Yusuf Yiğit

[Yusuf Yiğit @yigityu]

Even genocide supporters agree now - it is genocide.

[Yusuf Yiğit @yigityu]

Gemini ve Qwen reasoning modelleri halen İngilizce düşünüyor sadece.

[Yusuf Yiğit @yigityu]

Yeni deepseek modeli r1 0528 artık think tagları arasındaki düşüncelerini Türkçe olarak da ifade edebiliyor. Eskiden cevabı Türkçe olarak verse de düşünürken İngilizce kullanıyordu. Aynı yapı Qwen üzerine distill edilen modelde de korunmuş. Enteresan...

[Yusuf Yiğit @yigityu]

Agentic sistemler için memory katmanında yeni bir açık kaynak SOTA.

[mem0 @mem0ai]

We're excited to announce our latest advancement in building production-ready AI Agents with scalable long-term memory.

Mem0 outperformed six leading baselines across diverse tasks on the LOCOMO benchmark - from single-hop and multi-hop reasoning, to temporal and open-domain scenarios. Notably, Mem0 achieved up to 11% higher accuracy than leading competitive approaches on LLM-as-a-Judge metric, and surpassed OpenAI's memory by 26%.

And by intelligently leveraging both natural language and graph-based memory structures, Mem0 dramatically reduces computational overhead, resulting in 91% lower latency (p95) compared to traditional full-context methods.

This efficiency unlocks powerful possibilities - enabling AI agents to reason faster, handle more complex interactions, and scale effortlessly in production environments.

Read the paper here: https://t.co/kwjt2Q8FKY

[Yusuf Yiğit @yigityu]

LLM platform / uygulamalarının yapıtaşlarından biri olacağını düşündüğüm güvenlik araç seti - Meta'dan.

[AI at Meta @AIatMeta]

Major updates from LlamaCon!

We’re advancing AI security with new open-source Llama protection tools and new AI- powered solutions for the defender community.

Developers can now access:

-- Llama Guard 4, a customizable safeguard that supports protections for text and image understanding across modalities.

-- Llama Firewall, a security guardrail tool that helps build secure AI systems by detecting and preventing risks like prompt injection, insecure code, and risky LLM plug-in interactions.

-- Two new versions of Llama Prompt Guard: Prompt Guard 2 86M, which improves performance in jailbreak and prompt injection detection, and Prompt Guard 2 22M, a smaller, faster version that reduces latency and compute costs with minimal performance trade-offs.
We’re also investing in new AI-enabled solutions to help the community enhance their security systems.

-- CyberSecEval 4 is our latest suite of cybersecurity benchmarks for AI systems.

-- The Llama Defender Program will help trusted partners access a variety of open, early-access, and closed AI-solutions to address different security needs.

Learn more about our new open-source protection tools and how we’re advancing AI privacy and security: ➡️ https://t.co/WXBijN3ajY

[Yusuf Yiğit @yigityu]

Deepmind'dan derin bir paper: "Welcome to the Era of Experience".
https://t.co/Cy6nDqgmCa

[Yusuf Yiğit @yigityu]

Agentic sistemler geliştirenler için Langchain'den kritik noktalar: https://t.co/ffpjN63eF9

[Yusuf Yiğit @yigityu]

Anthropic AI asistanlarının konuşma sırasında kullandığı değerleri çıkarmış. Datası için: https://t.co/z3EQTCFOr0 , ve de makalesi için: https://t.co/z5aHBFRPGG

[Yusuf Yiğit @yigityu]

Yeni bir TTS, iki undergrad, ve SOTA seviyesi performans.

[Toby Kim @_doyeob_]

Two undergrads. One still in the military. Zero funding.

One ridiculous goal: build a TTS model that rivals NotebookLM Podcast, ElevenLabs Studio, and Sesame CSM.

Somehow… we pulled it off. Here’s how 👇

[☀️👀 @zei_squirrel]

they stopped ambulances, rounded up medics and massacred them, then tried to hide it by burying them in a mass grave, then lied and said they were hamas, then video came out showing they lied and the depraved NYT and BBC say it just "appears" as though they lied. Genocidal scum

[Yusuf Yiğit @yigityu]

@adapty Report

[Yusuf Yiğit @yigityu]

Genocide, massacre, words are not enough now. Evils of the world…

[TRT World @trtworld]

Dozens of casualties were reported as Israeli air strikes hit tents of displaced Palestinians near the UNRWA headquarters in Gaza's Rafah

[Yusuf Yiğit @yigityu]

@ZeybekciNihat Ve buna zekat düşeceğini zannetmiyorum.

[Yusuf Yiğit @yigityu]

@ZeybekciNihat Neye karşı hassasiyetin olduğu kayda geçti.

[John Cusack @johncusack]

One of the programs is called “ where’s daddy “
No joke

Lavender’: The AI machine directing Israel’s bombing spree in Gaza https://t.co/Vy5qDkvWTy

[Paul Graham @paulg]

Remember when we were told children were being killed in Gaza because Hamas was using them as human shields? Now we know the real reason. When Israel wanted to kill someone, they waited till he went home, then dropped a bomb on the building, killing the whole family.

[Robert Inlakesh @falasteen47]

THE FLOUR MASSACRE:

Israeli Tanks Opened Fire On Thousands Of Starving Palestinians Who Were Waiting For Humanitarian Supplies In Gaza City.

Over 100 confirmed dead & more than 1,000 injured.

All they wanted was flour, Israeli forces decided to kill them for fun. #GazaFamine

[Paul Biggar 🇵🇸🇮🇪 @paulbiggar]

To put the intercept article in broader context:
- Israel pushed lies about sexual violence
- NYT editors demanded an article to supprt it
- NYT got a first time "reporter" to "investigate"
- they found nothing. zero. nada
- they published it anyway
- Israel murders 30,000 people

[Miguel de Icaza ᯅ🍉 @migueldeicaza]

Hamas issues their version of the October 7th events and their rationale.

Good read:

[Yusuf Yiğit @yigityu]

Very well said… #FreePalestine

[Paul Biggar 🇵🇸🇮🇪 @paulbiggar]

I wrote about the genocide in Palestine, and how tech has been silent and complicit.

https://t.co/XjZHKMJqiX