Yes. Knowing is not enough to acquire the "language". The art of wanting requires lived experience caring about the human outcome—un vécu. If we stop relying on this then we lose control over how outcomes benefit us.
I think the same is true of creative fields, like I see artists really worrying about AI
but I wish so badly that I could sketch animation ideas and character concepts for my agent to flesh out instead of trying to find examples
I don't have the language!
Open-source infrastructure is capital to increase our understanding of LLM internals. This toolkit is awesome and will help standardize the development and benchmarking of steering methods!
Congrats @erikmiehling and the team 👏
We're excited to announce the Steerability toolkit!
There are a ton of methods on steering out there but not a lot of cohesion in how they are built. One of our main goals with Steerability is to provide some structure to this space. We've spent a good amount of time designing
A month ago, this upgrade made our Aletheia lie detection challenge possible. Today, it brings white-box inference closer to SOTA inference engines so you can pursue even more ambitious research.
Stay tuned for the official release on NDIF!
NNsight 0.8 is here (pre-release).
New execution engine, same API. Every HF Transformers task, vLLM interp at near-native throughput, and a shiny new engine built for intervention performance and flexibility:
🧵
