Recently I shared news about boosting #ai #planning using #LLMs.
It seems we can also boost LLMs #reasoning capabilities using #planners or #reasoners: https://arxiv.org/pdf/2409.08642

In this paper the #LLM is post-trained #machinelearning #deeplearning using the detailed arbitration of a MCTS (which next step is better) applied on domain-specific problems.

It appears the LLM can generalise to other domains, and as such exhibits some domain-independent reasoning capabilities.

📢📢NEW BLOG POST FROM SPONSOR

We interviewed Peter Crocker, the CEO of Oxford Semantic Technologies, and asked him a few questions about RDFox, its role in the Semantic Web Community, and about its future.

Read the interview here: https://semantics-org.github.io/page/news?page=2023-06-30

#semanticsconf #rdfox #reasoners