can these do training or only inference? currently working on learning machine learning and I'd love to have a physical machine I could aim to build real workloads on in a few years.
can these do training or only inference? currently working on learning machine learning and I'd love to have a physical machine I could aim to build real workloads on in a few years.
They're Turing complete. What else do you need?
There is a reason why Google has tpu8i and tpu8t
technically in order for something to be turing complete it needs infinite memory
The more (memory) you buy, the more you save!
It’s possible (likely, even) to have a chip fast enough for inference, but not fast enough or with enough memory to do meaningful training runs. Like the current DGX spark.
not for llm full training, but can do some finetuning for sure.
I believe training is way more processor intensive than inference.