4 ms·
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 worklo
by cultofmetatron 5mo ago
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.
- the_real_cher 5mo agoI believe training is way more processor intensive than inference.
- LogicFailsMe 5mo agoThey're Turing complete. What else do you need?
- porphyra 5mo agotechnically in order for something to be turing complete it needs infinite memory
- LogicFailsMe 5mo agoThe more (memory) you buy, the more you save!
- Our_Benefactors 5mo agoIt’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.
- Npovview 5mo agoThere is a reason why Google has tpu8i and tpu8t
- airjason 5mo agonot for llm full training, but can do some finetuning for sure.