Fasttext: Dimension of pretrained vectors does not match -dim option

[PGryllos]

I am trying to use the preatrainedVectors option to run a suvervised labeling training. I have .bin file of trained vectors and I using it like that

./fasttext supervised -input corpus_pn_1577277.txt -pretrainedVectors enwiki16_lr001_dim300_ws10_epoch2_neg6.bin -output small

where corpus_pn_1577277.txt is a set of labeled sentences and enwiki16_lr001_dim300_ws10_epoch2_neg6.bin is my pretrained model.

I get

Dimension of pretrained vectors does not match -dim option

Am I doing something wrong? thanks in advance!

[PGryllos]

The size of my pretrained vectors is 300 so I added the -dim 300 option but I get the same error

[jowagner]

How did you resolve the issue in the end? Your last message says that -dim did not help but you closed the issue.

[geovedi]

had similar problem earlier, solved by switching from pretrained model format .bin to .vec.

[jowagner]

@geovedi: Have you signed or can you sign the Facebook CLA (see CONTRIBUTING.md)? If yes, please fork the repository, correct the README.md and make a pull request. You can do all this directly on github. No need to clone the repository to your local machine. Should take only a few minutes.