deroholic

This problem still exists for release 96. If you --add-exclusive-node= to the command line, the issue goes away so it probably has something to do with being too aggressive with...

> I found the cause of it. When there is only one connected peer and this condition is false > > https://github.com/deroproject/derohe/blob/ec5da1c381a95129cd10be66b757d21798079d91/p2p/connection_pool.go#L427 > > it will never reach `goto done`...

> Your work is outstanding, and I admire the efficiency achieved in your mamba implementation. > > However, I’m concerned about its accessibility and broader adoption in comparison to transformer-based...

> In practice, depending on your setting, you may be able to simply concatenate the sequences and pass the whole sequence in (without enforcing state resetting at sequence boundaries). I've...

Downgrading to this will work: deepspeed 0.13.5 deepspeed-mii 0.2.2