Adventures in machine translation 😂
Conversation
Notices
-
Embed this notice
マリオ (Mario Menti) (mario@neko.cat)'s status on Wednesday, 27-Nov-2024 02:43:21 JST マリオ (Mario Menti)
-
Embed this notice
マリオ (Mario Menti) (mario@neko.cat)'s status on Wednesday, 27-Nov-2024 04:31:18 JST マリオ (Mario Menti)
@shiawase agree on all of this. But it's a interesting service (machine translation often comes up in my day job), and apart from the different styles, you can also provide context. If I add "this is travel related" as the context, it picks the correct term.
I haven't seen that context option in other translation tools, except of course in prompting of LLMs. -
Embed this notice
Robert Belton (shiawase@mastodon.social)'s status on Wednesday, 27-Nov-2024 04:31:19 JST Robert Belton
@mario
Didn't know this LLM.
Ignoring their self hype, it's an interesting idea how they flag areas of concern and offer different styles. I wonder if the styles are post processing on the translation.
Interesting choice / weighting that led to bowel movement as a default rather than flight in a contextless situation. trained on medical material?
The page I put through it made a similar bad choice in a key phrase. translating くさい as dull not smelly when the piece was about smells. -
Embed this notice
Robert Belton (shiawase@mastodon.social)'s status on Wednesday, 27-Nov-2024 04:31:19 JST Robert Belton
@mario
It's bs that any (self respecting) professional translator would use it though. I see all these as for casual users that can't afford a translator or cheapskate companies that won't pay for one. or churning out stuff they don't think anyone would read.
-
Embed this notice