Mamba, a state space language model (SSLM), can process long sequences with linear time and memory complexity for much larger context sizes

"Lars and the Real Girl" now feels like a prophetic commentary on our ever more complex relationship with artificial companions.

If we frame AI models as “grown” vs. “designed”, we can see the complex, iterative, and unpredictable nature of their development and usage

Llama allows developers to build and fine-tune model for coding, summarization or role-play, creating surges in natural language processing

Pursuing System 2 capabilities in AI will expand practical use of AI tech and I think deepen our understanding of our own cognition

om data collection practices to potential vulnerabilities, our digital footprints are larger and more detailed than ever.

All of this exposition helps us understand why they are giving us outputs that are wrong and what we can do to help mitigate the errors.

LLM plus search can find relevant web pages, understand the context, extract key information, and present it in a coherent, tailored manner

GRAIG benchmark provides a view of LLMs using a wider range factors, with traditional measures like reasoning and novel aspects like humor