Matthew Douglas

@HanGuo97 Thank you for reaching out! We're always interested in contributions and would be happy to help with PRs. It should be not an issue if we take an optimized...

The first option seems like a good place to start, especially since it does not sound like it would prevent later implementation of option 2. That said, I would like...

@HanGuo97 We've discussed this further internally, and thinking from a long-term perspective, it would be ideal for us to be able to build a GEMM kernel in-source here and prefer...

I would say this issue most likely would belong in [huggingface/transformers](https://github.com/huggingface/transformers), or maybe the model code directly, which is where `dotdict` is being used and breaking this. With that said,...

I would have expected `tie_weights()` to take care of that, but maybe it's not implemented in this model? I'm not very familiar with StripedHyena; this is probably a good question...

@SunMarc raises a good point about `get_output_embeddings` lacking an implementation. The other thing I find notable on this model's configuration is that the attribute is named `tie_embeddings` and not `tie_word_embeddings`...

Hi @adrienchaton, We noticed that there has been no recent activity on this issue. As a result, we will be closing it for now. If you continue to experience this...

cc: @pnunna93

We noticed that there has been no recent activity on this issue. As a result, we will be closing it for now. If you continue to experience this problem or...

There's an older PR (#1078) that I had started a while ago for the same thing. I'm OK superseding that one with this PR but with some more changes to...