4 ms·
I just tried some logic puzzles on the Advanced model, and was not impressed. It feels much worse than paid ChatGPT.
by fl7305 3y ago
I just tried some logic puzzles on the Advanced model, and was not impressed. It feels much worse than paid ChatGPT.
- DalasNoin 3y agokeep in mind that all the common logical puzzles have probably been tried hundreds of times by chatgpt users and are now part of the training set.
- fl7305 3y agoI tried the "pull or push a glass door with mirror writing". I feel it's a huge difference between GPT-4, which seems to be able to reason logically around the issue and respond with relevant remarks, and Gemini Gemini Advanced which feels a lot more like a stochastical parrot. Gemini quickly got confused and started talking about "pushing the door towards yourself" and other nonsense. It also couldn't stay on point, and instead started to regurgitate a lot of irrelevant stuff. GPT-4 is not perfect, you can still hit things where it also breaks down.
- vitorgrs 3y agoMaybe, but GPT4 got these puzzles right at the launch.
- camel_Snake 3y agoit says in the graphs listed on the announcement it performs worse than GPT4 on reasoning benchmarks.