Nii Osae

🚀 BREAKING: We just revolutionized AI pre-training We are excited to come out of stealth and launch our revolutionary product: Litespark, a language model framework which utilizes advanced algorithms to speed up training and inference workloads for generative AI applications. Litespark reduces training time from months to days. With just 16 H100s, you can train your own billion-parameter foundation model from scratch in 26 hours. Fortune 100 companies are already transforming their AI workflows with our breakthrough framework on AWS.

Litespark is bringing unprecedented training speed and performance to Generative AI. These benchmarks are redefining the status quo and unlocking new potentials for AI infrastructure.🚀

The first AGI system will be much worse than previous AI systems at pretty much everything -- except for tests that focus on measuring general intelligence (e.g. ARC). Like a baby, it will be generally intelligent -- able to learn efficiently and adapt -- but will lack any specialized competence. However, it will be able to learn using much less data, and the capabilities it picks up will generalize much more strongly to the variability and unpredictability of the real world. As a result, after a few years, it will beat previous AI systems at absolutely everything -- including many problems that were completely out of reach before.

OpenAI is about to blow your mind with Active Inference... So what is it? > First introduced in the early 2000s in a series of papers by neuroscientist and theoretical neurobiologist, Karl Friston - active inference is a theory of how the brain uses statistical inference and generative world models to predict sensory inputs and guide actions to minimize prediction errors - helping explain human perception, action and learning > Perception updates our generative world model to reduce errors in prediction while actions change our environment to align with our predictions - minimizing the probability of errors in our predictions > It is likely that with a combination of enough compute, advancements in continuous learning / information retrieval with causal grounding and layers of active inference methods like GoT, AoT, CoV and MCTS - we may be inching closer and closer to a generative, continually learning model that operates at near-human levels of cognition ✨ I suspect that we might start to see the emergence of “energy-based” models (EBM) that operate more dynamically and continuously learn with hot-swappable memory partitions that evolve over time. They will intelligently route and adjust the level of compute needed for more sophisticated means of active inference based on the complexity of a given query. These models will also allow users to explicitly define how much energy / reasoning strength is needed at inference time. I'll be posting more on how this works under the hood soon with layered graphs-of-thought (GoT) and algos like monte-carlo tree search (MCTS)... For more on active inference check out these incredible papers: > Friston, K. (2003). "Learning and inference in the brain." > Friston, K. (2005). "A theory of cortical responses." > Friston, K. (2006). "Free-energy principle for perception and action."