We pretrained three SYNTH 2B medical models with @pleiasfr and made the top 5 at NVIDIA's #OpenModelsCodefest, so we’re invited to present at GTC Berlin. They’re a new Pareto frontier on several medical benchmarks, and their reasoning is ~10x shorter than similar-sized models 🧵

Announcing that Pleias will be at Berlin for the @NVIDIAGTC and we'll host a synthetic data mixer on October 21: apéritif/diner all about synth and baguette models.

interpretability might be the most important scientific problem in the world right now, as I don't think we can solve alignment if we don't know what we're building. I think I have good news about it. I co-authored early work in mechanistic interpretability at OpenAI and came away pessimistic, because neural networks appeared to fold into arbitrary tangled knots. years later, I started openai's first large-scale automated interpretability project, finding about 2% of neurons legible with the primitive tools we had with modern agentic tools and a much closer look, the picture looks totally different. the vast majority in models up to trillions of parameters are legible, and their circuits beautiful. many famous examples of tangled neurons turned out to be misunderstood. The roadblocks I expected mostly aren't there. I now think ambitious interpretability will be solved by the time we need it most, and possibly quite soon, as automated scientists keep improving. it may be a bitter lesson for interpretability: the simple thing, at absurd scale, works already-clean.goodfire.com/