#placeholder sigma tag

[cousticks]

i fucked around and doodled a sigma because his design is cool. no i did not pay much attention to his canon outfit. do not ask me what weird galaxy box he is sitting on or whats up with the glowy stars because i do not know. i also dk anything about him yet i'm not there in The Media yet. no i do not know how to draw. yes this is just a sketch. no further questions thank you.

[bungobble-my-balls]

☆Introduction Post!☆

Despite my username, I'd rather be referred to as 'Sevsix' on this blog (like my page title says) over 'Bungobble-My-Balls'. That was a placeholder that I find too funny to change and am also too attached to.

What can you find?:

- Mostly reblogs

- any post that I make goes under the tag 'bungobble my post'

- A lot of Sigmatsu <3 (Sigma x Atsushi for anyone who doesn't know the pairing)

- my thoughts for every new chapter

- me playing with manga png cutouts like dolls

- content based around my fanfic 'Bugs in the Basement', which is Kyoukyuu centric and will eventually bring in the Kyouka-Kenji-Kyuusaku friendship dynamic.

- Dollhouse siblings content (the sibling relationship between Nikolai, Lucy and Q that I completely made up)

- The occasional analysis post, headcanon, theory, etc.

- Other random bsd related thoughts that aren't that important. These usually go under the tag 'ramblings'.

Other Accounts:

My Main - Crypteur (I don't use it really, but since this is a side blog that's the blog that will show up if I've liked a post or followed someone)

Gimmick side blog - atsushi-judging-you (I just used it to upload Atsushi expressions because I think they're funny)

My main art blog - everythingbrainsoup

My AO3: Cr7pt1c

[purplekoop]

any thoughts on hero 42? My theory is that Harriet is one of the "reject" Horizon animals, one who is a toad that was spliced with other reptilian and amphibian DNA, namely turtles, hence the shell we can see on her. Idk if she'd be good, evil, or neutral, but considering every tank since Orisa has been villain or Junker, I'm hoping she bucks that trend.https://www.reddit.com/r/Overwatch/comments/1e2f5ux/theory_new_overwatch_tank_is_a_genetically/

https://www.reddit.com/r/Overwatch/comments/17ythsp/atlantic_arcology_a_future_underwater_map/?share_id=39z04D8QgOF93Ov-7xr0U&utm_content=2&utm_medium=android_app&utm_name=androidcss&utm_source=share&utm_term=1 put this cuz of the Galapagos section

alright this sounded so utterly deranged BEFORE I checked the links. I was internally going "how much did I miss in the rodent's story", but these posts have some fascinating insight I'd encourage everyone check if nothing else. Gonna post the links cleanly here for the sake of usability:

For personal reasons I'd of course ADORE if this character actually is a giant buff spiky turtle mutant lady. if this sentiment confuses anyone, this is an open invite to the "#my art" tag on this account. But by default I'm not getting my hopes up because come on, that's TOO perfect for me, I shouldn't be allowed to get that happy. Then again I was WILDLY surprised by Junker Queen's in-game design, so you never know.

I've said before I don't like to base expectations on arbitrary categories (namely the current insistence on more villain supports but not villain tanks (even though we were clamoring for those before Sigma and then as far as I'm concerned the rest following were all warranted as catch-up)), though if I can have my own biases then I'd also prefer not having another Junker, but also I'm not wild on the idea of another Horizon animal guy, especially when I really don't like playing as the two we currently have (and playing in a lobby with a Wrecking Ball whatsoever). The idea of them being a Moira experiment both makes more sense and is a little more appealing, but like half of Talon seems to be Moira experiments already, so I don't know how upping the Level of Freak they are would really be shocking at this point.

I'd be astonished if the mutant-looking creature with glowing green spikes wasn't called "Hazmat" at at least some point in development, though based on the hero select in the Venture gameplay showcase, their internal codename seemingly is somewhere alphabetically between Sigma and Winston. Not necessarily their final name though, as evidenced by Juno in the same teaser being between Moira and Zenyatta (which works for "Space Ranger") as opposed to between Illari and Kiriko like they are in their release state. I speculated before their working name might be "Spike" or "Toxic", which could've been placeholders before "Haz" was used for the Ashe interaction.

I don't entirely want to rule out this possibly of this hero being someone named "Harriet Oris", but while that reddit theory points out the Horizon animals have dated english first names, it neglects to point out they don't have last names either. Winston is just Winston (named after the Horizon scientist Harold Winston) and Hammond is just Hammond. If "Hazmat" and Harriet Oris are the same, then that'd be weird if they were also a Horizon animal with a last name for some reason. The Waitron does also say "It has been a very long time!" to them in the unused dialogue, which could lend credence to the theory of them being a former Talon member and possible escaped Moira experiment. The same account points out similar dialogue for Mauga, and considering the Waitrons are in Circuit Royal (a Talon base of operations run by Maximilien), so maybe there's a parallel there.

Regarding the streamer mode name easter eggs potentially meaning this is a frog (or more accurately a toad, a buff one at that), I don't know how much stock I'd put in that part of the theory specifically. It's weird how few of the names aren't a reference to something, but the only ones the wiki doesn't attribute to anything are "BEEFToad", "BioHazard", "Ch00Ch00", "EZTarget", "GarlicBread", and "SillyGoose". I think most of these are pretty clearly jokes, maybe "Choo Choo" being a reference specifically to Reinhardt's one skin or the Busan train, but those first two are definitely more conspicuous. "BioHazard" is the least joke-like of the bunch, but could also be a reference to other things like the toxic waste area in New Junk City or Mauga's weird canisters. "BEEFToad" feels like a joke, but not as obvious as SillyGoose or GarlicBread. I dunno, to me it just feels like trying to stick on another data point that doesn't necessarily add anything. Making them a frog with turtle DNA feels like an arbitrary extra step and it feels like a weird thing to tease so specifically like this. The turtle thing in the other post feels more like a deliberate nod, they already look more turtle-ish and it feels like such a "nudge nudge wink wink future hero reference here" bit to emphasize.

This all does give me some more enthusiasm for a character that previously felt like a dead end to speculate on, and while I don't wanna get my hopes up, there's definitely something cool here to consider. Gonna be a bit of a wait before they get a proper reveal though, and not sure how it'll be done without a Blizzcon this year. We probably will still at minimum get a reveal and playtest around the same time we did last year for Mauga, so somewhere in November before the proper release in season 14 this December.

[siva3155]

300+ TOP SEO Objective Questions and Answers

SEO Multiple Choice Questions :-

1.  If a website’s search engine saturation with respect to a particular search engine is 20%, what does it mean? a. 20% of the web pages of the website have been indexed by the search engine b. Only 20% of the pages of the website will be indexed by the search engine c. 20% of the websites pages will never be indexed d. The website ranks in the first 20% of all websites indexed by the search engine for its most important search terms Ans: a 2.  10 people do a web search. In response, they see links to a variety of web pages. Three of the 10 people choose one particular link. That link then has a _______ click through rate. a. less than 30% b. 30 percent c. more than 30% Ans:  b 3.  Which of the following factors have an impact on the Google Page Rank? a. The total number of inbound links to a page of a web site b. The subject matter of the site providing the inbound link to a page of a web site c. The text used to describe the inbound link to a page of a web site d. The number of outbound links on the page that contains the inbound link to a page of a web site Ans: a 4.  What does the 301 server response code signify? a. Not Modified b. Moved Permanently c. syntax error in the request d. Payment is required e. The request must be authorized before it can take place Ans: b 5.  If you enter ‘Help site:www.iqsanswers.com’ in the Google search box, what will Google search for? a. It will open up the Google help pages applicable to www.iqsanswers.com b. It will find pages about help within www.iqsanswers.com c. It will only find page titles about help within www.iqsanswers.com d. It will direct you to the request page for re-indexing of www.iqsanswers.com Ans: b 6.  What is Anchor Text? a. It is the main body of text on a particular web page b. It is the text within the left or top panel of a web page c. It is the visible text that is hyper linked to another page d. It is the most prominent text on the page that the search engines use to assign a title to the page Ans: c 7.  Which of the following free tools/websites could help you identify which city in the world has the largest search for the keyword – “six sigma”? a. Yahoo Search Term Suggestion Tool b. Alexa c. Google Traffic Estimator d. Google Trends e. WordTracker Ans: d 8.  What term is commonly used to describe the shuffling of positions in search engine results in between major updates? a. Waves b. Flux c. Shuffling d. Swaying Ans: b 9.  Are RSS/Atom feeds returned in Google’s search results? a. Yes b. No Ans: b 10.  Which of the following statements regarding website content are correct? a. If you have two versions of a document on your website, Google recommends that you only allow the indexing of the better version b. Linking to a page inconsistently does not affect the way Google views the page/s. c. Syndicating your content could lead to Google viewing the material as duplicate d. Placeholders for pages which do not have content are never viewed as duplicate content by Google Ans: a

SEO MCQs 11.  What does the term Keyword Prominence refer to? a. It refers to the fact that the importance of choosing high traffic keywords leads to the best return on investment b. It refers to the importance attached to getting the right keyword density c. It refers to the fact that the keywords placed in important parts of a webpage are given priority by the search engines d. It refers to the fact that the keywords in bold font are given priority by the search engines Ans: c 12.  What is the term for Optimization strategies that are in an unknown area of reputability/validity? a. Red hat techniques b. Silver hat techniques c. Grey hat techniques d. Shady hat techniques Ans: c 13.  Which of the following statements is correct with regard to natural links? a. They are two way links (reciprocal links) b. They are from authority websites c. They are voluntary in nature d. They are from .edu or .gov extension websites Ans: c 14.  Which of the following can be termed as good keyword selection and placement strategies? a. Targeting synonyms of the main keyword b. Targeting the highest searched keywords only c. Copying competitor keywords d. Optimizing five or more keywords per page Ans: d 15.  What does the 302 server response code signify? a. It signifies conflict, too many people wanted the same file at the same time b. The page has been permanently removed c. The method you are using to access the file is not allowed d. The page has temporarily moved e. What you requested is just too big to process Ans: d 16.  Which of the following statements about FFA pages are true? a. They are greatly beneficial to SEO b. They are also called link farms c. They are paid listings d. They contain numerous inbound links Ans: b 17.  What is the name of the search engine technology due to which a query for the word ‘actor’ will also show search results for related words such as actress, acting or act? a. Spreading b. Dilating c. RSD (realtime synonym detection) d. Stemming e. Branching Ans: c 18.  What will the following robots.txt file do? User-agent:Googlebot Disallow:/*? User-agent:Scooter Disallow: a. It will allow Google to crawl any of the dynamically generated pages. It will also allow thealtavista scooter bot to access every page b. It will disallow Google from crawling any of the dynamically generated pages. It will also disallow the altavista scooter bot from accessing any page c. It will disallow Google from crawling any of the dynamically generated pages. It will allow the altavista scooter bot to access every page d. None of the above Ans: b 19.  Which of the following statements about RSS are correct? a. It is a form of XML b. It stands for Realtime streamlined syndication c. It is a good way of displaying static information d. It is a microsoft technology Ans: a 20.  Which of the following statements are correct with regard to using javascript within the web pages? a. It uses up the valuable space within the webpage, which should be used for placing meaningful text for the search engines b. Search engines cannot read Javascript c. It is a good idea to shift the Javascript into a separate file d. Most of the search engines are unable to read links within Javascript code Ans: b,c 21.  Which of the following options are correct regarding the Keyword Effectiveness Index (KEI) of a particular keyword? a. It is directly proportional to the popularity of the keyword b. It is inversely proportional to the competiton for the keyword c. It is directly proportional to the chances of the keyword ranking on the first page of the Google search results Ans: c 22.  What is the illegal act of copying of a page by unauthorized parties in order to filter off traffic to another site called? a. Trafficjacking b. Visitorjacking c. Viewjacking d. Pagejacking Ans: d 23.  Which of the following search engines offers a popular list of the top 50 most searched keywords? a. Google b. Yahoo c. AOL d. Lycos Ans: d 24.  Which of the following search engines or directories provides the main search results for AOL? a. Lycos b. DMOZ c. Google d. Yahoo e. Windows Live Ans: c 25.  Which of the following can be termed as appropriate Keyword Density? a. 0.01-0.1% b. 0.1-1% c. 3-4% d. 7-10% e. More than 10% Ans: c 26.  The following robots meta tag directs the search engine bots: a. Not to index the homepage and not to follow the links in the page b. Not to index the page and not to follow the links in the page c. To index the page and not to follow the links in the page d. Not to index the page but to follow the links in the page Ans: b 27.  What is Keyword Density? a. The number of times the keyword is used / (DIVIDED BY) the total word count on page – (MINUS)the total words in HTML on the page b. The number of times the keyword is used X (MULTIPLIED BY) the total word count on page c. The number of times the keyword is used in the page description d. The number of times the keyword is used in the page title e. The number of times the keyword is used / (DIVIDED BY) the total word count on the page Ans: a 28.  Which of the following are examples of agents? a. Internet Explorer b. Search engine spiders c. Opera d. SQL Server database attached to a website Ans: a,b,c 29.  If you search for the term “iq test” in the Word Tracker keyword suggestion tool, will it return the number of independent searches for the term “iq”? a. Yes b. No Ans: a 30.  Cloaking is a controversial SEO technique. What does it involve? a. Increasing the keyword density on the web pages b. Offering a different set of web pages to the search engines c. Hiding the keywords within the webpage d. Creating multiple pages and hiding them from the website visitors Ans: b 31.  Which of the following facts about Alexa are correct? a. Alexa provides free data on relative website visitor traffic b. Alexa and Quantcast provide information on visitor household incomes c. Alexa is biased towards US based traffic d. Quantcast only tracks people who have installed the Quantcast toolbar This is the last question of your test. Ans: a 32.  Google gives priority to themed in-bound links. a. True b. False Ans: a 33.  Which of the following methods can help you get around the Google Sandbox? a. Buying an old Website and getting it ranked b. Buying a Google Adwords PPC campaign c. Placing the website on a sub domain of a ranked website and then 301 re-directing the site after it has been indexed d. Getting a DMOZ listing Ans: a 34.  s Page is used to: a. Attract visitors from the search engines straight onto the Hallway Page b. Organizes the Doorway Pages c. Helps people navigate to different Doorway Pages d. Enables search engine bots to index the Doorway Pages Ans: d 35.  Which of the following options describes the correct meaning of Mouse Trapping? a. The technique of monitoring the movement of the mouse on the webpage b. The technique of monitoring the area on which an advertisement was clicked c. The web browser trick, which attempts to redirect visitors away from major websites through a spyware program d. The web browser trick, which attempts to keep a visitor captive at on a website Ans: d 36.  What is the most likely time period required for getting a Google page rank? a. 1 week b. 3 weeks c. 2 months d. More than 3 months Ans: d 37.  All major search engines are case sensitive. a. True b. False Ans: b 38.  Which of the following website design guidelines have been recommended by Google? a. Having a clear hierarchy and text links b. Every page should be reachable from at least one static text link c. If the site map is larger than 100 or so links, you should break the site map into separate pages d. Keeping the links on a given page to a reasonable number (fewer than 100) e. Use less than 30 images or graphics per page Ans: b 39.  How are site maps important for the search engine optimization process? a. Site maps help the search engine editorial staff to quickly go through a website, hence ensuring quicker placement b. Google gives credit to the websites having site maps. The GoogleBot looks for the keyword or title “Site Map” on the home page of a website. c. Site maps help the search engine spider pick up more pages from the website d. None of the above Ans: c 40.  Google looks down upon paid links for enhancing page rank. If a website sells links, what action/s does Google recommend to avoid being penalized? a. The text of the paid links should state the words “paid text link” for Google to identify it as a paid link b. Only Paid text links to non profit websites should be accepted c. Paid links should be disclosed through the “rel=nofollow” attribute in the hyperlink d. Paid links should be disclosed through the “index=nofollow” attribute in the hyperlink Ans: c 41.   Some words, when followed by a colon, have special meanings to yahoo. What is performed by the link: Operator? a.    it shows all the outbound links from the url b.    It shows how many pages of the site yahoo is pointing to c.    It shows all the pages that point to that url d.    It show url’s with broken links Ans: c 42.   Which blackhat Seo techniques is characterized by a method to decieve search engine, by detecting the search engine bot and “feeding” it with a different HTML code than the HTML actually served to users? a.   Coaling b.   Foisting c.   Slighting d.   Cloaking Ans: d 43.   Why is it bad idea from SEO perspective to host free articles and write ups that are very common on the internet? a.    Because they will not lead to fresh traffic b.    Because you could be penalize by search engine for using duplicate contents c.    Because you will not get the benefits of proper keyword targeting d.    Because people could turn up claiming copyright infringement Ans: b 44.   What will happen if you type the word ‘Certification-Networking’ in the Google search box? a.    Google will find the web pages about “certification” and also containing the word “networking” b.    Google will find ALL the web pages containing the word word “Certification” and “Networking” c.    Google will find ALL the web pages in which the words “Certification” and “Networking” appear together. d.    Google will find the web pages about Certification that do not contain the word Networking Ans: c 45.   Which of the following factors does Google take into account while accessing whether or not a website is an authority website? a.    The frequency with which the contents of the website is updated b.    the number of web pages containing relevant information on the main theme of the website c.    The number of in-bound natural links related to the website’s theme (or keywords) d.    None of the above Ans: c 46.   What is keyword density? a.    The number of times the keyword is used / (divided by) the total word count on page – (minus) the total words in HTML on the page b.    The number of times the keyword is used x (multiply by) the total word count on page. c.    The number of times the keyword is used in the page description d.    The number of times the keyword is used in the page title e.    The number of times the keyword is used / (divided by) the total word count on the page. Ans: e 47.   Which of the following statement about FFA pages are true? a.    They are greatly beneficial to SEO b.    They are also called Link Farms c.    They are Paid Listings d.    They contain numerous inbound links Ans: b 48.   What is the illegal act of copying of a page by unauthorized parties in order to filter off traffic to another site called? a.    Traffic jacking b.    Visitors Jacking c.    View Jacking d.    Page Jacking Ans: d 49.   What is the most likely time period required for getting a google page ranking? a.    1 week b.    3 weeks c.    1 month d.    More than 3 months Ans: d 50.   Which of the following can be termed as a good keyword selection and placement strategies? a.    Targeting synonyms of the main keyword b.    Targeting the highest searched keywords only c.    Copying competitor keywords d.    Optimizing 5 or more keywords per page (check) Ans: a, d 51. What does the term “Sandbox” mean in SEO? a) The box with paid ads that appear when you perform a search. b) This is where sites are kept till they get mature enough to be included in the top rankings for a particular keyword c) A special category of sites that are listed in kid-safe searches d) The first 10 search results for a particular keyword. Ans: b 52. High quality links to a site's homepage will help to increase the ranking ability of deeper pages on the same domain a) True b) False Ans: a 53. What aspects of a hyperlink are not important for SEO? a) The visibility of the link text. b) The anchor text, especially the keywords in it. c) The place from which the link originates. d) The place to which the link leads. Ans: a 54. Which of the following is NOT a "best practice" for creating high quality title tags? a) Make sure the title is unique for every page b) Include an exhaustive list of keywords c) Limit the tag to 65 characters, including spaces d) Write compelling copy that encourages users to "click" your listing Ans: c 55. What is the advantage of putting all of your important keywords in the Meta Keywords tag? a) There is no specific advantage for search engines b) It increases relevance to Yahoo! and MSN/Live, although Google & Ask ignore it c) They will be bolded in searches for that term d) You have to have a term in Meta Keywords in order to bid for AdWords to that page Ans: a 56. Which HTTP server response code indicates a file that no longer exists? (File Not Found) a) 401 b) 301 c) 500 d) 404 Ans: d 57. Which one of the following practice is ethical ? a) Buying links from link farms b) Having the same page twice - once in html, once in pdf. c) Using hidden text that users dont see but spiders can read d) Stuffing the metatags with keywords Ans: a 58. Is the following metatag appropriate for a site that sells ebooks? a) Yes, the following metatag contains all the keywords I need. b) No, it is rather short, I have more keywords to optimize for c) Why bother with metatags at all? d) Yes, this metatag is OK even though it could have had more keywords. Ans: d 59. What is Page Rank? a) The Alexa technology for ranking pages. b) The way Yahoo! measures how popular a given page is based on the number and quality of sites that link to it. c) The search relevancy of a page compared to the other pages in the search engine d) The way Google measures how popular a given page is based on the number and quality of sites that link to it. Ans: d 60. The de-facto version of a page located on the primary URL you want associated with the content is known as: a) Home Page b) Canonical Version c) Heretical Version d) Empirical Version Ans: b 61. What is the generally accepted difference between SEO and SEM? a) SEO focuses on organic/natural search rankings, SEM encompasses all aspects of search marketing b) No difference, theyre synonymous c) SEO refers to organic/natural listings while SEM covers PPC, or paid search d) SEO tends to be a West coast term, SEM is more East coast. Ans: a 62. Which of the following content types is most easily crawled by the major web search engines (Google, Yahoo!, MSN/Live & Ask.com)? a) XHTML b) Windows Media Player Files c) Java Applets d) Flash Plugin Files Ans: a 63. How can Meta Description tags help with the practice of search engine optimization? a) Trick question - meta description are NOT important b) They help to tell the engines which keywords are most important on your page c) They serve as the copy that will entice searchers to click on your listing d) Theyre an important ranking factor in the search algorithms Ans: c 64. Which of the following is the least important area in which to include your keyword(s)? a) Meta Keywords b) Meta Description c) Title d) Body Text Ans: a 65. If you want a page to pass value through its links, but stay out of the search engines' indices, which of the following tags should you place in the header? a) Use meta robots="noindex, nofollow" b) Use meta robots="index, nofollow" c) Use meta robots="noindex, follow" d) Use meta robots="autobot, decepticon" Ans: c 66. When do you apply for Re inclusion in a search engine's index? a) When you have made changes to your site. b) When you have changed your hosting provider and the IP address of your site. c) After you have been banned from the search engine for black hat practices and you have corrected your wrongdoings. d) When you are not happy with your current ratings. Ans: c SEO Questions and Answers pdf Download Read the full article

[mdsororitygifts]

Alpha Sigma Alpha Luggage Tags Make Great Gifts What makes our sorority luggage tags special? Super cute useful chapter gifts. Cute for dinner placeholders and take home gift for sorority guests, such as Alumnae members. Great addition to any sorority bid day bag! It is your Alpha Sigma Alpha luggage tag you decide how to use it! Alpha Sigma Alpha luggage tag measures 2 inches x 1.5 inches Material: Lead free pewter luggage tag with stainless steel ball chain Sorority Luggage Tags are not just for your luggage. Use as a door hangar, attach to your book bag, your gym bag, hang on your bulletin board... Cute $10 Big Little Gift Official Greek Licensed product. Purchase other Alpha Sigma Alpha Gifts and create your own gift set at our Alpha Sigma Alpha Store offering a complete line of Alpha Sigma Alpha Merchandise at affordable prices whether you are shopping for yourself, a daughter or a fellow sister. *Chapter discounts available. Give us a call. $9.98. Order here https://tinyurl.com/y5odgf9n. See what else is trending today for Alpha Sigma Alpha! https://manddsororitygifts.com

[floutingphysics-blog]

[ ⚘ ] — RP Character Stat Sheet

REPOST replacing the old information with your character’s information.

HOVER underlined ‘links’ for better explanations, should you need them.

PASS it on to your mutuals for a better understanding of their characters.

            TAGGED BY: @scientificmicroscope @forgedcold

BASICS —

      ♦ Designation: Genitus “Brainstorm       ♦ Face Claim: himself/Holoform - Tinkerbell       ♦ Age: -classified-       ♦ Gender: Agender       ♦ Nationality/Race: Cybertronian       ♦ Place of Creation: -classified-       ♦ Creation Date: -classified-       ♦ Sun Sign: -classified-       ♦ Residence: The Lost Light       ♦ Marital Status: Single       ♦ Alignment: Neutrual Good

LIKES —

      ♦ Drink: eliphino       ♦ Food: Energon       ♦ Day or Night: Night       ♦ Snack: energon sticks       ♦ Song: Every chance we get we run       ♦ Quote: “I do whatever the hell I want”

      ♦ Historical Character: Vector Sigma ((kinda placeholder for now sorry))       ♦ Colors: Yellow and Aqua       ♦ Pet/Animal: none       ♦ Books: unknown       ♦ Flower: Unknown       ♦ Sexuality: Pansexual

LOOKS —

      ♦ Frame Type: Seeker       ♦ Optic Color: Yellow       ♦ Hair Color: [Human] Blonde       ♦ Skin Color: [Human] idk       ♦ Body Reference: Tinkerbell?       ♦ Beauty Scale: Unknown.

TAGGING —

Nobody, if you want to do this, feel free!

[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

[cousticks]

On extremely rare occasions, during months when I am in a tight spot for money, I will go to the casino and use this skill of mine to grab some easy money and go home.

so where's the oda lives aus where oda is the most annoying customer at sigma's casino

[cousticks]

pls do sigma, akutagawa, and tecchou for the character opinion bingo :)

whoa three for one!!!!! im so sorry im a fake fan and haven't gotten to the hunting dogs or sigma yet, so i'm going off what i do know! (i know a decent bit about sigma, not as much about tecchou =( he seems like a really interesting character though) someday i'll get out of my time machine and figure out what's happened the last two seasons but for now learning via osmosis through my mutuals is quite fun

[cousticks]

bsd headcanons i have a lot [this is only a few]

+ tecchou is an insomniac

+ sigma is autistic [me too sigma, me too] + dazai has a cat, he called it soemthing stupid like glowbug when its a black cat with no ‘light/glowey’ spots [he has an infatuation with misnomers]

+ nikolai is basically a human heater and runs warm while everybody else in the decay [aside from fukuchi] runs cold

+ dazais love language is gift giving!! :) but he’s really subtle about it, instead of up and giving it he’d just leave it on somebody like kunikidas desk [kndz ftw!!!!!!!]

+ in a modern au chuuya owns a german shepard

+ like once a month chuuya and dazai will spend the day and night together and just fuck around like they did when they were 15. both claim to hate it and find it annoying and blame the other for inviting eachother, but at this rate there’s been no invitation for years they’ll just go to the meeting spot and trust the other is there.

i can't say anything about any of the doa or hunting dogs or related because i'm a horrible bsd fan and not actually there yet despite being here for going on five years (sorry i'm obsessed with precanon) but i am trusting your headcanons and will keep them in my heart when i get there anon

dazai having a Thing with misnomers is really really fun to me actually i really like that one. he would be a dastardly little contrarian like that. incredible. i like this one a lot a lot

the gift giving one is so fun. i always saw his as a really derived form of acts of service in a way, but i also do like the idea of him silently leaving little trinkets and do think it fits him. like... pissing off kunikida to the point he breaks another pen, then sneaking a new one or two into his drawer a couple days later. delightful. [kndz real you're so right]

german shepards are SUCH nice dogs. chuuya would love one

chuuya and dazai having a little ritual where they can go decompress and just. be stupid despite everything is so important to me. you get it

[mdsororitygifts]

Alpha Sigma Tau Luggage Tags Make Great Gifts What makes our sorority luggage tags special? Super cute useful chapter gifts. Cute for dinner placeholders and take home gift for sorority guests, such as Alumnae members. Great addition to any sorority bid day bag! It is your Alpha Sigma Tau luggage tag you decide how to use it! Alpha Sigma Tau luggage tag measures 2 inches x 1.5 inches Material: Lead free pewter luggage tag with stainless steel ball chain Sorority Luggage Tags are not just for your luggage. Use as a door hangar, attach to your book bag, your gym bag, hang on your bulletin board... Cute $10 Big Little Gift Official Greek Licensed product. Purchase other Alpha Sigma Tau Gifts and create your own gift set at our Alpha Sigma Tau Store offering a complete line of Alpha Sigma Tau Merchandise at affordable prices whether you are shopping for yourself, a daughter or a fellow sister. *Chapter discounts available. Give us a call. $9.98. Order here https://tinyurl.com/y46kwojx. See what else is trending today for Alpha Sigma Tau! https://manddsororitygifts.com

[mdsororitygifts]

Sigma Kappa Luggage Tags Make Great Gifts What makes our sorority luggage tags special? Super cute useful chapter gifts. Cute for dinner placeholders and take home gift for sorority guests, such as Alumnae members. Great addition to any sorority bid day bag! It is your Sigma Kappa luggage tag you decide how to use it! Sigma Kappa luggage tag measures 2 inches x 1.5 inches Material: Lead free pewter luggage tag with stainless steel ball chain Sorority Luggage Tags are not just for your luggage. Use as a door hangar, attach to your book bag, your gym bag, hang on your bulletin board... Cute $10 Big Little Gift Official Greek Licensed product. Purchase other Sigma Kappa Gifts and create your own gift set at our Sigma Kappa Store offering a complete line of Sigma Kappa Merchandise at affordable prices whether you are shopping for yourself, a daughter or a fellow sister. *Chapter discounts available. Give us a call. $9.98. Order here https://tinyurl.com/yxjrucrm. See what else is trending today for Sigma Kappa! https://manddsororitygifts.com

[mdsororitygifts]

Phi Sigma Sigma Luggage Tag I Pewter $9.98. Order here https://tinyurl.com/y2z8oklc. Phi Sigma Sigma Luggage Tags Make Great Gifts What makes our sorority luggage tags special? Super cute useful chapter gifts. Cute for dinner placeholders and take home gift for sorority guests, such as Alumnae members. Great addition to any sorority bid day bag! It is your Phi Sigma Sigma luggage tag you decide how to use it! Phi Sigma Sigma luggage tag measures 2 inches x 1.5 inches Material: Lead free pewter luggage tag with stainless steel ball chain Sorority Luggage Tags are not just for your luggage. Use as a door hangar, attach to your book bag, your gym bag, hang on your bulletin board... Cute $10 Big Little Gift Official Greek Licensed product. Purchase other Phi Sigma Sigma Gifts and create your own gift set at our Phi Sigma Sigma Store offering a complete line of Phi Sigma Sigma Merchandise at affordable prices whether you are shopping for yourself, a daughter or a fellow sister. *Chapter discounts available. Give us a call. See what else is trending! https://manddsororitygifts.com