🤔 LLMs Struggle With Causality

But other forms of reasoning might be well within reach

In this issue:

1. The right conclusions for the wrong reasons

2. LLMs not being great at causal reasoning

3. Halving models while keeping 100% performance