Maciej Szankin

Replace unavailable link with WebArchive

1

"Build your own error-correction fountain code with Luby Transform Codes" (https://franpapers.com/en/algorithmic/2018-introduction-to-fountain-codes-lt-codes-with-python/) is no longer available. Did not find an alternative link.

I did multiple runs of search and usually after 800-1000 iterations there is no improvement. Is it a desired result? Given large search space I would expect the search to...

When running `search.py` with pre-trained weights provided in the README I am getting consistent accuracy results independently of the latency constraint (120ms yields similar results to 8ms). How could that...

Hi, Would it be possible to disclose which version/commit of `fairseq` was used & modified for this implementation? Thanks!

Update DyNAS interface

3

Signed-off-by: Maciej Szankin

Checkpoint for RetNet

2

Hi, is there a pre-trained checkpoint available for RetNet? It is not mentioned in the [README](https://github.com/microsoft/unilm/blob/master/retnet/README.md). Thanks!

Reasoning behind Alapca's default split

1

For the Alpaca dataset, the default split comprises 51,800 samples for training and 200 samples for testing [1]. What is the rationale behind such a small test set? I haven't...

Hi, is there a pre-trained checkpoint available for RetNet, and if not - are there any plans to make it available? Thanks! Related to https://github.com/microsoft/unilm/issues/1474

Torch is now supported in Python 3.11. Moving our testing env from 3.10 to 3.11 to enable tests in a newer environment.