3 ms·
To your point on reputation: Didn't Meta recently unpublish a LLM after it was shown to hallucinate wrong answers? Smaller AI companies would have stuck to thei
by sangnoir 4y ago
To your point on reputation: Didn't Meta recently unpublish a LLM after it was shown to hallucinate wrong answers? Smaller AI companies would have stuck to their guns, but Meta has internal controls to guard against damaging the "brand". IIRC, the tweet announcing the retraction was particularly salty,<speculation> sounded like someone whose hand was forced</speculation>
- dr_zoidberg 4y agoYes, the Galactica LLM by Meta. Though LeCun isn't an author in the paper, he is "Chief AI Scientist for Facebook AI Research (FAIR)"[0], and he was quite angry about closing the Galactica demo[1]. [0] https://ai.facebook.com/people/yann-lecun/ https://ai.facebook.com/people/yann-lecun/ [1] https://twitter.com/ylecun/status/1593293058174500865 https://twitter.com/ylecun/status/1593293058174500865