5 ms·
Can LLMs give us AGI if they are bad at arithmetic?
- Terretta 10mo agoCode here: https://github.com/wesm/llm-arithmetic-benchmark https://github.com/wesm/llm-arithmetic-benchmark
- jqpabc123 10mo agoI found that the frontier models surprisingly struggled to do basic arithmetic LLM = Large *Language* Model. They are designed to do language prediction, not logical deduction. Attempting to extract truth or accuracy or facts or intelligence from a probabilistic algorithm is an exercise in frustration.