3 ms·
You cannot use an LLM to solve mathematic equations. That’s not a training issue, that’s a limitation of a technology that’s at its core, a text prediction eng
by hnlmorg 1y ago
You cannot use an LLM to solve mathematic equations.
That’s not a training issue, that’s a limitation of a technology that’s at its core, a text prediction engine.
- tim333 1y agoYet if you look at deepmind getting gold in the IMO it seems quite equationish. Questions and answers: https://storage.googleapis.com/deepmind-media/gemini/IMO_2025.pdf https://storage.googleapis.com/deepmind-media/gemini/IMO_202...