News for hackers. Stuff that matters.

A comment on Running a 180B parameter LLM on a single Apple M2 Ultra. Parent comment

Re:Running a 180B parameter LLM on a single Apple M2 Ultra by slt2021 on Thursday September 07, 2023 @01:00PM (#37421998)
it is not really a moat if one engineer can leave openai with all the secret sauce in his head and replicate it elsewhere (anthropic?)
[ Parent | on HN ]
Re:Running a 180B parameter LLM on a single Apple M2 Ultra by foobiekr on Thursday September 07, 2023 @01:40PM (#37422647)
Name one software-based tech company where this isn’t true.
[ Parent | on HN ]
Re:Running a 180B parameter LLM on a single Apple M2 Ultra by slt2021 on Thursday September 07, 2023 @02:02PM (#37423025)
microsoft? gogel? FB?
[ Parent | on HN ]
Re:Running a 180B parameter LLM on a single Apple M2 Ultra by Frannyies on Thursday September 07, 2023 @02:05PM (#37423076)
I only meant the trained model.

You would need to steal it all over again as soon as the next model is trained.

[ Parent | on HN ]
Re:Running a 180B parameter LLM on a single Apple M2 Ultra by slt2021 on Thursday September 07, 2023 @03:14PM (#37424110)
no need to steal the model if training process can be reliably replicated/adopted in clean room implementation with additional optimisations.

startup as legal entity has close to 0 value, most value is in intellectual property which is stored and transmitted by meatbags.

[ Parent | on HN ]