Azhar

[Ather Salem® @AthSal01]

کل نیپال میں آنے والے تباہ کن سیلاب کے بعد…

بھارت میں ایک ڈیم سیلابی ریلے کو اپنے اندر سمیٹ رہا ہے۔ پانی کا یہ خوفناک دباؤ دیکھ کر اندازہ لگایا جا سکتا ہے کہ قدرتی آفت کتنی طاقتور ہے۔

ایک طرف نیپال میں سیلاب نے تباہی مچائی، تو دوسری جانب بھارت میں ڈیم اس طوفانی پانی کو سنبھالنے کی کوشش کر رہا ہے۔

پانی کی طاقت… انسان کی تمام تدبیروں پر بھاری..!

[Alternative News @AlternatNews]

When the USSR dissolved, Ukraine was the richest part with the world’s best fertile soil, a developed industry, army and 52 million inhabitants.

In just one generation, Ukraine threw it all away. They started to antagonize Russia and Russia stopped buying Ukrainian goods.

Even the world’s best freighter aircraft manufacturer Antonov went bankrupt.

Then Ukraine threw away the deal of free gas for transporting Russian gas to Europe.

Then they ruined the entire industry, lost 4 provinces to Russia and could no longer live on the taxes they collect. They are an eternal support case.

The 52 million inhabitants are now 25 million. Many emigrated and millions left in 2022. Half are old and will die soon. With a fertility rate under 1 per woman, Ukraine will be down to 15 million soon.

After one generation the most successful country lies in ruins.

People in Ukraine will forever ask “what possessed us?”. The answer is Europe. The EU used Ukraine to unite around a common enemy, to strengthen central command and control.

Disgusting!

[Math Files @Math_files]

One of Einstein’s students asked him, “What does logic mean?”

Einstein replied with a question:

“Suppose two workers go down a chimney to clean it. One comes out with a dirty face, and the other comes out with a clean face. Who will wash their face?”

The student immediately answered, “The one with the dirty face.”

Einstein said, “That answer is not correct. The person with the clean face would wash their face because they would see that their colleague’s face was dirty and assume their own face was dirty too. The person with the dirty face would see that the other person’s face was clean and assume their own face was also clean.”

The student said, “That makes sense.”

Einstein replied, “No. The question itself is not logical. If both workers went into the same chimney at the same time, it would not make sense for one to come out clean and the other dirty.”

The lesson is simple: Sometimes the problem is not the answer. The problem is the question itself.

[𝕄𝕖𝕞𝕠𝕟𝕒𝕙 𝕎𝕒𝕣𝕒𝕚𝕔𝕙👒 @mw_233]

Always turn to Allah with complete faith, unwavering hope, and heartfelt persistence. Never hesitate or doubt when you ask, for nothing is too great, too difficult, or too much for Allah to give. 🤍
Assalamualaikum !
Have a blessed Morning ✨

[Tom Yeh @ProfTomYeh]

Backpropagation by hand ✍️ ~ 11 steps walkthrough below

Backpropagation is the algorithm that actually trains a neural network, and it is where most people stop following along. It is not calculus you cannot do. It is matrix multiplication, working backward, one layer at a time.

So I drew and calculated one entirely by hand.

Goal: push the loss gradient back through a 3-layer network and land on a new value for every weight and bias.

= 1. Given =

A 3-layer perceptron, an input X, predictions Ypred = [0.5, 0.5, 0], and the truth Ytarget = [0, 1, 0].

= 2. Backprop gradient cells =

Let us draw empty cells for every gradient we are about to compute. The shape of the answer comes first.

= 3. Layer 3 softmax =

We get dL/dz3 straight from Ypred minus Ytarget = [0.5, -0.5, 0]. No chain rule needed, and that shortcut is the whole reason softmax and cross-entropy are paired.

= 4. Layer 3 weights and biases =

Let us multiply dL/dz3 by [a2 | 1]. One multiplication gives the gradient for W3 and b3 together.

= 5. Layer 2 activations =

We multiply dL/dz3 by W3 to get dL/da2. The gradient moves back across a layer the same way the signal moved forward.

= 6. Layer 2 ReLU =

Let us pass it through the gate: keep the gradient where the activation was positive, zero it everywhere else.

= 7. Layer 2 weights and biases =

We multiply dL/dz2 by [a1 | 1]. The same figure as step 4, one layer up.

= 8. Layer 1 activations =

Let us multiply dL/dz2 by W2.

= 9. Layer 1 ReLU =

We apply the same gate again, now on a1.

= 10. Layer 1 weights and biases =

Let us multiply dL/dz1 by [x | 1], and every weight in the network now has a gradient.

= 11. Update =

We subtract, and the network has learned. In practice a learning rate scales this step.

The gradients:
dL/dz3 = [0.5, -0.5, 0]
dL/da1 = [1, -2, 2, -1]
dL/dz1 = [0, -2, 2, -1]

The takeaway: matrix multiplication is all you need. Just like the forward pass, backpropagation is matrix multiplications end to end. You can do every one by hand, slowly and imperfectly, which is exactly why a GPU's ability to do them fast mattered so much to deep learning.

💾 Save this post!

[Azhar @datazone]

@heyalexmoore Drive

[Alex Fedotoff @FedotOff90]

Pakistan is the 5th most populous country on earth and almost nobody in US ecom is paying attention.

240 million people. Median age 22. Mobile penetration above 85%. The 4th largest country in Asia by population after India, China, and Indonesia.

Almost zero US-style direct response presence in the entire country.

Urdu content works. English works in tier 1 cities (Karachi, Lahore, Islamabad). Cash on delivery dominant across all tiers (80%+ of online orders). Meta CPMs at $1-2 in beauty and supplements. EasyPaisa and JazzCash handle digital wallet payments but recurring billing is not standard infrastructure.

The Pakistani consumer market for personal care is estimated at $3B+ annually. Skincare and beauty alone clears $800M-$1B. Supplements around $500M. All categories growing double-digit YoY. None of these markets are served by US-style direct response brands.

A friend's cousin runs a hair growth brand exclusively in Pakistan and UAE. $410K/month run rate, 2 years in. Single product. CAC $11 on paid traffic, but COD changes the real math: 25-35% of cash-on-delivery orders get refused at the door, so effective cost per delivered order runs closer to $16. AOV $34. Reorder rate within 60 days around 31% (the entire retention play is repeat purchase via WhatsApp re-engagement, not auto-subscription, because the infrastructure isn't there yet).

What's hard about it: regulatory compliance for supplements requires DRAP (Drug Regulatory Authority of Pakistan) registration which takes 4-6 months. If you're entering with a skincare brand, the regulatory path is faster (cosmetic notification only, 60-90 days). The supplement category is bigger but the regulatory lift is real.

The next $20M brand in supplements isn't being built in the US. It's being built in Karachi by someone you've never heard of who runs Meta ads in Urdu and never plans to ship a single bottle to North America.

The world is bigger than 30M Americans you're bidding against.

[TRT World @trtworld]

The official match ball for the FIFA World Cup 2026 was made in Sialkot, Pakistan, a city that produces around 70% of the world’s footballs. From hand-crafted expertise to cutting-edge technology, Pakistan’s football industry has been at the heart of the global game for the last four decades

[اردو ورثہ @UrduVirsa]

"جدی پشتی نوکر "
ایک مزدور باپ اور گاؤں کے چودھری کے درمیان مکالمہ
بہت ہی خوبصورت اور دکھ بھرے الفاظ والی پنجابی نظم
جناب عاقب ستیانوی
بزبان شاعر
ایک بار ضرور سنیں۔

[Rubab @rubabbx]

Pakistan has a fort that has never been conquered.

Not once. In 500 years.

Rohtas Fort. Near Jhelum, Punjab. Built in 1541 by Sher Shah Suri after he defeated Mughal Emperor Humayun and sent him into exile.

He built it for one reason. To make sure Humayun never came back.

4 kilometres of walls. 68 bastions. 12 monumental gates. 1,900 battlements. 9,500 stairs. Built on a hillock 300 feet above its surroundings where the Kahan River meets the old invasion route from Afghanistan into Punjab.

Every army that looked at it turned away.

Humayun came back eventually. But not through Rohtas. No one ever breached those walls.

Sher Shah Suri spent the equivalent of millions in the currency of the time building it. Then died before it was completed. His greatest fortress. Built to stop one man. Never once breached by anyone.

UNESCO inscribed it as a World Heritage Site in 1997. Called it an exceptional example of early Muslim military architecture in Central and South Asia. A milestone in the history of fort construction.

And the man who built it did something else while he was at it.

Sher Shah Suri also built the Grand Trunk Road. 1,500 miles from Calcutta to Peshawar. Inns every 12 miles. Shady trees the entire way. Built in the 16th century.

It still exists today.

You know it as the GT Road.

A fort that was never conquered. A road that still carries traffic 500 years later. Both built by the same king in Pakistan.

The world has barely noticed.

Pakistan has always inspired extraordinary things. 🇵🇰

📍XH7F+X5V, Rohtas, fort, Dina, 49400, Pakistan

https://t.co/gjV2R4FbBG

@Tripadvisor
@PearlTours
@TourPakistan786
@UNESCOarabic
@UNESCO
@GovtofPunjabPK
@PakistanJannatt

[Thomas Maurer @ThomasMaurer]

Microsoft owns a global network with over 600,000 km of fiber and 200+ PoPs, underpinning 70 Azure regions and 400+ data centers. This massive infrastructure, built over decades, makes Microsoft a leading network provider #SovereignCloud #Azure #Microsoft #AzureNetworking

[Azhar @datazone]

@SaleemFarrukh @kdastgirkhan These dams are under construction for many years . Pace of construction is very slow. Cost has doubled due to poor management. By the time, dams power come into national grid, power need will be diminished.

[Rana Atif @Rana_Atif101]

Population growth= 2.5%
Average GDP growth of last 4 years =2%

This is called a deep recession. Thank you PMLN and Shahbaz Sharif.

No Mandate =No vision=No performance

[Azhar @datazone]

@Rana_Atif101 Our 70 years average is around 4.5 to 5. Same will continue.

[Bilal Pervez @rajabilal555]

@ARNOfficiall Allāh destroys interest and gives increase for charities. And Allāh does not like every sinning disbeliever.
— Saheeh International

Surah Baqarah (276)

https://t.co/eWay3z9SkD

[Huangbo Xiyun 🪷°•☸️•°🪷 @HongZhou1110308]

@Naturalphilosy Complete Context:

[Muhammed Faisal @Intl_Mediatior]

ایک ایسا شعبہ جس نے 32 سال میں پاکستان کو موت کے دہانے پر پہنچا دیا۔ وہ ہے IPPs۔ یعنی بجلی پیدا کرنے والے 90 پرائیویٹ ادارے۔
ان معاہدوں کی قاتلانہ شرائط سے آپ پہلے سے واقف ہیں لیکن ان خون آشام اداروں کے مالک کون کون ہیں وہ بھی آپکو پتہ ہونا چاہیے
28 فیصد شریف خاندان ۔۔ %16 نواز لیگ کے رہنما ۔۔۔ تقریبا % 16 ہی زرداری اور %8 پیپلز پارٹی کے رہنما اور %10 اسٹیبلشمنٹ جو اس ملکے کی سلامتی کی ٹھیکدار ہے ۔۔ %8 وہ جن کی دوستی سمندر سے گہری اور ہمالیہ سے اونچی ہے یعنی چینی کمپنیاں ۔۔ %8 ہمارے مسلم برادر عرب کے سرمایہ کار (قطری) بابو اور تقریبا %6 فیصد پاکستانی سرمایہ دار ہیں ۔۔
یعنی 78 فیصد صرف تین گروپس ( شریف ، زرداری، اسٹبلشمنٹ) کی ملکیت ہیں۔ ۔۔ اس سے آگے کا المیہ بھی سن لیجیے ۔۔۔ یہی وہ 90 لوگ یا گروپ ہیں جس کے پاس پاکستان کی تمام شوگر ملز، سٹیل ملز، سیمنٹ، کھاد، کپڑے، بنک، LPG اور گاڑیاں بنانے کے لائسنس ہیں۔۔۔۔
یہ بجلی گھر پاکستان کی کل ضرورت کا %125 فیصد پیدا کرنے کی صلاحیت بتا کر لگائے گئےلیکن یہ بجلی گھر صرف %48 فیصد بجلی فراہم کرتے ہیں """ لیکن قیمت عوام سے %125 فیصد کی وصول کرتے آ رہے ہیں، لہذا پچھلے 5 سال میں 6 ہزار ارب دینے کے باوجود یہ ملک ان کا اج بھی 2900 ارب کا مقروض ہے۔
یہ تحریر بہت قیمتی ہے ۔۔۔ اس کو شیئر کریں تاکہ ان بدبخت اور لٹیرے حکمرانوں کاحال اس عوام کو پتہ چلے ۔۔۔۔۔۔۔۔

[Jvnior @Jvnior]

🇪🇸 BREAKING: Spanish PM Pedro calls for the IMMEDIATE ARREST of Netanyahu

He said: "Netanyahu launched the worst possible unjustified attack against Lebanon. His contempt for life and international law is intolerable. He is a criminal that must be arrested immediately.”

He also adds that the EU will suspend its Association Agreement with Israel very soon.

This is what true leadership looks like.

[Formula 1 @F1]

Watching an F1 race start will never get old 🤩

#F1 #JapaneseGP

[Ali khizar @AliKhizar]

At current prices, Pakistan oil bill, assuming no change in the consumption, is estimated to increase by 65% or $8 bn annually.