search

Helsinki-NLP / OPUS-MT-train

Training open neural machine translation models
MIT License
323 stars 40 forks source link

how to train our dataset #89

Open bondeakshay opened 1 year ago

bondeakshay commented 1 year ago

hi i have my own dataset can someone please provide the procedure for how can we train our dataset

bondeakshay commented 1 year ago

@Helsinki please @veer66 @jorgtied @TommiNieminen @Traubert please provide support ,i have large dataset i want to train it using OpusMT i did all installation and run server.py for pretrained model now i have to train my own dataset please provide support on that

flutter-painter commented 11 months ago

According to the readme this could be a tricky path :

"... but documentation needs to be improved. Also, the targets require a specific environment and right now only work well on the CSC HPC cluster in Finland."

How about first using MarianMT ?

bondeakshay commented 11 months ago

Hii , Thanks for response.

I have trained model for Indian and multiple Asian languages using marianMT and able to translate text ,PDF file and Docx file and also text file ..

Thanks for providing such an amazing framework you worked great.

Thanks

Regards, Akshay Bonde 9860737353 Software Engineer, Pune, India

On Sat, Oct 28, 2023, 11:17 PM Pierre @.***> wrote:

According to the readme this could be a tricky path :

"... but documentation needs to be improved. Also, the targets require a specific environment and right now only work well on the CSC HPC cluster in Finland."

How about first using MarianMT https://marian-nmt.github.io/docs/ ?

— Reply to this email directly, view it on GitHub https://github.com/Helsinki-NLP/OPUS-MT-train/issues/89#issuecomment-1783882275, or unsubscribe https://github.com/notifications/unsubscribe-auth/A5X474O5WZM5K5IQFU6DCTTYBVAL5AVCNFSM6AAAAAAVG4LVDWVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMYTOOBTHA4DEMRXGU . You are receiving this because you authored the thread.Message ID: @.***>