwise until all subcomponents are considered. This makes C3iDE very effective on large-scale global optimisation problems (up to 1000 dimensions) compared to cooperative coevolutionary algorithm (CC) and Differential evolution. The improved algorithm has then been adapted for multi-objective optimization. == Algorithm == As shown in the pseudo code below, an iteration of C3 exists of two phases. In Phase I, the constructive phase, a feasible solution for the entire problem is constructed in a stepwise manner. Considering a different subproblem in each step. After the final step, all subproblems are considered and a solution for the complete problem has been constructed. This constructed solution is then used as the initial solution in Phase II, the local improvement phase. The CC algorithm is employed to further optimise the constructed solution. A cycle of Phase II includes optimising the subproblems separately while keeping the parameters of the other subproblems fixed to a central blackboard solution. When this is done for each subproblem, the found solution are combined during a "collaboration" step, and the best one among the produced combinations becomes the blackboard solution for the next cycle. In the next cycle, the same is repeated. Phase II, and thereby the current iteration, are terminated when the search of the CC algorithm stagnates and no significantly better solutions are being found. Then, the next iteration is started. At the start of the next iteration, a new feasible solution is constructed, utilising solutions that were found during the Phase I of the previous iteration(s). This constructed solution is then used as the initial solution in Phase II in the same way as in the first iteration. This is repeated until one of the termination criteria for the optimisation is reached, e.g. a maximum number of evaluations. {Sphase1} ← ∅ while termination criteria not satisfied do if {Sphase1} = ∅ then {Sphase1} ← SubOpt(∅, 1) end if while pphase1 not completely constructed do pphase1 ← GetBest({Sphase1}) {Sphase1} ← SubOpt(pphase1, inext subproblem) end while pphase2 ← GetBest({Sphase1}) while not stagnate do {Sphase2} ← ∅ for each subproblem i do {Sphase2} ← SubOpt(pphase2,i) end for {Sphase2} ← Collab({Sphase2}) pphase2 ← GetBest({Sphase2}) end while end while == Multi-objective optimisation == The multi-objective version of the C3 algorithm is a Pareto-based algorithm which uses the same divide-and-conquer strategy as the single-objective C3 optimisation algorithm . The algorithm again starts with the advanced constructive initial optimisations of the subpopulations, considering an increasing subset of subproblems. The subset increases until the entire set of all subproblems is included. During these initial optimisations, the subpopulation of the latest included subproblem is evolved by a multi-objective evolutionary algorithm. For the fitness calculations of the members of the subpopulation, they are combined with a collaborator solution from each of the previously optimised subpopulations. Once all subproblems' subpopulations have been initially optimised, the multi-objective C3 optimisation algorithm continues to optimise each subproblem in a round-robin fashion, but now collaborator solutions from all other subproblems' subspopulations are combined with the member of the subpopulation that is being evaluated. The collaborator solution is selected randomly from the solutions that make up the Pareto-optimal front of the subpopulation. The fitness assignment to the collaborator solutions is done in an optimistic fashion (i.e. an "old" fitness value is replaced when the new one is better). == Applications == The constructive cooperative coevolution algorithm has been applied to different types of problems, e.g. a set of standard benchmark functions, optimisation of sheet metal press lines and interacting production stations. The C3 algorithm has been embedded with, amongst others, the differential evolution algorithm and the particle swarm optimiser for the subproblem optimisations.
Generative engine optimization
Generative engine optimization (GEO) is one of the names given to the practice of structuring digital content and managing online presence to improve visibility in responses generated by generative artificial intelligence (AI) systems. The practice influences the way large language models (LLMs) retrieve, summarize, and present information in response to user queries. Related terms include answer engine optimization (AEO) and artificial intelligence optimization (AIO). The concept of GEO first appeared in response to generative AI technologies being integrated into mainstream search and information retrieval systems. Tools are used to monitor how websites and brands are cited, referenced, or incorporated into responses produced by large language models. == Terminology == Several overlapping terms describe related practices, and usage varies across practitioners, vendors, and publications. No consensus definition distinguishing these terms had been established in the academic literature as of early 2026, and the terms are frequently used interchangeably in trade and practitioner contexts. Other terms for the same concept include answer engine optimization (AEO), large language model optimization (LLMO), artificial intelligence optimization (AIO), and AI SEO. In 2026, Google released documentation entitled "Optimizing your website for generative AI features on Google Search." According to this documentation, "optimizing for generative AI search is optimizing for the search experience, and thus still SEO.” This position had previously been shared at conferences, with 2026 being the first time Google released official documentation stating it. == Factors influencing generative engine optimization == By early 2026, the focus of GEO practitioners shifted from simple keyword placement to "semantic relevance", a metric driven by the integration of advertising into conversational AI. OpenAI and Google began monetizing AI search results, which is not currently considered an aspect of generative engine optimization but is adjacent.
Learning to rank
Learning to rank (LTR) or machine-learned ranking (MLR) is the application of machine learning, often supervised, semi-supervised or reinforcement learning, in the construction of ranking models for information retrieval and recommender systems. Training data may, for example, consist of lists of items with some partial order specified between items in each list. This order is typically induced by giving a numerical or ordinal score or a binary judgment (e.g. "relevant" or "not relevant") for each item. The goal of constructing the ranking model is to rank new, unseen lists in a similar way to rankings in the training data. == Applications == === In information retrieval === Ranking is a central part of many information retrieval problems, such as document retrieval, collaborative filtering, sentiment analysis, and online advertising. A possible architecture of a machine-learned search engine is shown in the accompanying figure. Training data consists of queries and documents matching them together with the relevance degree of each match. It may be prepared manually by human assessors (or raters, as Google calls them), who check results for some queries and determine relevance of each result. It is not feasible to check the relevance of all documents, and so typically a technique called pooling is used — only the top few documents, retrieved by some existing ranking models are checked. This technique may introduce selection bias. Alternatively, training data may be derived automatically by analyzing clickthrough logs (i.e. search results which got clicks from users), query chains, or such search engines' features as Google's (since-replaced) SearchWiki. Clickthrough logs can be biased by the tendency of users to click on the top search results on the assumption that they are already well-ranked. Training data is used by a learning algorithm to produce a ranking model which computes the relevance of documents for actual queries. Typically, users expect a search query to complete in a short time (such as a few hundred milliseconds for web search), which makes it impossible to evaluate a complex ranking model on each document in the corpus, and so a two-phase scheme is used. First, a small number of potentially relevant documents are identified using simpler retrieval models which permit fast query evaluation, such as the vector space model, Boolean model, weighted AND, or BM25. This phase is called top- k {\displaystyle k} document retrieval and many heuristics were proposed in the literature to accelerate it, such as using a document's static quality score and tiered indexes. In the second phase, a more accurate but computationally expensive machine-learned model is used to re-rank these documents. === In other areas === Learning to rank algorithms have been applied in areas other than information retrieval: In machine translation for ranking a set of hypothesized translations; In computational biology for ranking candidate 3-D structures in protein structure prediction problems; In recommender systems for identifying a ranked list of related news articles to recommend to a user after he or she has read a current news article. == Feature vectors == For the convenience of MLR algorithms, query-document pairs are usually represented by numerical vectors, which are called feature vectors. Such an approach is sometimes called bag of features and is analogous to the bag of words model and vector space model used in information retrieval for representation of documents. Components of such vectors are called features, factors or ranking signals. They may be divided into three groups (features from document retrieval are shown as examples): Query-independent or static features — those features, which depend only on the document, but not on the query. For example, PageRank or document's length. Such features can be precomputed in off-line mode during indexing. They may be used to compute document's static quality score (or static rank), which is often used to speed up search query evaluation. Query-dependent or dynamic features — those features, which depend both on the contents of the document and the query, such as TF-IDF score or other non-machine-learned ranking functions. Query-level features or query features, which depend only on the query. For example, the number of words in a query. Some examples of features, which were used in the well-known LETOR dataset: TF, TF-IDF, BM25, and language modeling scores of document's zones (title, body, anchors text, URL) for a given query; Lengths and IDF sums of document's zones; Document's PageRank, HITS ranks and their variants. Selecting and designing good features is an important area in machine learning, which is called feature engineering. == Evaluation measures == There are several measures (metrics) which are commonly used to judge how well an algorithm is doing on training data and to compare the performance of different MLR algorithms. Often a learning-to-rank problem is reformulated as an optimization problem with respect to one of these metrics. Examples of ranking quality measures: Mean average precision (MAP); DCG and NDCG; Precision@n, NDCG@n, where "@n" denotes that the metrics are evaluated only on top n documents; Mean reciprocal rank; Kendall's tau; Spearman's rho. DCG and its normalized variant NDCG are usually preferred in academic research when multiple levels of relevance are used. Other metrics such as MAP, MRR and precision, are defined only for binary judgments. Recently, there have been proposed several new evaluation metrics which claim to model user's satisfaction with search results better than the DCG metric: Expected reciprocal rank (ERR); Yandex's pfound. Both of these metrics are based on the assumption that the user is more likely to stop looking at search results after examining a more relevant document, than after a less relevant document. == Approaches == Learning to Rank approaches are often categorized using one of three approaches: pointwise (where individual documents are ranked), pairwise (where pairs of documents are ranked into a relative order), and listwise (where an entire list of documents are ordered). Tie-Yan Liu of Microsoft Research Asia has analyzed existing algorithms for learning to rank problems in his book Learning to Rank for Information Retrieval. He categorized them into three groups by their input spaces, output spaces, hypothesis spaces (the core function of the model) and loss functions: the pointwise, pairwise, and listwise approach. In practice, listwise approaches often outperform pairwise approaches and pointwise approaches. This statement was further supported by a large scale experiment on the performance of different learning-to-rank methods on a large collection of benchmark data sets. In this section, without further notice, x {\displaystyle x} denotes an object to be evaluated, for example, a document or an image, f ( x ) {\displaystyle f(x)} denotes a single-value hypothesis, h ( ⋅ ) {\displaystyle h(\cdot )} denotes a bi-variate or multi-variate function and L ( ⋅ ) {\displaystyle L(\cdot )} denotes the loss function. === Pointwise approach === In this case, it is assumed that each query-document pair in the training data has a numerical or ordinal score. Then the learning-to-rank problem can be approximated by a regression problem — given a single query-document pair, predict its score. Formally speaking, the pointwise approach aims at learning a function f ( x ) {\displaystyle f(x)} predicting the real-value or ordinal score of a document x {\displaystyle x} using the loss function L ( f ; x j , y j ) {\displaystyle L(f;x_{j},y_{j})} . A number of existing supervised machine learning algorithms can be readily used for this purpose. Ordinal regression and classification algorithms can also be used in pointwise approach when they are used to predict the score of a single query-document pair, and it takes a small, finite number of values. === Pairwise approach === In this case, the learning-to-rank problem is approximated by a classification problem — learning a binary classifier h ( x u , x v ) {\displaystyle h(x_{u},x_{v})} that can tell which document is better in a given pair of documents. The classifier shall take two documents as its input and the goal is to minimize a loss function L ( h ; x u , x v , y u , v ) {\displaystyle L(h;x_{u},x_{v},y_{u,v})} . The loss function typically reflects the number and magnitude of inversions in the induced ranking. In many cases, the binary classifier h ( x u , x v ) {\displaystyle h(x_{u},x_{v})} is implemented with a scoring function f ( x ) {\displaystyle f(x)} . As an example, RankNet adapts a probability model and defines h ( x u , x v ) {\displaystyle h(x_{u},x_{v})} as the estimated probability of the document x u {\displaystyle x_{u}} has higher quality than x v {\displaystyle x_{v}} : P u , v ( f ) = CDF ( f ( x u ) − f ( x v ) ) , {\displaystyle P_{u,v}(f)={\text{CDF}
Stable Diffusion
Stable Diffusion is a deep learning, text-to-image model released in 2022 based on diffusion techniques. The generative artificial intelligence technology is the premier product of Stability AI and is considered to be a part of the ongoing AI boom. It is primarily used to generate detailed images conditioned on text descriptions, though it can also be applied to other tasks such as inpainting, outpainting, and generating image-to-image translations guided by a text prompt. Its development involved researchers from the CompVis Group at LMU Munich and Runway with a computational donation from Stability and training data from non-profit organizations. Stable Diffusion is a latent diffusion model, a kind of deep generative artificial neural network. Its code and model weights have been released publicly, and an optimized version can run on most consumer hardware equipped with a modest GPU with as little as 2.4 GB VRAM. This marked a departure from previous proprietary text-to-image models such as DALL-E and Midjourney which were accessible only via cloud services. == Development == Stable Diffusion originated from a project called Latent Diffusion, developed in Germany by researchers at LMU Munich in Munich and Heidelberg University. Four of the original 5 authors (Robin Rombach, Andreas Blattmann, Patrick Esser and Dominik Lorenz) later joined Stability AI and released subsequent versions of Stable Diffusion. The technical license for the model was released by the CompVis group at LMU Munich. Development was led by Patrick Esser of Runway and Robin Rombach of CompVis, who were among the researchers who had earlier invented the latent diffusion model architecture used by Stable Diffusion. Stability AI also credited EleutherAI and LAION (a German nonprofit which assembled the dataset on which Stable Diffusion was trained) as supporters of the project. == Technology == === Architecture === Diffusion models, introduced in 2015, are trained with the objective of removing successive applications of Gaussian noise on training images, which can be thought of as a sequence of denoising autoencoders. The name diffusion is from the thermodynamic diffusion, since they were first developed with inspiration from thermodynamics. Models in Stable Diffusion series before SD 3 all used a variant of diffusion models, called latent diffusion model (LDM), developed in 2021 by the CompVis (Computer Vision & Learning) group at LMU Munich. Stable Diffusion consists of 3 parts: the variational autoencoder (VAE), U-Net, and an optional text encoder. The VAE encoder compresses the image from pixel space to a smaller dimensional latent space, capturing a more fundamental semantic meaning of the image. Gaussian noise is iteratively applied to the compressed latent representation during forward diffusion. The U-Net block, composed of a ResNet backbone, denoises the output from forward diffusion backwards to obtain a latent representation. Finally, the VAE decoder generates the final image by converting the representation back into pixel space. The denoising step can be flexibly conditioned on a string of text, an image, or another modality. The encoded conditioning data is exposed to denoising U-Nets via a cross-attention mechanism. For conditioning on text, the fixed, pretrained CLIP ViT-L/14 text encoder is used to transform text prompts to an embedding space. Researchers point to increased computational efficiency for training and generation as an advantage of LDMs. With 860 million parameters in the U-Net and 123 million in the text encoder, Stable Diffusion is considered relatively lightweight by 2022 standards, and unlike other diffusion models, it can run on consumer GPUs, and even CPU-only if using the OpenVINO version of Stable Diffusion. ==== SD XL ==== The XL version uses the same LDM architecture as previous versions, except larger: larger UNet backbone, larger cross-attention context, two text encoders instead of one, and trained on multiple aspect ratios (not just the square aspect ratio like previous versions). The SD XL Refiner, released at the same time, has the same architecture as SD XL, but it was trained for adding fine details to preexisting images via text-conditional img2img. ==== SD 3.0 ==== The 3.0 version completely changes the backbone. Not a UNet, but a Rectified Flow Transformer, which implements the rectified flow method with a Transformer. The Transformer architecture used for SD 3.0 has three "tracks", for original text encoding, transformed text encoding, and image encoding (in latent space). The transformed text encoding and image encoding are mixed during each transformer block. The architecture is named "multimodal diffusion transformer (MMDiT), where the "multimodal" means that it mixes text and image encodings inside its operations. This differs from previous versions of DiT, where the text encoding affects the image encoding, but not vice versa. === Training data === Stable Diffusion was trained on pairs of images and captions taken from LAION-5B, a publicly available dataset derived from Common Crawl data scraped from the web, where 5 billion image-text pairs were classified based on language and filtered into separate datasets by resolution, a predicted likelihood of containing a watermark, and predicted "aesthetic" score (e.g. subjective visual quality). The dataset was created by LAION, a German non-profit which receives funding from Stability AI. The Stable Diffusion model was trained on three subsets of LAION-5B: laion2B-en, laion-high-resolution, and laion-aesthetics v2 5+. A third-party analysis of the model's training data identified that out of a smaller subset of 12 million images taken from the original wider dataset used, approximately 47% of the sample size of images came from 100 different domains, with Pinterest taking up 8.5% of the subset, followed by websites such as WordPress, Blogspot, Flickr, DeviantArt and Wikimedia Commons. An investigation by Bayerischer Rundfunk showed that LAION's datasets, hosted on Hugging Face, contain large amounts of private and sensitive data. === Training procedures === The model was initially trained on the laion2B-en and laion-high-resolution subsets, with the last few rounds of training done on LAION-Aesthetics v2 5+, a subset of 600 million captioned images which the LAION-Aesthetics Predictor V2 predicted that humans would, on average, give a score of at least 5 out of 10 when asked to rate how much they liked them. The LAION-Aesthetics v2 5+ subset also excluded low-resolution images and images which LAION-5B-WatermarkDetection identified as carrying a watermark with greater than 80% probability. Final rounds of training additionally dropped 10% of text conditioning to improve Classifier-Free Diffusion Guidance. The model was trained using 256 Nvidia A100 GPUs on Amazon Web Services for a total of 150,000 GPU-hours, at a cost of $600,000. === Limitations === Stable Diffusion has issues with degradation and inaccuracies in certain scenarios. Initial releases of the model were trained on a dataset that consists of 512×512 resolution images, meaning that the quality of generated images noticeably degrades when user specifications deviate from its "expected" 512×512 resolution; the version 2.0 update of the Stable Diffusion model later introduced the ability to natively generate images at 768×768 resolution. Another challenge is in generating human limbs due to poor data quality of limbs in the LAION database. The model is insufficiently trained to replicate human limbs and faces due to the lack of representative features in the database, and prompting the model to generate images of such type can confound the model. In addition to human limbs, Stable Diffusion is unable to generate legible ambigrams and some other forms of text and typography. Stable Diffusion XL (SDXL) version 1.0, released in July 2023, introduced native 1024x1024 resolution and improved generation for limbs and text. Accessibility for individual developers can also be a problem. In order to customize the model for new use cases that are not included in the dataset, such as generating anime characters ("waifu diffusion"), new data and further training are required. Fine-tuned adaptations of Stable Diffusion created through additional retraining have been used for a variety of different use-cases, from medical imaging to algorithmically generated music. However, this fine-tuning process is sensitive to the quality of new data; low resolution images or different resolutions from the original data can not only fail to learn the new task but degrade the overall performance of the model. Even when the model is additionally trained on high quality images, it is difficult for individuals to run models in consumer electronics. For example, the training process for waifu-diffusion requires a minimum 30 GB of VRAM, which exceeds the usual resource provided in such consumer GPUs as Nvidia's GeForce 30 series, w