Minimal implementation of Mamba, the new LLM architecture, in 1 file of PyTorch

Re:Minimal implementation of Mamba, the new LLM architecture... by TrackerFF on Wednesday December 20, 2023 @01:50PM (#38712177)

The people that are constantly up to date on this stuff tend to be AI/ML researchers and engineers. In academia, industry research groups, or startups.

They literally get paid to read papers, and implement models on a day-to-day basis.

I wouldn't worry too much not being up to date or things sounding a bit foreign. The names themselves are just that, names, the models themselves tend to be incremental versions of some previous model.

[ Parent | on HN ]

Re:Minimal implementation of Mamba, the new LLM architecture... by toasted-subs on Wednesday December 20, 2023 @01:57PM (#38712245)

Most of the startups I've chatted with seem to prioritize finding people who build products. The complaint/regret I've heard from 3-5 organizations was hiring researchers.

Researcher is more for highly funded organizations. Starrups can get by with off the shelf models.

[ Parent | on HN ]