You can find me at [email protected] if my account gets suspended btw.

Ohio Amerika'nın Sivas'ıdır.

İnternete erişim sağlamak bir insan hakkıdır. #Türkiye ve hükümetin, halkına özgür bilgi akışına erişim kısıtlaması uygulamaya çalıştığı her ülkedeki topluluğumuzu desteklemeye devam edeceğiz. #VPN hizmetimiz sonsuza dek ücretsizdir: protonvpn.com/tr/free-vpn/ #acildeprem

Today, starting from ~13:30 UTC, some networks in #Turkey have started to restrict access to @Twitter. OONI network measurement data shows evidence that the block is being implemented through targeted throttling and DNS interference. 📉 A 🧵 👇

You can monitor the ongoing testing of Twitter in #Turkey though OONI's real-time open data: explorer.ooni.org/chart/mat?prob…

21 yıldır yurt dışındayım. Dün üniversite bölüm toplantısında millet kakara kikiri yaparken kendimi çok yabancı hissettim, burada ne işim var diye düşünüp durdum.

Twitter's internal chaos is slowing Turkey earthquake relief efforts, volunteers say ti.me/410Zndt

Birisi Amerikan banliyösünün resmini paylaşır, öteki ovalardan dağlara yerleşmemiz gerekir der; her işi biliyordunuz bir bulaşmadığınız şehircilik kalmıştı. 1. ABD banliyösü yapmanız için her 100 kişiye 95 otomobil sağlamanız gerekiyor, kabaca yılda 3,5-4 milyon otomobil eder.

I suspect GPT-4's performance is influenced by data contamination, at least on Codeforces. Of the easiest problems on Codeforces, it solved 10/10 pre-2021 problems and 0/10 recent problems. This strongly points to contamination. 1/4

Just decided to make a thread to show how many decisions and pre-processing steps you need to train large language models (LLMs) such as LLaMA. This is mainly based on CommonCrawl (CC) dataset and the pipeline used in LLaMA to generate its dataset. 1/n

1/BREAKING: We were featured in this Rest of World article about the grave dangers of AI machine translation. These tools are on the rise, increasingly used by for-profit government contractors and aid organizations that work with refugees & migrants. bit.ly/41Z9Sxu 🧵

Wow new signage on the Valencia St bike lane

Has there been a single LLM-powered product unveiling that did not have factual errors in its literal marketing materials? Happened with Google Bard, Bing chat, perpelexity, now this. How hard can it be to fact check your own marketing materials?

- How would you like your steak sir? - Verifiably rare

cries in downtown Oakland and its 4 lane streets