I stand with the people of Iran 👊❤️
I have a lot of Iranian friends, including my childhood best friend. Amazing people.
It's one of the great cultures and peoples in the history of the world.
Stay strong ❤️
میانگین اجاره یک متر مسکن در محلات متوسط تهران در حال حاضر نزدیک به 300 هزارتومان هست. برای اجاره یک خانه 70 متری نزدیک به 21 میلیون درماه باید اجاره پرداخت بشه. میانگین حقوق یک استادیار 15 یک دانشیار 21 و معلمین 10
🛠️ Tooling Tuesday 🛠️
We use @PyTorchLightnin to provide a framework for our @PyTorch code, cut down on boilerplate, and get easy access to the configurable bells and whistles that make our training more efficient and effective.
🧵 1/N
If you are still at CVPR, visit the JRDB-Act poster in the tomorrow afternoon poster session!
JRDB-Act is a large-scale, multi-modal, multi-task dataset which serves as a benchmark for human detection, tracking, social grouping and action understanding. #CVPR2022@JRDB_Benchmark
Y2K, aka Year 2 kilo, is a phrase representing the millennials.
Creators of the generation, find your interpretation of Y2K and make creative expressions out of those ideas!
#Y2KChallenge
فرصت برنده شدن #NFT از دست ندید☕️
🏆2 برنده
🔸برای شرکت کردن:
1⃣ پیج های زیر فالو کنید
@persiancosmos & @CoffeeDrop_io
2⃣ 3 نفر از دوستانتان در قسمت کامنت ها تگ کنید
3⃣ این توییت لایک و ریتوییت کنید
⏰24 ساعت فرصت دارید #Giveaway
🔸با هولد کردن این NFT شامل ایردراپ هم خواهید شد
TIL in #100DaysOfCloud Twitter Space
Getting Hired at @awscloud with @mikeapted and @isahuerga
🔁RT for eBook draw
🎙️ Listen to the recording while it lasts
👇 Summary, Highlights and Resources in the thread
https://t.co/Mw2tReg0AA
#JRDB new updates:
1- The JRDB server has moved from Stanford University to Monash University. The leaderbords include new metrics and new singup procesure is now in place.
2- JRDB2022 will be relased soon including improved annotations and new set of annotations and challenges.
@wightmanr@amaarora@wightmanr its interesting that this problem was known to exist for low level tasks like super-resolution, hence the move from simplified super-resolution to blind super-resolution, but the same treatment was never taken seriously for high level tasks like classification
@haltakov 🤝I’ve often achieved the biggest return on investment with careful data augmentation during training,i.e, only use augs that help synthesize samples you would expect the model encounter during inference. If you can afford, model ensembling+test-time augmentation for inference.
@moaminfahim1 @galacticgeckosg @moaminfahim1 جان اگه فرصت بود لطفا یه توضیحی بده که ملاکت برای انتخاب کالکشنا چیه؟ الان solsouls و cavemen هم حدود ۱.۵ سولانا هستن، جای رشد دارن به نظرت؟