I’ve often said that nothing is normal about Julian Assange’s case. Here’s my list of the top 10 least normal aspects, strictly on the logistical side of monitoring the extradition proceedings (this is not to mention the many substantive issues that are also far from normal).

🕵️‍♂️ How Google's PageRank algorithm works

The PageRank algorithm gives each page a rating of its
importance, which is a recursively defined measure of importance, based on if important pages link to it.
It's recursive because the importance of a page refers back to the importance of other pages that link to it

Here's how it works in practice:
1⃣ We start with some pages and crawl them for links

2⃣ Each page has 1/N points (where N as the total number of pages)

3⃣ Add points to each page for the amount of links to it, divided by the number
of links emanating from the sources of these links

Again sharing bank nifty intraday weekly straddle. How to play by selling straddle.
Mostly will work on Monday to Wednesday. Sometimes on Friday.
Why to pay your money to learn this if u can get it here.
U need only practice.
Let’s go.

Say bank nifty future is at 32000.
Sell both 32000 call and put.
Considering current volatile market lest say premium on Monday morning is 350 on both sides.
Sold with 700 points combine premium.
If bank nifty will trade in narrow range say +\- 100 hold it.
If moving 200 points

then exit previous sold straddle. Sell new one. Let’s say BN moved +200
Now sell 32200. If bank nifty will move here within range then u will get profit through theta decay. Follow same procedure for next 150-200 pts move.

This is only for single straddle sell position.

Now in second case sell three straddle. 31900/32000/32100
In case of 100 points move no need to change.
If 200 + then future is 32200.
Exit 31900 and sell 31200 straddle.
If moving another 100 then exit 32000 and sell 32300. Continue same with 100 points move.
If no move then

it will consolidate between 32000-32200 or 32100 to 32300 and u will get theta decay.

This is for normal move. Sudden big move may come anytime. So keep SL in system.
If u r selling at 350-350 then keep SL in system 475 for both legs.
Practice it. No need to attend webinar.

Great seeing one of the fastest natural drivers of the last decade Robert Kubica today with @WilliamsRacing at @COTA #USGP #f1 #robertkubica #Poland #richenergy

Thrikkangode Sree Randu Moorthy Temple
Ottappalam, Kerala. 🚩
#keralatemples

This is very unique temple as Shiva and Vishnu both are in the “linga” form in the same Garbhagriham. The Srikovil is in a circular pattern. Temple's Prathishta was done by Parasurama.

Mahadeva and Maha Vishnu in the same Srikovil is very rare Prathishta, but poojas are done separately Dhara is done to Maha Vishnu here. People suffering from “KAIVISHAM” and mental illness are fully cured on drinking a medicinal herb supplied by a nearby family here.

The root of the banyan tree around the Ayyappa Sannidhanam is said to have great powers for skin ailments.

Very ancient temple but was covered with forest for many years.
Later two persons Chakki and Chaathan, while cutting trees their axe hit on the nearby
@SriramKannan77

Vigrahams and it suddenly started bleeding. Later Devaprashnam was held and thus found out that these were murthy's of Shiva and Vishnu which was worshipped long back. And thus temple is constructed here.
Om Namah Sivaye
Om Namo Narayanaya
🕉🙏🕉

Important paper from Google on large batch optimization. They do impressively careful experiments measuring # iterations needed to achieve target validation error at various batch sizes. The main "surprise" is the lack of surprises. [thread]

https://t.co/7QIx5CFdfJ

The paper is a good example of lots of elements of good experimental design. They validate their metric by showing lots of variants give consistent results. They tune hyperparamters separately for each condition, check that optimum isn't at the endpoints, and measure sensitivity.

They have separate experiments where the hold fixed # iterations and # epochs, which (as they explain) measure very different things. They avoid confounds, such as batch norm's artificial dependence between batch size and regularization strength.

When the experiments are done carefully enough, the results are remarkably consistent between different datasets and architectures. Qualitatively, MNIST behaves just like ImageNet.

Importantly, they don't find any evidence for a "sharp/flat optima" effect whereby better optimization leads to worse final results. They have a good discussion of experimental artifacts/confounds in past papers where such effects were reported.