Lorenzo Lupo

No solutions so far, unfortunately.

Yes, I did not have the error on single GPU.

Hi @marcwww , I suspect it is an error as pointed out in #62 and #82 .

did anyone manage to generate reasonable stuff?

> > did anyone manage to generate reasonable stuff? > > I was able to, with the suggestion in #48 . Thank you for your reply @kandykan. May I ask...

Same question here! @tsungruihon did you find a solution?

@tsungruihon I calculate the likelihood of an input sentence by summing the log-probabilities output by the model for each word of the input sentence. It looks like this: def score(self,...

Hi @gailweiss , an approximation of the log probability of the first token in the sentence should be given by `logProba[0][idxs[0]]`, right? However, I might have misunderstood your doubt.

Oh, I see! That's an excellent remark. Then, I think you could rewrite the above scoring function as: ``` def score(self, sentence): tokens = text_utils.getTokens( " " + sentence) #...

Indeed, training in this repo is performed over a long tensor representing the concatenation of all the sentences of the corpus, with the tag `` appended at the end of...