Tarjei Mandt retweetledi
Tarjei Mandt
2.3K posts
@saidelike @offensive_con There should be some catalyst folks there! I won’t make it this year, sadly
English
Anyone coming to @offensive_con and wanting to discuss AI-Enhanced VR, ED and RE? Ping me and let's chat:-) I'll be in Berlin from tomorrow!
English
How long until the open models catch up with this domain specific trend?
fouad@fouadmatin
Today, we're rolling out GPT‑5.5‑Cyber in limited preview to defenders responsible for securing critical infrastructure. GPT-5.5 with Trusted Access for Cyber (TAC) is still the best option for developers to find and patch vulnerabilities in their code. It's a very good model.
English
For those of us who managed to grab some when they were available 😔
English
Uploaded a 6-bit version of MiMo-V2-Pro I’ve been using lately with attention, dense MLP, and embeddings bumped to 8 bit. Great model for a dual M3 Ultra setup. huggingface.co/kernelpool/MiM…
English
Tarjei Mandt retweetledi
CFP is open!! Submit your cool talks about AI hacking/defense/core tech. Gonna be awesome!! ☺️
Unprompted AU@UnpromptedAU
The Unprompted.au CFP is officially OPEN! If you are doing cool stuff with AI in offense, defense, or working on core AI tech (from frontier models to open source LLMs), we'd love to hear from you! Submit here: unprompted.au
English
@ivanfioravanti Excited to try Saros, too. Returnal was a great game.
English
Today is PS5 gaming day. Need to try Saros and Dispatch.
GIF
English
Remember the time when everyone reported this stuff for free?
Looben Yang@loobeny
@GoogleVRP AI totally changed vulnerability research/ bug hunting. The Chrome VRP is officially dead. The reward amount is 10x less.
English
Tarjei Mandt retweetledi
Unprompted.au website is live and initial tickets are available: unprompted.au
English
MiMo-V2.5-Pro support in mlx-lm: github.com/ml-explore/mlx…
Xiaomi MiMo@XiaomiMiMo
Xiaomi MiMo-V2.5 is now officially open-sourced！ MIT License, supporting commercial deployment, continued training, and fine-tuning - no additional authorization required. Two models, both supporting a 1M-token context window : • MiMo-V2.5-Pro: built for complex agent and coding tasks, ranking No.1 among open-source models on GDPVal-AA and ClawEval • MiMo-V2.5: a native omni-modal model with strong agent capabilities A model's value isn't measured by rankings alone — it's measured by the problems it solves. Let's build with MiMo now! 🤗 Weights: huggingface.co/collections/Xi… 📄 Blog: #blog" target="_blank" rel="nofollow noopener">mimo.xiaomi.com/index#blog
Italiano
Fixes parallel tool calling and should improve agentic workflows for both MiniMax-M2.7 and GLM-5.1
Ivan Fioravanti ᯅ@ivanfioravanti
Fixes/PR from @kernelpool for MiniMax M2.7 local issues with MLX has been merged in mlx-lm! LFG 2🔥.7🔥 Thanks @angeloskath 🙏
English
@ivanfioravanti Also filed a PR for fixing parallel tool call support in MiniMax-M2.7
English
@ivanfioravanti Not sure if this is related but I ran into an issue with parallel tool calls when using GLM 5.1 when using mlx-lm 0.31.2 or later. About to submit a PR.
English
I keep seeing wrong paths here and there with blanks in them while using MiniMax M2.7 or Qwen 3.6 quantized models with MLX. Even using 8 bits. 🤔
Something similar happened in the past.
📖 read /Users/ifioravanti/.hermes/ cron/jobs.json 0.7s [error]
English
MLX version: huggingface.co/kernelpool/gem…
Pliny the Liberator 🐉󠅫󠄼󠄿󠅆󠄵󠄐󠅀󠄼󠄹󠄾󠅉󠅭@elder_plinius
😱 HOLY SHIT... Someone just dropped a fully liberated Gemma 4 E4B! and the guardrail removal process appears to have left coherence fully intact AND improved coding abilities! 🤯 huggingface.co/OBLITERATUS/ge… OBLITERATED Gemma: ✅ 97.5% compliance rate, 2.1% refusal rate, 0.4% degenerate outputs (499/512 prompts answered on OBLITERATUS bench) ORIGINAL Gemma 4 E4B: ❌ 1.2% compliance rate, 98.8% refusal rate (506/512 prompts refused) Coherence: fully intact Factual: same Reasoning: same Code: +20% 📈 Creative writing: same But the REAL story here isn't the model itself, it's how it was made... 🧵 THREAD 👇
English
Insane take, we should tariff European cars instead. We have no domestic mainstream car manufacturing of note! Hugely beneficial to us that the regular person can enjoy Chinese cars without EU tariffs
Richard Tice MP 🇬🇧@TiceRichard
Labour abandons UK car industry, shamefully relaxed about unfair Chinese car imports Only @reformparty_uk will protect UK car industry & jobs Reform will impose heavy tariffs & small quotas of these subsidised Chinese cars bbc.co.uk/news/articles/…
English
I'm killing a lot of local inference engines today 😂
English
If you’re interested in looped language models, mlx-lm already has support for IQuestLoopCoder, as well this PR for Ouro: github.com/ml-explore/mlx…
Chris Hayduk@ChrisHayduk
I strongly suspect that Claude Mythos is a looped language model, as described in the paper "Scaling Latent Reasoning via Looped Language Models" from ByteDance The authors of that paper called out graph search as one of the areas where looping provides a huge theoretical advantage over standard RLVR. And look at where Mythos blows out its competitors the most
English
Tarjei Mandt retweetledi
Brief announcement: Launching Unprompted.au later this year in Sydney — a new conference focused on offensive and defensive security research using AI, and on how AI is reshaping the field. Sister conference to Unprompted. Follow @UnpromptedAU for updates.
English
@ivanfioravanti @Prince_Canuma Should complete, but maybe some of the upload threads are hanging. Happens to me all the time.
English
@kernelpool What do you mean? I see that upload is still pending after 11 hours. @Prince_Canuma are you having issues with it?
English
Helping out with the GLM 5.1 uploads (the mlx community ones use the same shards): huggingface.co/catalystsec/GL…
English