🔥Telegram İfşa

3 days. 3 venues. 3 terrible reviewing examples. In ICCV: reviewer rejects paper for not having equation numbers (for equations that were not referred anywhere in the text). In CCS (for a paper with “all” positive scores pre rebuttal): reviewer rejects paper for missing code

An excellent and clearly written primer/tl;dr for those exploring DP auditing by Meenatchi Sundaram Muthu Selva Annamalai et al.

Wrote up some notes on python/ML project setup stuff over the weekend. Might save you some headaches (or might just be obvious things I was slow to figure out lol) anshumansuri.com/blog/2025/ml-c… If you have tricks that make your life easier, would love to hear them 👀

Introducing Kitten TTS, a SOTA tiny text-to-speech model - Just 15M parameters - Runs without a GPU - Model size less than 25 MB - Multiple high-quality voices - Ultra-fast - even runs on low-end edge devices Github and HF links below

OpenAI hasn’t open-sourced a base model since GPT-2 in 2019. they recently released GPT-OSS, which is reasoning-only... or is it? turns out that underneath the surface, there is still a strong base model. so we extracted it. introducing gpt-oss-20b-base 🧵

🙌🙌🙌

Reading books in a foreign language is a fun + effective way of learning that language. However, current ebook readers don’t support that use case well. As our submission to @LiquidAI_'s hackathon, we've designed a Leap-based app specifically for reading foreign-language books.

Very happy to share our paper with Adam Smith and Jon Ullman (Jonathan Ullman) on the “sample complexity” of membership inference (MIA). We ask: Is the number of data typically used in practice enough to build strongest possible MIAs? 🧵 link: arxiv.org/abs/2508.19458

Work on AI agents and want actual autonomous deployment instead of hacky localized implementations? Check out our solution ✨SAGA✨ : solves the security & governance issues everyone else ignores. Ready to deploy: github.com/gsiros/saga

My friends, today I am excited to announce AI Sequrity (AI Sequrity Company). Our mission is to provide developers and enterprises a painless and stress-free deployment of AI that is secure by design. You heard this right. You can deploy your AI agents and get guaranteed security. The

New blog post: The bug that taught me more about PyTorch than years of using it started with a simple training loss plateau... ended up digging through optimizer states, memory layouts, kernel dispatch, and finally understanding how PyTorch works!

🌟 Weekly AI Evaluation Spotlight 🌟 🤖 Did you know malicious actors can exploit trust in AI leaderboards to promote poisoned models in the community? This week's paper 📜"Exploiting Leaderboards for Large-Scale Distribution of Malicious Models" by Anshuman Suri explores this!