Dhravya Shah
@dhravya
beautiful. a friend shared a database of AI Hallucination cases, their reasons and outcomes. so much alpha and research opportunity omg
4 replies
3 recasts
16 reactions
Colin Charles
@bytebot
Yeah - this is for legal use cases. Very useful (and probably not even complete). Pair this with say, the China censorship datasets on huggingface, and you might be able to train Qwen/R1 to be “free” (perplexity did the latter), and then tune it to be more of a legal eagle avoiding said mistakes
0 reply
0 recast
0 reaction