Hi Stepan,

Good to know that the English to Russian model is working. When you install OPUS-CAT plugin to Trados, it adds a batch task to the Trados batch tasks, and you can start the fine-tuning by running that batch task for your project. Running this batch task will initiate fine-tuning in the OPUS-CAT MT Engine (the standalone program which the plugin communicates with). The result of this fine-tuning will be a new fine-tuned model, which should perform better with the kind of texts it has been fine-tuned with.

There are detailed instructions on using the batch task here: https://helsinki-nlp.github.io/OPUS-CAT/tradospluginfinetune

You can also finetune with a TMX file directly in the OPUS-CAT MT Engine: https://helsinki-nlp.github.io/OPUS-CAT/enginefinetune

-Tommi

Oh, thank you so much Tommi!
Just one minor question before I go to read your links: according to my understanding, I should completely translate a file first so that all segments are translated and tidied up, and then I fine tune the MT with the said batch task. Is that correct?

Hi Stepan Konev

You might find this of interest: https://community.sdl.com/product-groups/translationproductivity/f/openexchange_applications/35718/opus-cat-mt-fine-tuning-graph

Daniel

Wow! Will carefully read this. Thank you!
But am I right in thinking that terms "fine-tuning" and "training" are used interchangeably?

The fine-tuning functionality is meant more as a preprocessing step before starting work a project. The workflow I had in mind when I designed was as follows:

1. You receive a translation job with existing translations or a useful TM (project-specific translations).
2. You initiate the fine-tuning of the base model in OPUS-CAT MT Engine with the project-specific translations using the batch task.
3. Wait an hour or so for the fine-tuning to complete.
4. When the fine-tuning is complete, select the fine-tuned model in the OPUS-CAT plugin (by selecting its model tag) and start on the project.

The fine-tuning is ideally suited for large-scale projects that have a lot of existing translations. The fine-tuning will usually pick up the terminology and sentence structures used in the existing translations, which makes the MT much more useful (depends on the project, though).

It might also be useful to train a fine-tuned model for a specific client, and then use this model for all jobs for this client. This would be easiest to do with the TMX fine-tuning in the engine.

Amazing! Thank you!

As for the difference between fine-tuning and training, fine-tuning means continuing the training of an existing model with a fairly small batch of highly-relevant data (thousans or tens of thousands of sentence pairs). Training usually means training a model from the scratch using a massive parallel corpus (tens of millions of sentence pairs). Then in-between those two terms there is customization, where you use a larger batch of domain-specific data to customize a model. But the boundaries between the terms are fuzzy.

Ah, I've got it now. Loads of thanks.

Hi Tommi Nieminen,
I experience a problem like this: when I use the plugin, it only works when 'Show MT as suggestion origin' is checked. Once I uncheck it, all segments remain empty. Also, I cannot filter by NMT segment status. When I select the 'Automated translations' filter, it does not detect NMT segments. Is it by design?
Ah! My bad. I should use the 'Nearal Machine Translation' filter. Ok. But the first question still valid.

(this was originally mistakenly posted in the main thread)

That's strange, the MT match should be shown as a 0 percent TM match. It might be due to Trados settings that prevent matches below a certain fuzzy percentage showing, but the 0 percent OPUS-CAT translation is displayed for me in 2019, even if with the Minimum match value being 70. I think 2021 might handle this differently, are you using that?

I think it might be best to keep Show MT as suggestion on in any case, there's no downside to it. The only benefit of disabling it is if you don't want to advertise the use of MT, but others could probably still detect from the segment metadata that MT had been used, so it's not terribly useful.

The plugin does actually contain a shortcut key that you can use to copy the MT into the target segment in a completely untraceable way, it's Ctrl+Shift+Z by default.

(btw. I'm aware that hiding the use of MT is a contentious issue, but my personal opinion is that professional translators should be able to choose their tools freely, without outside control, so I've provided this option).

(this was originally mistakenly posted in the main thread)

That's strange, the MT match should be shown as a 0 percent TM match. It might be due to Trados settings that prevent matches below a certain fuzzy percentage showing, but the 0 percent OPUS-CAT translation is displayed for me in 2019, even if with the Minimum match value being 70. I think 2021 might handle this differently, are you using that?

I think it might be best to keep Show MT as suggestion on in any case, there's no downside to it. The only benefit of disabling it is if you don't want to advertise the use of MT, but others could probably still detect from the segment metadata that MT had been used, so it's not terribly useful.

The plugin does actually contain a shortcut key that you can use to copy the MT into the target segment in a completely untraceable way, it's Ctrl+Shift+Z by default.

(btw. I'm aware that hiding the use of MT is a contentious issue, but my personal opinion is that professional translators should be able to choose their tools freely, without outside control, so I've provided this option).

Yes, I use 2021.

Tommi Nieminen said:

but others could probably still detect from the segment metadata that MT had been used, so it's not terribly useful.

Ah, ok. That makes sense. I can live with it, because in my case the translation will be imported into another CAT tool with almost all metadata lost (except Confirmed/Not cionfirmed).

Tommi Nieminen said:

Ctrl+Shift+Z

This shortcut does not seem to work for me... Or I am doing something wrong. No effect. Can I change it or just get to know all the shortcuts?
==========================
Regardless of all the above written, the plugin works just perfect. Notably after fine-tuning. I couldn't expect that. Thank you again.

I had changed the shortcut in my Trados, so Ctrl+Shift+Z was incorrect. The shortcut is unset by default, but you can set in Options -> Keyboard Shortcuts -> OpusCatTranslationProvider. Nice to hear that you find OPUS-CAT useful.

Yes, I have found it. Alt+F8 by default. Thanks.