#THATW update

[altfire-archive]

the horror and the wild

an elder scrolls fanfic series

Édouart Langstone and Brandr Steel-Blood aren't exactly on the run. A necromancer expelled from the College of Winterhold and a former bandit leader from the Rift, they've settled down on a farm in Falkreath to try and live a normal life. Unfortunately for them, the wood hides dangers that were waiting for this chance to come out, and their peace and quiet is destined to shatter.

series on ao3 prologue: wretch [3.7k] part one: the horror and the wild [19.1k] part two: like real people do [38.2k]

my thatw tag/old blog version édouart ref/my ed tag brandr ref/my brandr tag

[orphudice]

HWLP. thatw so real tbh .. u have some ocs 2 yes?! can i hear abt them

I DO HABE OCS THANKU FIR ASKING!!!! they are technically for my fangan but im mostly developing their bacjstories rn instead <33

the protagonists name is chikako(ive mentioned her a Lot but ive never rlly gone in detail about her plus shes a lil updated) and basically she comes from a super wealthy and important family of doctors and lawyers and attorneysand judges etc etc !! so theres like a lot of academic pressure on her from her family especially since shes the eldest daughter of the newest generation of her family. chikako ends up also putting a lot of pressure on herself which then translates into her pressuring other people so her arc is kinda just about her learning to seperate herself anf her academic acchievements and learning to be more compassioante towards other and um yeah thats her <33

one of my other ocs is named rei and hes a funky fantasy writer guy !!he got bullied a lot in school due to him being a lot meeker than other people so he spent a lot of time by himself in the school library immersing himself in fantasy worlds that fistracted him enough from reality, and decided he wanted to make his own funky story world !! rei soent a few years working and editing an original book of his and it helped builf his confidence and stuff and etcetra etcetra ^__^hes a lot simpler than most of my ocs but i love him dearly

[creativemessbyvd]

Chapter 7 is up! READ ON AO3!

Summary:

Georgie sat up gasping, wetness running down his cheeks. It was morning, and he was back in Ben’s living room. Mike sat a short distance away, hands help up, talking to him. But the loudness of his heart made it hard to understand. Maybe he was still trapped in the dream, and just to confirm it, his hand clasped his amulet, thankfully around his neck this time. Mike now was demonstrating inhales and exhales exaggeratedly, leading Georgie to copy it, until he could hear normally again and his heart didn’t seem like it would jump and leave him behind.

[laerrynseelie]

so, now that the AMA is out, I thought I’d make something inspired by things said in the AMA as well as the general dear hearts and discord chaos. from the AMA, “we make our own mythology”, and I’ve surrounded this quote with lyrics and jokes and references throughout. explanation of every reference below the cut!! this took me 6 hours I’m dying @joeybateyofficial

(if you want you can grab prints of a slightly updated version of this here! if the art is too difficult to see here’s a gdrive link for it. I also put it on twitter if y’all care to give it some love there too https://twitter.com/misscocochen/status/1254104159902711809)

“we make our own mythology” - something madeleine and joey said in the discord AMA regarding the mythology or folklore that may inspire them. actual quote is creating our own mythology but oh well

your interpretation - response in discord AMA about what the songs are about and if they’re up for interpretation

robot vampire + I DUNNO - wild blue yonder lyrics

13/04/2020 - the discord AMA day

the horror and the wild - title track and title of second album

dead sheep - in the AMA, Joey said that madeleine slaughters a lamb before every gig to warm up

where’s the vodka - lyric from Not Yet/Love Run

glug - reference to Jackie downing vodka during the AMA

pruning shears - pruning shears title/lyrics

blue flower - it’s a forget me not, reference to lyrics in Elsa’s Song

dungeon master / tiefling bard - reference to J&M’s roles in DND, as answered in the AMA

blue furious boy / scarlet scarlet - reference to the characters J&M took on for the album Love Run, more info on the official TAD instagram

that’s what she said - lyrics from fair

THE SPANISH INQUISITION - it’s a long story but it happened in the discord

bottle of wine + glass - reference J&M showing up to the discord drinking wine

HALT! - lyrics of wild blue yonder

so much fucking hair - lyrics of fair

beef and brambles + pie adjacent - reference to discord AMA where joey is answering what dessert TAD would be

chords? no - reference to AMA where joey says they won’t release chords

1953 - reference to AMA where madeleine says they met in 1953

be good to me / no no not I - that unwanted animal lyrics

stag skull - reference to album photoshoot for the horror and the wild, also subtle nod to everyone’s need for TAD tattoos

fairy lights - joey described THATW album as fairy lights and fire in an instagram post

DRINK WATER / NEVER! - as J&M were drunk during the AMA, mods prompted them to drink water after, with “NEVER!” as the response

fae realm - let’s face it, J&M are probably fae, and also something a lot of dear hearts say to new TAD fans (ex. “welcome to the fae realm!”)

croissant - the discord made jokes about joey pronouncing croissant the french way

ice cream + fork - no that’s not a pitchfork I just can’t draw forks reference to an instagram post where joey says that isolation is making him so mad he’s having ice cream with a fork

devils / tumblings / witchlings - multiple things joey called the gang on instagram before settling on dear hearts, thus why they’re crossed out

bitrate - bees taught joey about bitrate before the AMA

shoulder blades - those tender fucking lyrics in fair

SQUINTS - jackie. suspicion.

under the covers - lyrics of wild blue yonder

shower head + letters of lyrics coming out as the “water” - shower day

and if you’re alone, just know you’re not - something joey wrote in an instagram post

SHAKESPEARE - J&M have both worked in shakespeare productions and madeleine said is a likely influence for her songwriting in the AMA

with my yoghurt - lyrics of fair, with hurt underlined because fair, well, hurts

OOH SCARY tumblr - reference to joey saying tumblr scares the shit out of him in the AMA, also reference to s1ep4 of the witcher where jaskier goes “ooh, scary face”

onion - TAD’s songs have layers. like onions.

HIVE MIND - reference to J&M saying we have a hive mind thing going with similar logo designs

mom said it’s my turn with the brain cell - hive mind. plus we’re chaos. also small nod to Dani, the server mom

broom - emoji used by mods in the discord to move us to another channel. also, jackie is the fastest sweeper.

tender - need I say more?

unreasonable - lyric from fair, the blue line represents the gorgeous, gorgeous voice crack

tent with stick figures - “two men in a tent” pruning shears lyrics

BARMY / mental - words joey used to describe the fandom during the AMA when responding to if it’s overwhelming

O of OWN with red lines through it - reference to our deep analysis of the album Love Run, referencing the red strings meme

red boots - reference to Welly Boots, lyrics “scarlet welly boots”

from Oxfam - reference to Pruning Shears, “got my shoes from Oxfam”

GIN - reference to joey and madeleine saying they were going to have gin after the AMA

lion king meme - the meme Dani made about discord being the shadowy place over there

can you feel the luuv? - reference to AMA where joey and madeleine said this multiple times in a row. also subtle lion king reference hence the location

wolf hall - a production J&M did together (the one they met on (?) not sure about that)

unemployed - reference to lyrics in Not Yet/Love Run

feral - because we’re all absolutely feral for TAD

the hexagonal box - I tried to draw the box from the King music video, not sure it worked

fret not - lyrics from the horror and the wild

sacred heart + nyts + and the candle we lit - oh boy this has LAYERS. like an onion. sacred hearts are a traditional tattoo, again a nod to everyone wanting TAD tattoos, including J&M as they mentioned in the AMA. sacred hearts, while referencing a specific religion and event, also have the meaning of divine love and its transforming power, which I think we’ve all experienced from TAD. the heart of the sacred heart is typically more rounded to look like an apple, Big Apple = New York, while “torch song” is its own thing torch, fire, you get it. “and the candle we lit” is lyrics from wild blue yonder, again here because of the fire of the sacred heart

ship with thought bubble lyrics - from Not Yet/Love Run, in a thought bubble because “sing me awake with a song about pirates” so dreams? idk what the logic was either. the lyrics making up the thought bubble are “sing me awake with a song about pirates / and I will try to harmonize / and sip the sunlight from your eyes”

it’s so good - we once spoke about Not Yet/Love Run in the general channel on discord, ended with multiple people only replying with “it’s so good”, and no one used the channel for hours to preserve this chain

slipknot - reference to the AMA where joey says he enjoys slipknot, also a reference to the only sound tardi makes in the recording which is to gasp “slipknot!” at this time

[jasiper]

thsi is the anon from a while ago thatw as sending harrassing messages. im sorry. i was being immature, rude, and dumb. for that im sorry. im being fully honest with this ive felt guilt for the distressing and downright horrible messages. i know you might not forgive me (i dont blame you) but im sorry.

hi anon

gonna be real honest, every day i think about what you said to me and most times i wondered when you were coming back and what you were gonna say next to me. i limited my original posts, stopped updating my fics, tried not to do anything to put a target on my back. what you said did hurt me. a lot. you went out of your way to work around my blocks to bring me down. i still find it hard to write and a part of me thinks i’ll never be able to write again.

but (this comes as a surprise to me), when i read this ask, there wasn’t a doubt in my mind that i wasn’t going to forgive you. you went out of your way to hurt me, but you also went out of your way to apologize. so i do forgive me and i hope this is a learning lesson for you. i hope you don’t feel the need to tear another person down unjustifiably. i hope you decide to spread love and support for creators. and if there ever comes a time where someone (myself included) writes/makes something that you don’t like, i sincerely hope you don’t make them feel how you made me feel. even though you tore me down, i’m not going to hold onto the anger i feel towards you anymore, so i forgive you.

[suzanneshannon]

Where Does Logic Go on Jamstack Sites?

Here’s something I had to get my head wrapped around when I started building Jamstack sites. There are these different stages your site goes through where you can put logic.

Let’s look at a special example so you can see what I mean. Say you’re making a website for a music venue. The most important part of the site is a list of events, some in the past and some upcoming. You want to make sure to label them as such or design that to be very clear. That is date-based logic. How do you do that? Where does that logic live?

There are at least four places to consider when it comes to Jamstack.

Option 1: Write it into the HTML ourselves

Literally sit down and write an HTML file that represents all of the events. We’d look at the date of the event, decide in whether it’s in the past or the future, and write different content for either case. Commit and deploy that file.

<h1>Upcoming Event: Bill's Banjo Night</h1> <h1>Past Event: 70s Classics with Jill</h1>

This would totally work! But the downside is that weu’d have to update that HTML file all the time — once Bill’s Banjo Night is over, we have to open your code editor, change “Upcoming” to “Past” and re-upload the file.

Option 2: Write structured data and do logic at build time

Instead of writing all the HTML by hand, we create a Markdown file to represent each event. Important information like the date and title is in there as structured data. That’s just one option. The point is we have access to this data directly. It could be a headless CMS or something like that as well.

Then we set up a static site generator, like Eleventy, that reads all the Markdown files (or pulls the information down from your CMS) and builds them into HTML files. The neat thing is thatwe can run any logic we want during the build process. Do fancy math, hit APIs, run a spell-check… the sky is the limit.

For our music venue site, we might represent events as Markdown files like this:

--- title: Bill's Banjo Night date: 2020-09-02 --- The event description goes here!

Then, we run a little bit of logic during the build process by writing a template like this:

  <h1>Past Event: </h1>

Now, each time the build process runs, it looks at the date of the event, decides if it’s in the past or the future and produces different HTML based on that information. No more changing HTML by hand!

The problem with this approach is that the date comparison only happens one time, during the build process. The now variable in the example above is going to refer to the date and time the build happens to run. And once we’ve uploaded the HTML files that build produced, those won’t change until we run the build again. This means that once an event at our music venue is over, we’d have to re-run the build to make sure the website reflects that.

Now, we could automate the rebuild so it happens once a day, or heck, even once an hour. That’s literally what the CSS-Tricks conferences site does via Zapier.

The conferences site is deployed daily using a Zapier automation that triggers a Netlify deploy,, ensuring information is current.

But this could rack up build minutes if you’re using a service like Netlify, and there might still be edge cases where someone gets an outdated version of the site.

Option 3: Do logic at the edge

Edge workers are a way of running code at the CDN level whenever a request comes in. They’re not widely available at the time of this writing but, once they are, we could write our date comparison like this:

// THIS DOES NOT WORK import eventsList from "./eventsList.json" function onRequest(request) {   const now = new Date();   eventList.forEach(event => {     if (event.date > now) {       event.upcoming = true;     }   })   const props = {     events: events,   }   request.respondWith(200, render(props), {}) }

The render() function would take our processed list of events and turn it into HTML, perhaps by injecting it into a pre-rendered template. The big promise of edge workers is that they’re extremely fast, so we could run this logic server-side while still enjoying the performance benefits of a CDN.

And because the edge worker runs every time someone requests the website, we can be sure that they’re going to get an up-to-date version of it.

Option 4: Do logic at run time

Finally, we could pass our structured data to the front end directly, for example, in the form of data attributes. Then we write JavaScript that’s going to do whatever logic we need on the user’s device and manipulates the DOM on the fly.

For our music venue site, we might write a template like this:

<h1 data-date=""></h1>

Then, we do our date comparison in JavaScript after the page is loaded:

function processEvents(){   const now = new Date()   events.forEach(event => {     const eventDate = new Date(event.getAttribute('data-date'))     if (eventDate > now){         event.classList.add('upcoming')     } else {         event.classList.add('past')     }   }) }

The now variable reflects the time on the user’s device, so we can be pretty sure the list of events will be up-to-date. Because we’re running this code on the user’s device, we could even get fancy and do things like adjust the way the date is displayed based on the user’s language or timezone.

And unlike the previous points in the lifecycle, run time lasts as long as the user has our website open. So, if we wanted to, we could run processEvents() every few seconds and our list would stay perfectly up-to-date without having to refresh the page. This would probably be unnecessary for our music venue’s website, but if we wanted to display the events on a billboard outside the building, it might just come in handy.

Where will you put the logic?

Although one of the core concepts of Jamstack is that we do as much work as we can at build time and serve static HTML, we still get to decide where to put logic.

Where will you put it?

It really depends on what you’re trying to do. Parts of your site that hardly ever change are totally fine to complete at edit time. When you find yourself changing a piece of information again and again, it’s probably a good time to move that into a CMS and pull it in at build time. Features that are time-sensitive (like the event examples we used here), or that rely on information about the user, probably need to happen further down the lifecycle at the edge or even at runtime.

The post Where Does Logic Go on Jamstack Sites? appeared first on CSS-Tricks.

You can support CSS-Tricks by being an MVP Supporter.

Where Does Logic Go on Jamstack Sites? published first on https://deskbysnafu.tumblr.com/

[suejenleymon]

Upscale New York Asian Escorts Make You Relax And Happy

Exotic New York Asian Escort: Japanese Escorts & Korean Escorts Most Satisfied AgenciesWhat separates our natural relax Asian Escorts girls from other escort agencies out there? The answer to this question is very clear. All New York Asian escorts we presented are elite girls who are exceptional in every way and it isn't just about their looks or their bodies. It is about their brains and everything else.This means that they are the whole package with their own uniqueness, smarts, and other things that make them stand out the most from other escorts. Our female Super Asian Models's Asian escorts are right on target every time! The female Asian escorts that we do offer professional nightlife booking.This means that they are all the Wonderful Things that an Asian escort for men should be. They are caring, sharing, and have their own individual ways about them that make them stand out from a sexy and sexual aspect. They are perfect party girls in every sense of the word that know how to please a man and to please themselves as well. They are waiting to join you for a good time. Because this good time will be something for both of you to remember together for always.More girls coming soon, But we need your Support, its very very important to usWelcome! Our Escort service is No.1 in New York ,New Jersey, Long Island.We are offering best Exotic Asian female Companion service and exotic massage service to your place.We value Your satisfaction and you will enjoy a dreamed exotic NYC Asian escort service whatever it is. Just make a phone call reservation for Your ideal escort, and we will send NYC Asian escort girls to your doors of your hotel in New York. Better Call or Text.About usOur New York Asian Escort girls that work with us come from a wide variety of backgrounds. We can provide you with girls of whatever nationality you prefer. Japanese, Korean, Thai, Singaporean, Filipino, Vietnamese, even Asian-American hot girls. Most of our Asian escorts are under 24 years old, but we can also provide you with more experienced, mature ladies if you like.Most of the beautiful New York Japanese & Korean escorts that work with us are oriental and upscale looking, others are the mature and sweet type. When you call us, just let us know what type girls you like. We will know the right type Escort Models for you. please Call us.Disclaimer**This New York Asian escort Website contains adult material that some may find to be offensive. If you are offended by such material, or if you are under 21 years old, you have the option not to view by going back. By continuing to view the following pages you are making the following legal statement: * I am an adult as defined by the country from where I am accessing the site. * I am connecting from a country where accessing adult sites is legal. Content Updated & Edited by Super Asian EscortsOur New York Asian escorts give you hearty services worthy of your moneyYou will get experience an aura of breathtaking and wonder at the bewitching beauty of our arm candy. You will not only get the best value for your money in terms of our sophisticated GFE services. Our NY Outcall Girls are not only beauteous but also well endowed and possess a bewitchingly voluptuous body features.Our New York Asian Escort Beauties Are The Most Sophisticated And Gorgeous CreaturesIf you are seeking to raise your canal desires to unfathomable levels, let our NYC GFE Asian escorts charm you throughout their stay. Are you looking for a companion to talk to or an attractive beauty to accompany you to an event? Well, look no further because you are in the right place for that matter.  If you prefer are seeking for out-call New York Asian escort services, the cozy feeling our New York Asian escorts provide leave you reminiscing on those moments even after their departure. You would not want to miss out on the hearty yet intellectual conversations that spin you to open up and explore your desires.Be It An Out-Call Escortall of our New York Asian Escort beauties are trained and committed to making you happy and eventually add more flavor to the entire entertainment experience. They leave you wondering how these beautiful creatures can possess all the alluring qualities and at the same time intellectually gifted. They are conversant with the modern day cultures without having to abandon their own culture and lifestyle. They are a rare breed of playmates who take their time in pleasing men and their exceptional ability to pinpoint on the touchy subjects which are important in the creation of a homely feeling. Their Asian super models culture compliments the attractiveness of our ladies. They are further trained and well-mannered. They are just like a sweet girl friend and delivers exemplary escort services. From a tender age, our New York Asian escort candies are trained to submit and treat men with uttermost respect. They treat men as worthy masters. If you need her to submit, you don't even have to lay down the details on how to achieve that. It is a natural scenario for them.Having A Rough Day Or A String Of Stressful Meetings Or Events?Are you jet lagged or just want a fun night out accompanied by a beautiful woman by your side? We have all sorts of New York Asian Escorts beauties at your disposal for adult fun and entertainment. Their ability to entice men and make them feel loved, welcomed and cherished is just incomparable to any other women in the whole world. They will leave you concurring why New York Asian escorts are considered a rare gem to mankind. They are able to easily adapt and sync to your tune in terms of conversation, manners as well as your desires. If you desire a companion to cure your ego when it has been previous bruised; maybe in a business deal gone sour. Our escorts are here on call to help you get through the next day leaving you rejuvenated and ready to tackle the next day with more vigor.Due To The Nature Of Their Upbringing Out Beauties Are Conscious Of Their Diet And LifestyleThey know what to eat and at what time. If your gym partner need comes up, please contact us because are just a phone call away. Ring us up when you need to attend the gym where you get to show off to your business associates, the class of women whom you associate with. They can keep up with your diet and healthy lifestyle because that is what they are accustomed to. Imagine a charming siren that is extremely gifted in beauty sparks off a sense of confidence in them. As it is said that beauty lies in the eyes of the beholder, there is no better way to express beauty if not in the way you view yourself. They leave you hooked in their dream boat and you often find yourself wanting to hang out with them. You are in for a ride when you try our New York Asian escort services.When Seeking For A Candlelit Dinner And Have No Idea Whom To Accompany You?Well, why not try our gorgeous New York Asian Escort ladies. Be it a Thai, Philippine or Japanese chic, they possess a resemblance in their nature. Their soft, vibrant and naturally tanned smooth faces leave you glued to them under dim light during your dinner. The dolled-up beauties fit very well in any aspect of the event as well as fashion. Their petite bodies complement the dinner gowns that perfectly fit their figures, their sparkling eyes consequently lighten up your moods. They color your entire world literally. They fit perfectly in any mood, event or conversation. Their professional experience is put to test every time they entertain a client as they get to interact with people of diversified age groups and settings. Our New York Asian Escorts are more versed in terms of cultural differences and more importantly are able to conquer the language barriers. Most of them are able to communicate effectively in one or two languages with English leading in their direct communication with their clients.About Our Best ServicesJust before you conclude to unwind our day, we are here to help you make it as relaxing and peaceful as possible. Through direct referrals from previous satisfied clients, our New York Asian escorts get to grant you your desires in all areas. Their main duty is to please her vip clients body again, they can make your happiness and comfort their priority. You will experience a series of tasteful events where you will feel her presence and want to have our GFE services anytime you feel the need for companionship without the daunting task of courtship that is expected in a relationship. You will find yourself specifically requesting for a certain beauty because you really enjoyed your moments with her. Their ability to connect to you is smooth as they are introduced to the art of reading out obvious body language. You will find yourself requesting for a previous escort due to the qualities they possess such as discreetness and strong principles. You are in a situation where you want to vent your frustrations without having to visit a counselor, just talk and without the fear of your secrets finding a way into the public.We Are Ready For ThatWe take caution in preserving our clients' integrity because there is no day where what was said in confidence will find a way into the public limelight. Discreetness is a delicate and sensitive matter to us. paying for our New York Asian Escort services, your feeling and security won't compromised while interacting with our escorts. Our clients include top government officials and world leaders either in call or out call, nothing said or witnessed will make way to the public. It is just against our business goals and sabotages our business ambition. Something which we would not want to be a part of so be sure that your adult entertainment needs will surely be catered for to your satisfaction levels. In case you ever find yourself in a medical emergency, have no worries as our charming beauties are capable of handling basic first aid techniques and eventually get you the medical assistance you need. Hit us up for a smooth ride in the adult entertainment. You could start by asking for a recommendation from any of your trusted friends who have had one before. However, if you would like to keep it private, then the next factor should come in handy. The beauty about our New York Asian escort agencies is that you have an opportunity to choose the best among the many that are out there. How do you know the best? Always look out for the reviews from other customers and the ratings issued. This will always be a sure credibility test. Once you have settled on one agency, it is important to find out the payment method beforehand. Have all your questions answered before making that call. Let the agency have all your preferences known, that way, they can place the best New York Asian Escort for you, which means full satisfaction for you!Look For An Array Of Services?Look for an NYC Asian escort agency that has a lot to offer. Ranging from girlfriend experience services, to basic escort services and even Asian massage services among others. Speaking of the Asian Massage Services, it is one of the most relaxing and very erotic service you could get yourself. For newbies, you better put it in your bucket list as soon as possible. Would you like a special New York Asian Escort to share some time with you? If you do please read on further. We are the complete Asian escort agency and we have some of the most beautiful and breathtaking of all female Asian escorts for you to choose from to be your one escort on an out-call basis.Luxury NYC Asian Escort Service of New York City Most Experienced New York Asian Escorts in the City These New York Asian Escort party girls that work with us come from a wide variety of backgrounds. We can provide you with girls of whatever nationality you prefer. Japanese, Korean, Thai, Singaporean, Filipino, Taiwanese, Vietnamese, even Asian-American hot girls. Most of our Asian escorts are under 25 years aged, but we can also provide you with more experienced, mature ladies if you like. Most of the beautiful New York Japanese & Korean escorts that work with us are oriental and upscale looking, others are the mature and sweet type. When you call us, just let us know what type girls you like. We will know the right type Model girls for you. Top Asian Escorts for Couples in NYC? It is also a great idea to set up the boundaries prior proceeding to hiring the escort for couples. Think about the three-way scenarios and discuss it with your partner. If you and your partner think that it won’t work, then it will probably be true. But with the discussion, you both will be able to make an informative decision later. This website (New York Luxury Asian Escort Site ) contains material that may be offensive to some individuals. Beyond this warning page, there are links to material of adult. 1. Under perjury of law, I solemnly affirm that I am at least 18 years of age and have the legal right to posses adults material in my community. 2. I understand the laws and standards set in my community, site, and computer to which I am transporting this material, and am solely responsible for my own actions. By clicking “Enter” you are electronically signing to the fact that you agree with all the statements in this Disclaimer. If you do not agree, please leave the site now. You might have seen many Asian Escort ads in the NYC If you have come across an Asian Escort ads saying that premier NYC escorts are ready to accompany your night with a happy ending, then you cannot go wrong with it. You will be hiring the premier escorts to enhance your wonderful night at NYC.  Whether you’ve seen this ads on Craigslist, Backpage, or any other source, you will attain such nice sensual experience that is hard to forget.These classified ads are a great source to find your attractive women. However, just like many other demanding topics, it can get easier to encounter scams. It is going to cost your time and effort to search around the women there. So, where to find your premier New York Asian escorts? Avoid the classified ads and focus on the official site of Asian escorts. In NYC, you can find many Asian escort services. Some of them offer you cheap service. Don’t get tempted easily by those. Some of them are low standard services. You don’t want to waste your time with that. If you know someone who has ever used the premier NYC escorts’ service before. Covered Full Service Escort Service If you are coming to NYC by yourself, or with your group of friends, you will want to get some companions for the event. You cannot go wrong with the full service escort in NYC. There’s a chance that you come across some online ads out there that offer you full service escort. These agencies come with different kinds of escorts which can suit your preferences and likeness. Why choose us? New York Asian escorts are professional women of high-class who enjoy the pleasures of SWEET and whom can only be found through the right agency or organization. Gentlemen who come to New York by means of themselves and want to explore this superb metropolis inside the corporation of an attractive, funny and communicative Asian girl, have the possibility of resorting to the offering of an unbiased New York Asian escort. This manner, you do not spend your time all on your own, but rather you get the treat to enjoy the organization of beautiful Asian escorts in our escort agency with women who will probably show you a wonderful time. After all, there’s no factor in refusing everything that New York has to provide, particularly if you like having fun. Our agency is dedicated to offering these Asian female escorts by giving you access to them wherever you are in New York. You can ask him for references We cannot overlook the fact about the word “premier” in the service. That means it is a service which is not look anything like the hookers on the street. Indeed, it is much easier to find girls who want to have a sex with you for few dollars. But it does not work the same with the premier escorts’ service.

[topicprinter]

Hi everyone!I recently launched my entry level product which is a quarterly beauty experience. We help women get the makeup look they want for any occasion in just a few clicks. Target market is 30 year olds and mom's who need help with their makeup. The initial challenges I've had are thatWe aren't a beauty box but everyone who comes by our website thinks it is. They are shocked at the price of the experience because they compare it to beauty boxes who are monthly, not quarterly, and we have a ton more value in the experience. The website has been updated to showcase the value of the product more since. Also, we are designed for women who aren't good at their makeup. Most beauty boxes are designed for women who are extremely passionate and talented with makeup. So women who are good at their makeup don't see the value that our target market sees, but women who do need the help don't know our solution is out there.Organic Social media on Instagram isn't converting well. Influencers don't convert at ALL .When we tested decreasing the price we definitely got more bites but people cancelled after they got the first box discount, despite them providing great reviews. When the price was higher, less people quit.So, now I would like us to diversify marketing to Target women who are willing to spend money to have a better, more convenient experience with makeup to get the look they want.My current plan is to launch a sales funnel that starts with a Facebook video advertisement for a free makeup tutorial, and then move them down the funnel to educate them on the value and buy the product. Inspiration is coming from Click Funnels.Other activities I have in mind:increasing SEO by blogging/cross posting in hopes of getting exposure on a media site.posting in mom groups on Facebookhosting a major "launch" party in the springHow should we prioritize these for the greatest impact? Would welcome any feedback or ideas!

[brandingheight22-blog]

Best E-commerce Solution as Per Client’s Need - Branding Height Technologies

Branding Height Technologies is the established name in the market of Information Technology, whether website design or social media marketing We are engaged in serving every kind of digital solutions to our clients. We are being the master in our services, Branding Height Technologies encouraged new occupation fro startup to a well developed business all need the right blend of promotion according to the analysis of market eCommerce is the first step to initiate online business and to enlarge it across globe. With the constant changes happening in the field of technology and digital instruments are getting advanced day by day, ecommerce become a basic need to take a business ahead and we professionally skilled to serve the best ecommerce solution as per client’s need. The Digital world has no limits and such has to be you business and consumer can attain the things anytime anywhere he wants’ commerce is the platform to increase the horizon of your business. Word Press Ecommerce With Word Press powers 25% of the web. It gives wide array of hosting options and allows you, managing and maintaining power-packed websites. Word Press offers, you can get a high-performance, responsive website achieves high search engine rankings. We have a high professional team to understand your requirements and for the development of your website. Branding Height Technologies offers you the ecommerce website that is Handy to use by anyone and Responsive to adapt any screen size. It is Reliable and secure as well as Economic and Cost effective platform. because we guaranteed to give you India’s best Ecommerce Word press website at cheapest price It is Customizable with Platform, search engine and social media friendly. Exclusive profile page for Seller and buyer.

PHP Ecommerce Website

With the dimension to fulfill unique demands of a clients like page design, adaptability. Being a digital solution provider company, what made us different thatWe are punctual and dedicated to our job to serve our clients the fully responsive and competitive e-solution on time and at affordable price Provide you strong and secured mechanism to create a safe and easy to use payment gateway for ecommerce participants to make and receive online payment. Our motive to generate  for comprehensive business frameworks based on new technologies that give you an edge to survive themselves in harsh fiercely competitive environments

Magento Ecommerce Website Development

We excelled in design and developing professional and fully responsive eCommerce website for your business. With the use of Magneto, platform for ecommerce development Branding Height the only one which create advance market features are addressed for multiple vendor ecommerce websites in which you can customize seller details as well. We create a fully competent mobile supportive website.. With the use of innovative marketing strategies we globalize your product and services,  there is nothing which gonna stop to grow

OPEN CART E-commerce website

Cart being the most important feature of an ecommerce website, therefore, we build a cart system for ecommerce application We are the best open cart E-commerce website service, provider. We are proud of our efficient and dedicated teamWe offer you after development services to repair and maintain the instrument as per latest changes and updates in technology With the goal of delivering global creative digital marketing solutions we bring on the fluid dynamics of digital world and it’s rapidly changing character delivering the results to small as well as starts up enterprise. We are present here at Delhi, with our work is speaking for us .Our client testify the quality of our work, you too can be part of our big family. Beat Branding Heights and get ready to be the brand of your own.

[altfire-archive]

might do a midweek update for lrpd on wednesday bc i feel like ch3 is weak and dont want to do a whole week of waiting just to post it on its own

[wcrcleaders-blog]

We Follow Two Lions for Many Kilometres ... | Tusk Photo | World-Wide Photographic Safaris & Tours

We Follow Two Lions for Many Kilometres … | Tusk Photo | World-Wide Photographic Safaris & Tours

Ben’s Update from Ta Sebube Lodge on the on the Kgalagadi Photo Safari:

Let’s start with a photo of the WIld Cat thatwe spotted yesterday …

Knowing that our final game drive would be over in a few hours, we were all ready to leave the Lodge as soon the game reserve allowed. We had a very successful game drive that started with a sighting of two lions. A male and female were together, with the…

[creativemessbyvd]

Chapter 5 is UP! Click here to read it on Ao3!

Chapter summary:

“I told you, Maddie, just give me a few minutes, please.” Bev hadn’t yet turned around and didn’t even notice it was not her friend there with her. Georgie felt bad for scaring her, but they needed to do this now, or their window of opportunity would close.

“Ms. Marsh, I’m sorry to interrupt, but we have something very important to discuss with you.”

Georgie, Ben, Mike and Eddie race to stop a wedding, will they manage to save Bev from marrying Rogan?

[yve-1428-blog]

Diary ♥

September 26 Today we have our summative test in Science. And to the other subjects we just listen to the lectures of our teachers. After our class me and my friend went to our house and eat our lunch , after that we decuded to continue our activity that we didn't finished in school. It took us an hour to finish that math problem (nit good at solving math problems) , but atleast we got the real answers. September 27 An exciting and interesting day.♥ One of our teacher gave us a game which we raelly enjoyed. Our other teachers continued their lectures, and we enjoyed listebing too because she's connecting a story which is matched to our topics. We also experienced mananaging a microscope, but this is not the first time that i hold a microscope because i already do when I am at Junior High School. September 28 Surprising day! ♥♥ We didn't expect that our Aunt will arrive today that's why we were very surprise when we saw her entering our house. We didn't even prepared something for her because we didn't know that she'll arrive. But she said that's "it's alright because she's here to be with us not to eat." She even gave me allowance before I go to school. So I thank her because even though we're not close enough she still knows my name and was updated on me. September 29 On this day at our classroom most of our time was being spent watching movies, our teacher allows us to watch because they are busy, and some of them finished their lessons for this week. September 30 A very tiring day but I enjoyed. I can't resist seeing our house dirty. So I decided to clean up my room first. Then I cleaned our house and washed our clothes. Then went to take care of my little brother at our other house. Then I went home to take some rest, while taking my rest someone messaged me and it's my classmate telling that our project in Personal Development was uploaded and we need to mention friends so they can react on it. October 1 New chapter for this month! :-D October 1st and it is Sunday. Sunday means having a quality time with family and God. With my family we went to atend mass. Then after that, we went home and our grandma asked us to cook "meryenda" (miki and sinuman). Then we asked, "Why miki and sinuman?" Then she replied "Beacuse today is the death anniversary of your grandfather." So we immediately run to the market to buy some ingredients. Then we cooked. At 3:00 p.m I went to Miramar to help continue our project. October 2 1st day of school for this week and month. New lessons and discussions, we had our surprise quiz in Philosiphy. We planned for our project in Komunikasyon. October 3 Accompanying my friend was FUN! ♥ After school me and my friend went to Vigan. We went there because I awe my friend a treat. We ate at McDonald's. While eating we talked about lot of things. Then questioning each other and sharing thoughts. And then she said that before we go gome, we will going first at NBI near Arcade. Then we're there, we waited for almost 20 mins., beacuse my friend needs to checkbif all the info about her was correct. After that the girl tells us thatwe need to pay at seven eleven (7/11) infront and near St. Paul. After paying, they asked us again to go at the station of NBI (Vigan) to get the papers there. After that we find a photoshop so my friend can have a 2x2 pics. because that's one of her requirements. And lastly we went home. October 5 Practice makes perfect! ♥ ;-) (Afternoon) We are asked to attend the pracrice for choir. After practicing the songs for the mass our classmate asked the teacher who is practicing us if He could play the piano for us and play "Tuloy Parin." So then we practiced the song that we will dedicate to our teacher-adviser-mother. October 6 ♥♥♥♥♥ Double Celebration ♥♥♥♥ Today was a big day because we celebrated the First Friday Mass on the morning and the Tacher's day on the afternoon. While waiting for the mass to start we practiced the mass songs again. It almost took us 1 hour practicing then the mass started. After mass we went to the photobooth to have a class picture with our adviser. October 7 ♥ Yey! Weekends! ♥ When i woke up this morning my siblings, and my cousin and my grandma already left the house, because my sister went to apply work, my brother and my cousin went to their classmates house to make their projects and my grandma went to attend Senior Citizens practice for their dance. Well I am the only one remained, so it means I did all the household chores. :-( October 8 ▶BLESSED SUNDAY◀ Today I attended the mass. I went early today because K know that lot of people will attend today because our Bishop is the one who will take incharge for the mass. It is already past 9 but the mass didn't started yet, the start of our mass here is 8. Some of people didn't wait for the Priest to come so they decided to go home. But me and my friend finished the mass even though our Priest taked over because Bishop can't come. After mass me and my friend decided to go at pagempanadahan to eat. And since we missed each other we stayed there for a few hours talking at our new environment now and having fun. October 9 No classes for this day! 100% I took this day as an opportunity to make my assignments and finish my requirements. And on the afternoon I went to Miramar to help finish our project in Science, even if it's raining. Thank God we finished already, just going to past it. October 10 NATIONAL ACHIEVEMENT TEST We took our NAT today. Our surname was alphabetically arranged l, since my surname starts with "U," let's just say that I'm at room six the last classroom. Thank God, our proctor was my teacher when I'm at Junior High School at St. William's Institute. I don't have close friend on that room so I just talked with my former teacher. October 11 Passing Requirements Today we passed our project in Science. We also took our summative test in Physical Education. Actually, I didn't took a review because I forgot that we will have a test, good thing that our teacher in Komunikasyon was not around so me and my classmates took the opportunity to review on her time. Thanks to my classmate who lend me some of her reviewers. October 12 Finding a nice and good plant for our dish garden. On this afternoon me and my classmate went to their house (Sarsaracat) to find a good Bermuda Grass. Then we went to Patong to get our classmate. (short ROADTRIP ♥) and see if they have good plants in their backyard so we can get it. After that we went to our house to continue making our dish garden. October 13 CLASSES ARE SUSPENDED! We have no class today because there was storm which cause signal number 1, here in Ilocos. This whole day I took care of my little brother so my mother can work. October 14 Time for my work again.! It's weekend so I need to clean and do household chores. I washed our clothes and cleaned our house. Then went to take care my little bro. October 15 ♥♥♥FAMILY TIME♥♥♥ Today I didn't go anywhere after the mass because Sunday is a family day for us so all of us stayed home and enjoyed the company of each of us. ~~~~~~~¤♥¤♥¤♥¤~~~~~~~ -Yvette Marie L. Urcia

[eurekakinginc]

"[P] Sigma – Creating a machine learning framework from scratch (Update on high school thesis advice thread)"- Detail: TLDR: Asked this subreddit for advice in deciding on ML topic for high school thesis 2 years ago (see original thread), ended up writing a machine learning framework from (almost) scratch in C#/F#. It can’t do as much as all the others, isn’t as fast or as pretty, but we still think it’s kind of cool. Here it is: our github repo and a short UI demo.ResultsUpfront the current feature set of our framework Sigma, to give you an idea of what the next few paragraphs are about:Input, Output, Dense, Dropout, Recurrent, SoftmaxCE / SquaredDiff cost layersGradient descent, Momentum, Adadelta, Adagrad optimisersHooks for storing / restoring checkpoints, timekeeping, stopping (or doing other things) on certain criteria, computing and reporting runtime metricsEasy addition of new layers with functional automatic differentiationLinear and non-linear networks with arbitrarily connected constructsDistributed multi- and single- CPU and GPU (CUDA) backendsNative graphical interface where parameters can be interacted with and monitored in real-time  1. IntroductionThis is the story of us writing a machine learning framework for our high school thesis, of what we learned and how we went about writing one from scratch. The story starts about 3 years ago: we saw a video of MarI/O, a Super Mario AI that could learn to play Super Mario levels. We thought that was about the coolest thing of all time and wanted to do something at least kind of similar for our high school thesis. 1.1 The Original PlanFast forward, over 2 years ago we asked for help in deciding what kind of machine learning project we could feasibly do for our senior year high school thesis (see original thread). Quite ambitiously, we proposed a time investment of about 1000 hours total (as in 500 hours each over the course of 8 months) – for what project, we didn’t know yet. In that thread, we were generously met with a lot of help, advice ranged from reproducing existing papers to implementing specific things to getting to understand the material and then seeing what peaked our interest. After some consideration, we figured we would implement something along the lines of DeepMinds arcade game AI and then make it more general, figuring that would be easy for some reason. When planning our project in more detail we however quickly realised thatwe had no idea what we were doing andit would be a shame to do all that work from scratch and have it be so arbitrarily specific. 1.2 PivotingBefore doing anything very productive, we had to properly study machine learning. We figured this might take a while and allocated that part of our time to writing the theoretical part of our thesis, which conveniently overlapped. But because around that time we had to hand in an official target definition for our thesis, we set the most generic “goals” we could get away with. For reference, for a concerning number of months our project was officially named “Software framework for diverse machine learning tasks” with an even longer and even less specific subtitle. During further study and first attempts to draft the actual target definition for our project, our plans gradually shifted from a machine learning framework for playing specific types of games pivoted to an “any kind of visual input” learning framework and then finally to an “anything” machine learning framework – because why not, it seemed like an interesting challenge and we were curious to see how far we would get.  2. Research and PlanningAlright, so we've decided to write a machine learning framework. How does one create a machine learning framework? It takes many weeks to get reasonably proficient in just using a given framework, and that with proper guides, video tutorials and forums to ask for help. Creating a machine learning framework is a whole other story, with no 12 step guidelines to follow. For a considerable amount of time, we were at a loss at what we actually needed to implement – constantly encountering new and conflicting terms, definitions and not-so-obvious-“but the actual conclusion is obvious”-articles. After a little over a month we slowly got a very basic grasp of how this whole machine learning thing worked – something with functions that are approximated at certain points in steps typically using differentiation to make some metric go down – still magic, but a bit less so (until we read about CNNs, LSTMs and then GANs, each of which confused the heck out of us for some time). 2.1 Sketching our FrameworkAs soon as we knew a bit about the art of machine learning we got more serious about the writing a new framework part. Because there are no guides for that, we resorted to reading the source code of established frameworks – all to us relevant parts, many times, until it made some sense. In the meantime, we had decided to use C# as our primary language – mostly because we were already very familiar with it and didn’t want to also have to learn a new language, but officially also because there were no other proper neural network frameworks for .NET. Alongside reading the source code of machine learning libraries (mainly Deeplearning4J, Brainstorm and Tensorflow) we sketched out how we wanted our own framework to be used. After some time, we felt like there was some unnecessary confusion in getting to know machine learning frameworks as an outsider and we set out to design our API to avoid that. Note that because our design makes sense to us doesn’t mean that it makes more sense than the existing ones to other people, nor do we recommend everyone wishing to use machine learning to write their own framework, just to spare their own sanity.Our naïve ideas on how a machine learning should look like was clearly inspired by our C#/Java based programming experience, as is evident from the code example we drafted a few weeks in:Sigma sigma = Sigma.Create("minsttest"); GUIMonitor gui = (GUIMonitor) sigma.AddMonitor(new GUIMonitor("Sigma GUI Demo")); gui.AddTabs({"Overview", "Data", "Tests"}); sigma.Prepare(); DataSetSource inputSource = new MultiDataSetSource(new FileSource("mnist.inputs"), new CompressedFileSource(new FileSource("mnist.inputs.tar.gz"), new URLSource("http://....url...../mnist.inputs.targ.gz"))); DataSetSource targetSource = new MultiDataSetSource(new FileSource("mnist.targets"), new CompressedFileSource(new FileSource("mnist.targets.tar.gz"), new URLSource("http://....url...../mnist.targets.targ.gz" [, output: "otherthandefault"]) [, compression: new TarGZUnpacker(), output: "mnist.inputs" , forceUpdate: false])); DataSet data = new DataSet(new ImageRecordReader(inputSource, {28, 28}).Extractor({ALL} => {inputs: {Extractor.BatchSize, 1, 28, 28}}).Preprocess(Normalisor()), new StringRecordReader(targetSource).Extractor({0} => {targets: {Extractor.BatchSize, 1}} [, blockSize: auto/all/1024^3]); Network network = new Network("mynetwork"); network.Architecture = Input(inputShape: {28, 28}) + 2 * FullyConnected(size: 1024) + SoftmaxCE() + Loss(); Trainer trainer = sigma.CreateTrainer("mytrainer"); trainer.SetNetwork(network); trainer.SetInitialiser(new GaussianInitialiser(mean: 0.0, standardDeviation: 0.05)); trainer.SetTrainingDataIterator(MinibatchIterator(batchSize: 50, data["inputs"], data["targets"]); trainer.SetOptimiser(new SGDOptimiser(learningRate: 0.01); trainer.AddActiveHook(EarlyStopper(patience: 3)); trainer.AddActiveHook(StopAfterEpoch(epoch: 2000)); gui.AccentColor["trainer1"] = Colors.DeepOrange; gui.tabs["overview"].AddSubWindow(new LineChartWindow(name: "Error", sources: {"*.Training.Error"}) [, x: 1, y: 0, width: 2, height: 1]); gui.tabs["overview"].AddSubWindow(new LineChartWindow(name: "Accuracy", sources: {"*.Training.Accuracy"})); sigma.Run(); And skipping ahead a bit, it should be noted that the final framework is extremely similar to what we envisioned here: merely changing around a few syntax things and names, the above example from about a year ago can be used 1:1 in our current framework. The jury is still out on whether that’s a sign of really good or really bad design. Also note the python-style kwargs notation for layer constructor arguments, which was soon discarded in favour of something that actually compiles in C#. But back to the timeline. 2.2 The Sigma ArchitectureAfter defining the code examples and sketching out the rough parts we felt a machine learning framework needed, we arrived at this general architecture for “Sigma.Core”, divided into core components (which translate almost 1:1 to namespace in our project):Util: Mostly boring, well, utility stuff, but also registries, a key part of our architecture. Because we wanted to be able to inspect and visualise everything we needed a global way to access things by identifier – a registry. Our registry is essentially a dictionary with a string key which may contain more registries. Nested registries can be resolved using registry resolvers in dot notation with some fancy wildcards and tags in angel brackets (e.g. “network.layers.*.weights”).Data: Datasets, the records that make them up in various formats, the pipeline to load, extract, prepare and cache them from disk, web, or wherever they come from and make them available as “blocks”. These blocks are parts of an extracted dataset, consist of many individual records, and are used to avoid loading all of a potentially very large dataset into memory at once. Also, data iterators, which slices larger blocks from datasets into pieces that are then fed to the model.Architecture: Abstract definitions for machine learning models, consisting of layer “constructs”, which are lightweight placeholder layers defining what a layer will look like before its fully instantiated. These layers may be in any order and connected with however many other layers they would like.Layers: Unfortunately named since we started out with just neural networks, but these are the individual layers of our machine learning networks – they store meta-parameters (e.g. size) and actual trainable parameters (e.g. the actual weights).Math: Everything that has directly to do with math and low-level computations. All the automatic differentiation logic (which is very much required for doing proper machine learning) and everything that modifies our data is processed here in various backends (e.g. distributed CPU / GPU). To support calculating derivatives with respect to anything we opted for an approach with symbolic objects – essentially an object for a number or an array where the actual data was hidden (it can be fetched, but only via copies). These symbolic objects are passed around through a handler which does the actual data modification. This abstraction proved to be useful when implementing CUDA support where, due to the asynchronous execution of the CUDA stream, the raw data could not exposed to the user anyway, at least not without major performance hits (host-device synchronisation is very slow).Training: The largest component with many subcomponents, all revolving around the actual training process. A training process is defined in a “trainer”, which specifies the following:Initialisers, that define how a models parameters are initialised, which can be configured with registry identifiers. For example, trainer.addInitialiser(“layers.*.biases”, new GaussianInitialiser(0.1, 0.0)); would initialise all parameters named “biases” with a Gaussian distribution of 0.1 (mean 0).Modifiers, that would modify parameters at runtime, for example to clip weights to a certain range.Optimisers, that define how a model learns (e.g. gradient descent). Because we mainly considered neural networks we only implemented gradient based optimisers, but the interface theoretically supports any kind of optimisation.Hooks, that “hook” into the training process at certain time steps and can do whatever you want (e.g. update visualisations, store / restore checkpoints, compute and log metrics, do something (e.g. stop) when some criteria are satisfied).Operators, that delegate work to workers which execute it with a certain backend computation handler according to some parameters. Notable is our differentiation of “global” and “local” processing, where global is the most recent global state. This global state is fetched by local workers that then do the actual work, publish their results to the operator which merges it back into the global scope. A global timestep event is only ejected when all local workers have submitted their work for that timestep, enabling more fine control in distributed learning, at least in theory.Sigma: The root namespace that can create Sigma environments and trainers. An environment may contain multiple trainers, which are all run and, if specified, visualised simultaneously (which was supposed to be helpful in hyperparameter search).Monitors: Technically outside of the core project, but still a component. These monitors can be attached to a Sigma environment and can then, well, monitor almost everything about the trainers of that environment using the aforementioned registry entries. Behaviour can be injected using commands, a special form of hooks that are only invoked once. This way monitors can be used almost independently of the core Sigma project and can be pretty much anything, like a graphical interface or a live, locally hosted website.  3. ImplementationAnd that’s what we implemented, step by step. We started out with me mainly working on Sigma.Core and my partner on our visualisation interface, working to a common interface for months until we could finally combine our individual parts and have it miraculously work in a live graphical interface. The specifics of implementation were very interesting and quite challenging to us, but most of the particulars are probably rather dull to read – after all, most of the time things didn’t work and when we fixed something, we moved on to the next something that didn’t. 3.1 Low-level Data and Mathematical ProcessingThe very first thing we did was getting the data “ETL” (extract transform load) pipeline up and running, mainly fetching data from a variety of sources, loading them into a dataset and extracting them as blocks. I then focused on the mathematical processing part – everything that had to do with using math and calculating derivatives in our framework. I based our functional automatic differentiation, aptly named “SigmaDiff”, on an F# library for autodiff named “DiffSharp”, which I modified heavily to support n-dimensional arrays, improve performance significantly, fix a few bugs, support multiple simultaneous non-global backends, variable data types, and some more stuff that I’m forgetting. The specific details of getting that to work aren’t very interesting – a lot of glue code, refactoring and late-night bug-chasing because the backpropagation didn’t work as it should with some specific combination of operations. One memorable bug was that when remapping backend operations to my own OpenBLAS-based backend I forgot that matrix transposition did more than just change its shape – a mistake that cost me weeks in debugging efforts down the line, because things just didn’t work properly with large layers or more than 1 record per minibatch (duh). 3.2 Performance OptimisationsCompared to all the backend work, the “middleware” of layers and optimisers was rather trivial to implement, as there are hundreds of tutorials and papers on how to create certain layers and optimisers, where I only had to map them to our own solution. Really, that part should have taken a few weeks at most, but took that much longer because we only then discovered dozens of bugs and stability issues. Skipping over a lot of uninteresting details here, it should be noted that at this point performance of the framework was quite bad. I’m talking 300ms/iteration of 100 MNIST records with just a few dense layers on a high-end computer bad. This bad performance not only slowed training but also actual development down by quite a lot, hiding a few critical bugs and never letting us test the entire framework in a real-world use case within a reasonable time. You might wonder why we didn’t just fix the performance from the get-go, but we wanted to make the actual training work first so we would have something to show for our thesis. In hindsight not the ideal choice, but it still worked out quite well and otherwise we wouldn’t have been able to demonstrate our project adequately in time for the final presentation. 3.2.1 A Self-Adjusting BufferIt took many months before we finally got around to addressing the performance issues, but there was no single fix in sight, rather a collection of hundreds of small to medium sized improvements. A major issue was the way our SigmaDiff math processor handled operations: for every operation, a copy was created for the resulting data. That added up. The copying was necessary because backwards differentiation requires all intermediate values, so we couldn’t just not copy things. We couldn’t even create all required buffers in a static way ahead of time because there was (and still is) no way to traverse the operations that will be executed – the computation graph is constructed anew every time, and we can’t completely rely on them to remain constant. To introduce a reliable way of buffering anyway, we introduced the concept of sessions: A session was meant to be a set of operations that would be repeated many times. Iterations, essentially. When a session is started we would start storing all created arrays in our own store and when an array of the same dimensions was requested in the next session we could return the one from last session, all without allocating any new memory. If more memory was required than last time, we could still allocate it, if less was used, we could discard it for the next session, rendering this neatly self-adjusting. To not overwrite data that was created within a session but was needed for the next one (e.g. parameters) we added an explicit “limbo” buffer, which was basically just a flag that could be set at runtime for a certain array that marked it as “do not reuse”. 3.2.2 SIMD and Avoiding Intermediate AllocationOther significant performance improvements were adding SIMD instructions (which enables processing of typically 8 values at the same time for CPU-bound arithmetic operations) wherever possible and reducing other memory allocation to a minimum by adding some in-place operations wherever intermediate values weren't strictly needed. For example, copying results when accumulating gradients during backpropoagation on nodes with multiple operands is unnecessary because the intermediate values aren't used. By analysing profilers to death, I eventually got the iteration time for my MNIST sample down to an acceptable 18ms in release configuration (speedup of about 17x). Incidentally, the core was now so fast that our visualiser sometimes crashed because it couldn’t keep up with all the incoming data.3.3 Monitoring with Sigma 3.3.1 The monitoring SystemWhen developing Sigma, we not only focused on the “mathematical” backend but also implemented a feature rich monitoring system which allows any application to be built on top of Sigma (or better said Sigma.Core). Every parameter can be observed, every change hooked, every parameter managed. With this monitoring system, we built a monitor (i.e. application) that can be used to learn Sigma and machine learning in general. 3.3.2 The WPF MonitorUsers should be able to not only use Sigma, but also learn with Sigma. To address this issue, we built a feature-rich application (with WPF) that allows users to interact with Sigma. plot learning graphs, manage parameters and control the AI like controlling a music player. This monitor, as every other component of Sigma, is fully customisable and extensible. All components were designed with reusability in mind, which allows users to build their own complex application on top of the default monitor. But why describe a graphical user interface? See it for yourself, here is the UI (and Sigma) in action. (Example builds of Sigma can be downloaded on GitHub).Learn to learn at the press of a buttonDirect interaction during the learning processSave, restore and share checkpoints3.4 CUDA Support and Finishing TouchesOnly 2 months ago we started finalising and polishing our framework: adding CUDA support, fixing many stability issues and rounding off a few rough spots that annoyed us. The CUDA support part was particularly tricky as I could only use CuBLAS, not CuDNN, because our backend doesn’t, by design, understand individual layers but just raw computation graphs. A problematic side effect of the previously described session-logic was that there was no guarantee when buffers would be freed, as that was the job of the indeterministic GC. To not leak CUDA device memory I added my own bare-bones reference counter to the device memory allocator, which would be updated when buffers were created / finalised, which works surprisingly well. With CuBLAS, many custom optimised kernels and many nights of my time we achieved around 5ms/iteration for the same sample on a single GTX 1080, which we deemed acceptable for our envisioned use cases.  4. ConclusionApproximately 3000 combined hours, tens of thousands of lines of code and many long nights later we are proud to finally present something we deem reasonably usable for what it is: Sigma, a machine learning framework that might help you understand a little bit more about machine learning. As of now, we probably won’t be adding many new features to Sigma, mainly because we’re working on a new project related to it that’s now taking up most of our available time. Even though it lacks a lot of default features (most importantly the host of default layer types other frameworks offer), we’re quite happy with how far we got with our project and hope that it’s an adequate update to our original question 2 years ago. We would be happy if some of you could check it out and give us some feedback. 4.1 The Cost of Creating a Machine Learning FrameworkExcluding time, it’s quite cheap. Honestly, with some solid prior programming experience (so that the low-level programming part doesn't become an issue), the whole thing isn't terribly difficult and is probably something most people could do, given enough time. A lot of time. Overall, we it took us approximately:Some 600 hours of researchSome 2400 hours of development2 tortured souls, preferably sold to the devil in exchange for less bugsWe have long since stopped properly counting, so take these numbers with a grain of salt, but they should be in the right ballpark. 4.2 Final remarksAll in all, an undertaking like this is extremely time intensive. It was very much overkill for a high school thesis from the get-go, and we knew that, but it just kept getting more and more elaborate, essentially taking up all of our available time and then some. It was definitely worth it though, for now we have a solid understanding how things work on a lower level and, most importantly, we can say we’ve actually written a machine learning framework, which grants us additional bragging rights :). Caption by flotothemoon. Posted By: www.eurekaking.com

[creativemessbyvd]

Chapter 3 of The Horror and the Wild, an IT Chapter 2 Fix-it Fic is up!

Chapter Summary:

“Excuse me, but are you Eddie Kaspbrak?”

Eddie shot them both wary looks, and Georgie noticed that the sun was going down and no one else was around, so he guessed Eddie was thinking on the worst-case scenarios of two (apparent) strangers asking for him. Georgie even guessed he might lie to them, so he directed a bright smile towards him to ease the tension. It seemed to work on the shorter man.

“I… yeah, I am. Who’s asking?”

“I’m… my name is Mike, Mike Hanlon. We used to be old friends, back in … back in Derry.” Mike was pained upon the mention of Derry and Eddie’s next reaction.

Eddie’s eyebrows went down upon hearing both Mike’s and Derry’s name. His face went through so many emotions that Georgie felt it was a little funny until his eyes got big and he paled.

“Mike…. from Derry?”

Georgie and Mike start their road trip to locate their fellow Losers. Will the first of them believe them about their impossible tale?

If you are interested on a fix it fic with Georgie coming back to help out the Losers 11 years after his death, then check out my fic!

[creativemessbyvd]

Read Chapter 2 of The Horror and the Wild on Ao3!

Chapter summary:

“It was raining. I- I wanted to go out and play with my brother. But he was starting to get sick so I went alone. He made me a paper boat and it went into a storm drain. And that’s when he- IT- attacked me. He took me away from my family… and then he went after my brother too.”

By this point, Georgie had tears running down his cheeks, as he held his missing poster in one hand and handed it towards Mike, who set down his books and slowly approached Georgie. Taking one look at the poster, and then back towards him, Mike lost color and stumbled, dropping the paper and almost falling from one of the many piles of books around.

Mike Hanlon is the first Loser on Georgie’s list, but will he believe him when he tells him who he is?

If you are interested in a story with lots of Georgie bonding with the Losers and them getting to fight Pennywise much earlier, than check out the Horror and the Wild!