This site uses cookies.
Some of these cookies are essential to the operation of the site,
while others help to improve your experience by providing insights into how the site is being used.
For more information, please see the ProZ.com privacy policy.
1 Day conference on the state of MT and LLM, reality check
Thread poster: Philippe Locquet
Philippe Locquet Portugal Local time: 19:11 English to French + ...
Nov 10, 2023
So, after attending a highly relevant one day online event, here’s the skinny:
LLM are performing on par or worse that state of the art MT for big languages. Once you leave big languages (En, Es, Fr, De and maybe one or two more) the quality of LLM output drops tremendously. LLM have some flexibility to perform tasks that are harder for MT (gender style, applying a glossary with correct agreement). So, on translation tasks, in Q4 2023 we can still consider LLM to... See more
So, after attending a highly relevant one day online event, here’s the skinny:
LLM are performing on par or worse that state of the art MT for big languages. Once you leave big languages (En, Es, Fr, De and maybe one or two more) the quality of LLM output drops tremendously. LLM have some flexibility to perform tasks that are harder for MT (gender style, applying a glossary with correct agreement). So, on translation tasks, in Q4 2023 we can still consider LLM to be the same as a new MT provider. Relevance depends on language pair and pricing policy.
If you don’t use MT in your translation pipeline, this information is irrelevant to you.
Some information will be presented in an upcoming video, a lot more will come out around the time of Wordfast Forward 24 (I’ll do a full presentation there). Hope this is useful.
The contents of this post will automatically be included in the ticket generated. Please add any additional comments or explanation (optional)
Dan Lucas United Kingdom Local time: 19:11 Member (2014) Japanese to English
It is
Nov 10, 2023
Philippe Locquet wrote: Hope this is useful.
I don't/can't use MT, but it's an interesting perspective.
Thank you Dan
Philippe Locquet
Michele Fauble
Subject:
Comment:
The contents of this post will automatically be included in the ticket generated. Please add any additional comments or explanation (optional)
Neirda China Local time: 02:11 Chinese to French + ...
I mostly agree
Jan 11
Having tested extensively the quality between LLMs and MT, I could never notice a huge difference in quality when the game is equal (just prompting the LLM to translate with no extra fluff). The advantage of LLMs really relies on the ability to provide context.
I remember a paper from Tencent claiming they managed to improve the quality of a translation by prompting the LLM to refine its own work, but I could never replicate this.
Subject:
Comment:
The contents of this post will automatically be included in the ticket generated. Please add any additional comments or explanation (optional)
Manage your TMs and Terms ... and boost your translation business
Are you ready for something fresh in the industry? TM-Town is a unique new site for you -- the freelance translator -- to store, manage and share translation memories (TMs) and glossaries...and potentially meet new clients on the basis of your prior work.
Exclusive discount for ProZ.com users!
Save over 13% when purchasing Wordfast Pro through ProZ.com. Wordfast is the world's #1 provider of platform-independent Translation Memory software. Consistently ranked the most user-friendly and highest value