13 comments

[ 2.7 ms ] story [ 35.2 ms ] thread
Attorney General Barr is not resigning; not before President Trump does. Barr is the same as Mueller,Pelosi,Schiff, & Nadler: feign opposition to cover the true motive of obstruction to keep Trump in power. Supreme Court Justice Alito & FBI Deputy Director Wray also on board. See latest updates

"Impeachment" Is A Diversion And Delay - Part II: Blocking of the "impeachment" witnesses was collusion planned before the new year. Listen to an FBI agent's disclosure from January 1, 2O2O here. The President was to resign late summer securing election for DNC. See latest updates.

Here is the zip file, which was also made available in the 3Jan2O2O update. The file within is VID_20200101_201948.mp3. Turn up the volume and put on headphones.

BB10Mp3Footage31Dec1Jan.zip 122.4mb

https://drive.google.com/file/d/1IXOOhQhHybwky8Z5pGdr9ZXhWpI...

The dialogue about the impeachment starts near the beginning. Having Biden in the White House is as good as Trump or anyone else in their organization. Obviously Schiff and Nadler pledged their allegiance to the organization by raping boys on the record, with their task being to drag out an impeachment designed to obstruct and delay any real efforts to remove the President, thus keeping Trump in power. The witness blocking was to cause an apparent uproar delaying things with legal actions until late Summer. Soon after, the President would resign, leaving any other candidate with not enough time or support to compete with an opportunistic Biden, who is as good as Trump or any other Illuminati friendly politician in the Presidency.

170 pg PDF [last updated: February|27|2O2O]:

https://drive.google.com/file/d/1S7T_kDv48E40eHzus6CTXHxcm0W...

Previously reported:

\Wag The Dog: first was feigned impeachment hearings meant to obstruct, now an attack on Iranians in Iraq. Here is what they are trying to distract from & cover up to retain power. $100+ billion in bribes to the highest offices in this country. 815+ deaths from child rapes to prove loyalty!

See the latest PDF updates: FBI Director Wray, AG Barr, SoD Shanahan, & SoS Pompeo each raped boys and were paid billions in bribes for a Soros & Koch funded child rape org. So did Trump & his "impeachment" team Nadler,Schiff,Mueller.So did media moguls Redstone,Murdoch,Moonves. What are they trying to set up? Who can arrest them since they are all bribed and in on it ? tuINBCWE, kncvewrq revermvpm vr;emvr omcwc.

Their strategy to stay in every office and obstruct until forced to leave no matter what. Feigning impeachment: see page 13O.

\\if;Download the video/audio file, put on headphones and turn up the volume. You will hear these people committing these crimes. Audio was broadcast into my apartment by outdated surveillance equipment illegally embedded within my walls. This very same technology was being used to broadcast me to the internet for five years without my consent. I own this footage. Please use this to prosecute all found within. Note:: I am obliviously speaking throughout the video, and it can be quite loud at times relative to the desired content. The are dozens more links, including these, that can be found in this PDF that was last updated on 27 FEB 2O2O:

https://drive.google.com/file/d/1S7T_kDv48E40eHzus6CTXHxcm0W...

All members of the "Illuminati"; "....an underground organization of homosexuals and child rapists..." (from pg 26: Barack Obama with Jack Dorsey).

President Donald Trump:

Demands a $4 billion dollar bribe here at 10:1...

Some really interesting work lately on "contrastive" learning, where the accuracy is really getting on par with supervised learning, e.g. https://arxiv.org/abs/2002.05709
For those of us out of the loop, could you summarize the idea of contrastive learning as a whole?
A fully illustrated article [1]

And Lilian Weng blog on self-supervision [2]

.. CPC is .. translating a generative modeling problem to a classification problem... uses cross-entropy loss to measure how well the model can classify the “future” representation amongst a set of unrelated “negative” samples...

[1] https://ankeshanand.com/blog/2020/01/26/contrative-self-supe...

[2] https://lilianweng.github.io/lil-log/2019/11/10/self-supervi...

(one variant of) The task is: Given a crop of an image (or a short audio snippet etc.), can you find the matching crop that also comes from the same image from a set containing a lot of negative samples (crops of others, unrelated images)?

To succeed, the encoder needs to be able to extract the underlying, useful information (called slow features) contained in the patch and discard the noise as this will make the retrieval process much easier.

This yields an encoder that gives pretty good representations of your inputs and you can then finetune some additional layers on top of it for your final task.

So instead of image annotation, self-supervised learning performs image manipulation to train a model. Then what? Is this network then piped into the original task at hand which would have required human annotations or is it simply for these made up tasks?
You then add a few additional layers on top, and you train those new layers in a classic supervised way. But because a lot has already been learned you need way less labels.
Optimistically if a self-supervised algorithm is capable of understanding a concept then it shouldn't need all that many examples to make it useful. Ideally you could just show it what cats look like (with just a couple of examples) and ask it to find more of them.
Two main things you can do:

1) Transfer learning -- start with self supervised model and either fine tune parameters or freeze parameters + add another layer to train your task (with way fewer params/necessary labels since you already learned about the input distribution)

2) Nearest neighbor/clustering -- no need to label all classes, simply fetch similar examples (eg find semantically similar sentences in a corpus).