Is OPUS worthwhile? Автор темы: CafeTran Trainer
|
|---|
OPUS seems to be hot at the moment. And there are connectors for mQ an Trados.
What are your experiences with OPUS?
| | | | John Di Rico 
Франция
Local time: 22:17
Член ProZ.com c 2006
французский => английский
+ ...
| Good results, integrates with my tech stack | Jan 16, 2023 |
Hi Hans,
I found it relatively easy to deploy after watching Jamie’s session “Your Own Neural MT: Opus-CAT and Wordfast” available here: https://www.wordfast.com/conference/2022/session_recordings
I installed Opus on my server, then connected it to Wordfast Server so my team can access it via their instance of Wordfast Pro, Anywhere, or Classic.<... See more Hi Hans,
I found it relatively easy to deploy after watching Jamie’s session “Your Own Neural MT: Opus-CAT and Wordfast” available here: https://www.wordfast.com/conference/2022/session_recordings
I installed Opus on my server, then connected it to Wordfast Server so my team can access it via their instance of Wordfast Pro, Anywhere, or Classic.
My next project is to train the models using my own datasets.
The quality is comparable to other MT engines (Google, DeepL, MS, etc.).
Best,
John ▲ Collapse
| | | | | Connector for OmegaT | Jan 16, 2023 |
Hans Lenting wrote:
OPUS seems to be hot at the moment. And there are connectors for mQ an Trados.
What are your experiences with OPUS?
In the Czech language we have a saying: there is nothing older than yesterday's newspaper; yet every year the newspapers say that spring is coming, summer, autumn and winter.
Free local translator for Windows: the Opus-CAT, Jul 18, 2021
https://www.proz.com/forum/czech/352042-bezplatný_lokální_překladač_pro_windows_opus_cat.html
Bezplatný lokální překladač pro Windows: Opus-CAT
Is it worth using a free tool? Is it worth using a tool that doesn't send data away? Is it worth using a machine translator that translates between two languages without using English as a pivot language? Is it worth it to use a tool that you can refine with your data?
I can use Opus-CAT-MT in parallel with other web translators in OmegaT. Sometimes the machine translations are identical, sometimes they differ. MT in OmegaT M-translate a segment-by-segment. In Basic-CAT is available to pre-translate a whole document.
Milan
| | | | |
|
|
|
| | | You're welcome | Jan 16, 2023 |
Hans Lenting wrote:
Thank you, Milan
Hans, you're welcome.
Milan
| | | | | OPUS CAT MT Engine | Jan 17, 2023 |
I always used OPUS when I had a single major client with a specific domain. They had a huge TM I could use to train the offline OPUS engine and it worked just perfect for me. Now I rather have random clients with different domains, and using small TMs to fine-tune another model for every new client is not practical for me. However I'd like to say thank you to Tommi Nieminen for that un... See more I always used OPUS when I had a single major client with a specific domain. They had a huge TM I could use to train the offline OPUS engine and it worked just perfect for me. Now I rather have random clients with different domains, and using small TMs to fine-tune another model for every new client is not practical for me. However I'd like to say thank you to Tommi Nieminen for that unevaluable gift to us translators. Weirdly, there is not a single hint at his great work in his proz bio but I appreciate it very much.
[Edited at 2023-01-17 14:36 GMT] ▲ Collapse
| | | | Michael Beijer
Великобритания
Local time: 21:17
Член ProZ.com
голландский => английский
+ ...
| a question for Stepan/Milan | Jun 28, 2023 |
How do I know which model is being used in memoQ?
I've got it all working and am seeing the MT output in the Translation results pane in memoQ.
I installed 4 models. How do I know which one is actually being used?
###
I think i figured it out already: "Set as override model"
[Edited at 2023-06-28 10:23 GMT]
| | |
|
|
|
Michael Beijer
Великобритания
Local time: 21:17
Член ProZ.com
голландский => английский
+ ...
| Denis Fesik
Local time: 23:17
английский => русский
+ ...
| In my case, it's just a typing aid | Jun 28, 2023 |
And so was Google some time back. This thing has a problem with capitalized and first-letter capitalized words, so I can be sure that every sentence will begin with a completely random word (sometimes even one that doesn't exist). Also, it messes up alphanumeric codes and often makes changes to words that are written in a language other than the selected source language (as wise people say, you never know with neural networks). However, I'm still happy with it as a free tool: I've never been loo... See more And so was Google some time back. This thing has a problem with capitalized and first-letter capitalized words, so I can be sure that every sentence will begin with a completely random word (sometimes even one that doesn't exist). Also, it messes up alphanumeric codes and often makes changes to words that are written in a language other than the selected source language (as wise people say, you never know with neural networks). However, I'm still happy with it as a free tool: I've never been looking for an MT tool that I could trust with anything but the basics. Opus CAT has a finetune option, but somehow it doesn't work with my TMs (maybe because they are all bigger than 50k segments, which, their website says, is the upper limit on training databases)
[Edited at 2023-06-28 13:25 GMT] ▲ Collapse
| | | |
Michael Beijer wrote:
a question for Stepan/Milan
How do I know which model is being used in memoQ? I didn't know that (even though I asked that question to myself too) before you answered. Seems right, thank you.
| | | | | Any tool is just an aid | Jun 28, 2023 |
The 50K TUs limit is only applicable for Trados plugin, that is when you want to finetune directly from Trados. OPUS CAT MT Engine itself can fine-tune as many as 281149 segments (I fine-tuned it recently).
| | |
|
|
|
Robert Rietvelt
Local time: 22:17
Член ProZ.com c 2006
испанский => голландский
+ ...
| Sorry for the question, but ...... | Jun 28, 2023 |
......what is OPUS, and why should I want to have it?
[Edited at 2023-06-28 21:52 GMT]
| | | | | OPUS CAT MT engine is a free | Jun 28, 2023 |
Robert Rietvelt wrote:
what is OPUS offline machine translation engine that helps those who know how to use it. As any other MT engine it helps you type letters and words faster. Unlike other NMT engines, it combines three properties simultaneously: you can train it using your own translations, you don't have to pay for it, you don't have to expose your sacred data because it is offline.
[Edited at 2023-06-28 23:45 GMT]
| | | | Robert Rietvelt
Local time: 22:17
Член ProZ.com c 2006
испанский => голландский
+ ...
Stepan Konev wrote:
Robert Rietvelt wrote:
what is OPUS offline machine translation engine that helps those who know how to use it. As any other MT engine it helps you type letters and words faster. Unlike other NMT engines, it combines three properties simultaneously: you can train it using your own translations, you don't have to pay for it, you don't have to expose your sacred data because it is offline. [Edited at 2023-06-28 23:45 GMT]
Thank you for the info. I downloaded the program, and will have a look at it as soon I find some time.
[Edited at 2023-06-29 11:18 GMT]
| | | | To report site rules violations or get help, contact a site moderator: You can also contact site staff by submitting a support request » Is OPUS worthwhile? | Protemos translation business management system | Create your account in minutes, and start working! 3-month trial for agencies, and free for freelancers!
The system lets you keep client/vendor database, with contacts and rates, manage projects and assign jobs to vendors, issue invoices, track payments, store and manage project files, generate business reports on turnover profit per client/manager etc.
More info » |
| | LinguaCore | AI Translation at Your Fingertips
The underlying LLM technology of LinguaCore offers AI translations of unprecedented quality. Quick and simple. Add a human linguistic review at the end for expert-level quality at a fraction of the cost and time.
More info » |
|
| | | | X Sign in to your ProZ.com account... | | | | | |
Login to reply/comment 
