Leon Derczynski ⚒️☁️🏔️🌲

NLP/ML/language/security. Principal research scientist @NVIDIA, & Prof @ITUkbh. Views ostensibly professional. llmsec stan acct

[Leon Derczynski ⚒️☁️🏔️🌲 @LeonDerczynski]

Proud to announce:

💫 garak - an LLM vulnerability scanner💫

🔎 Check if a model is susceptible to common attacks
🦜 Supports HuggingFace, OpenAI, ggml, Cohere, ...
🔧 >70 probes: prompt injection, false claims, toxicity, encoding evasion, ..

https://t.co/sNjmBIQaeU

[Leon Derczynski ⚒️☁️🏔️🌲 @LeonDerczynski]

@Miles_Brundage what quality of copper are we using here

[Leon Derczynski ⚒️☁️🏔️🌲 @LeonDerczynski]

@deliprao peer - sure, seems relatable

[Leon Derczynski ⚒️☁️🏔️🌲 @LeonDerczynski]

@ChShersh push tests that turn it red. forces reading and thinking

[Leon Derczynski ⚒️☁️🏔️🌲 @LeonDerczynski]

@moyix it's not polite to rowhammer your spouse brendan

[Leon Derczynski ⚒️☁️🏔️🌲 @LeonDerczynski]

@nikaletras Below 25% gives unacceptable errors

[Rhys @RhysSullivan]

it's starting

[🎭 @deepfates]

im curious about the history of "prompts" -- as in, the > or $ or whatever in your terminal, not the text string for AI models. unfortunately this is impossible to google now

[Leon Derczynski ⚒️☁️🏔️🌲 @LeonDerczynski]

@verena_rieser @GoogleDeepMind nice work!

[Verena Rieser @verena_rieser]

What will AI safety look like in an agentic world? 🕵🏼‍♀️ important foresight piece led by my team @GoogleDeepMind

[Leon Derczynski ⚒️☁️🏔️🌲 @LeonDerczynski]

@segyges "form carries meaning" section is /really/ short on argumentation and citations. the arguments are definitely out there, in droves - do you not want to bring them in instead of the sloppy hand-waving here?

[NVIDIA AI Developer @NVIDIAAIDev]

30M downloads and counting for the NVIDIA Nemotron family on @huggingface 🤗

We're grateful for the incredible community that has made this possible.

Get started with Nemotron: https://t.co/4AtDcnOCXS

[Leon Derczynski ⚒️☁️🏔️🌲 @LeonDerczynski]

@s4rah_dev @guyrleech i'd defer to English English for this one

[Leon Derczynski ⚒️☁️🏔️🌲 @LeonDerczynski]

@ajrgd @s4rah_dev

[Leon Derczynski ⚒️☁️🏔️🌲 @LeonDerczynski]

@s4rah_dev

[Leon Derczynski ⚒️☁️🏔️🌲 @LeonDerczynski]

@arvidkahl should be green

[Leon Derczynski ⚒️☁️🏔️🌲 @LeonDerczynski]

@sarahookr 哇..!

[Leon Derczynski ⚒️☁️🏔️🌲 @LeonDerczynski]

@ObsoleteSony

[Tri Dao @tri_dao]

Nvidia continues to put out some of the strongest and fastest open models. Pretraining and post training data are released as well, something very few orgs have done

[Nathan Lambert @natolambert]

It's an honor to be competing with Nvidia for the best models with open data, checkpoints, and code.

Super excited about Nemotron 3 and Nvidia's new focus on fully open models in 2025.

[Leon Derczynski ⚒️☁️🏔️🌲 @LeonDerczynski]

@soumyesinghal @nvidia cracked