LM self-models are built from indirect sources

LessWrong
Read the full article
Researchers on LessWrong discuss how language model (LM) self-models are constructed using indirect data sources rather than direct introspection. This insight sheds light on the internal mechanisms of LMs and their self-representation.

More in LLM & Text Generation

Moonshot releases Kimi K3.1 Pro and says it cut its hacking skills

Covered by 4 sources

[AINews] Kimi 3.1 Pro 104BA-2.8T: Fable-5.1-Class Open Weights narrows Closed-AI lead to 1 month... and Pro Highspeed API with 2-3× speed

Latent.Space

Moonshot AI Releases Kimi K3.1 Pro, First Model in New K3.1 Series

Covered by 2 sources