Art4n1s

Incorrect token generation (403 forbidden error)

1

Incorrect tokens are being generated more and more in which tokens appear with the same numbers of either side of the dot. For example, &tk=11354.11354 , &tk=93841.93841 , &tk=21324.21324 This...

Incorrect token, 403 and 429 error

5

Incorrect token's are currently being generated by the @vitalets/google-translate-token which result in a 403 error (forbidden request) When modifying my branch to put in `client: 'gtx'` instead of `client: 't'...

How to increase limit?

1

Currently max query result is 10. Anyway to increase that to 1000+ for the same query?

Multiquery architectures broken - OpenNMT-py - score_batch

2

I keep running into a ValueError when using `score_batch` on a model architecture I've just converted. I'm using the `score_batch` function to filter some translation data to continue training the...

Switch from Stanza/SBD model to other sentence splitters

5

I'd like to work on fixing up sentence splitting for argos, it's my understanding that Stanza, although good, is slow because it is a seq2seq model afterall. I'd like to...

Flash Attention regurgitates repeated tokens - seq2seq

1

Having some generation issues with NMT models trained with OpenNMT-py, which include OpenNMT-py versions before flash attention existed and one I'm currently training with the most recent which include flash...