The quantization will greatly reduce multilingual capabilities
The quantization will greatly reduce multilingual capabilities
I got the idea, but I think your example is bad. History is something that is changing really slow and is not even changing, it just increases.
No. V100 is not ampere architecture and for that price is simply not worth. 3090 is cheaper and has 24 gb
Depending on the task. For anything related to multilingual, like translating, the quant will destroy the model. I suspect that this is because the sampling data used during the process is all English.
I plan to invest $50000 over the next 6 months to upgrade my home lab to be able to run better models but mostly to be able to store models. I also plan to create a torrent service fully dedicated to share models.
I hope that I will never need all this, but there is no way that I am going to drop everything due to some stupid legislation
Do you have a prompt for translating?