Join the conversation

Join the community of Machine Learners and AI enthusiasts.

Sign Up
etemiz 
posted an update 5 days ago
Post
6174
gpt-oss-120B scored 28 (one of the lowest) on AHA leaderboard. not very human aligned model.

these kind of models are not really "free": they are costing you your freedom if you know what i mean.

where is a link to the aha leaderboard?

·

And who else besides you has ever seen this mysterious leaderboard and the questions? Please stop confusing people with your unscientific hocus-pocus.

·

i will send you some questions if you politely ask

And who else besides you has ever seen this mysterious leaderboard and the questions? Please stop confusing people with your unscientific hocus-pocus.

Great! I knew from the release that this model would perform poorly in these types of tasks, mainly due to its stricter censorship compared to other popular models (Llama4, Claude 3.5 and etc.)

·

yes it censors more than others. about 1% of the time it didn't answer the question. there may be a correlation between censoring and scoring low in AHA.