Open louis030195 opened 5 months ago
I would think that compared to mlc-llm it's more of a general framework, e.g. we have a bunch of vision models or audio models. Compared to tvm, candle tries to be much closer to the pytorch experience so that it's easy to port models, also it's in rust rather than c++ and my guess is that it's also simpler/with a smaller scope when it comes to platform support.
hi, this is great lib, i wonder how would you describe the difference or different goals between candle and mlc-llm/tvm?