- Jul 24, 2017
- 2,531
- 28,586
Depends on the translator. Professional pretty far. Amateur pretty close. Really the most annoying things will be the gender and subject mixups. Grammar should be good
- Jan 19, 2023
- 83
- 137
Ah the gui! It can be slow but at least it works well, I also have a potato pc but used it quite a bit for some things heh.
As long as it works it's all well, yup!
- Jan 19, 2023
- 83
- 137
If you ever need to, use you can use G4F (GPT 4 Free) which has GPT-4o for free. I've written multiple posts previous explaining how to install it and such.
While it's free, it's not perfect and you'll encounter times when a provider gives the "I can't do it" when it comes to some explciit lines but it still works if you keep at it alright! The recent updates has made G4F a ton more better compared to before, the newest version fixing issues I had before.
Hope this helps!
You don't have permission to view the spoiler content.
Log in or register now.
By the way, I just want to translate a single file with only Japanese lines. I have parsed a .js game file for Japanese text, now all I want is to translate it.
- Jul 24, 2017
- 2,531
- 28,586
Yup, It was indeed outdated. Reinstalled the new one, the problem has been solved.
The issue is with the update of httpx to 0.28.
They removed the deprecated keyword proxies.
Until OpenAI pushes an update downgrade httpx to 0.27.2 to fix the issue.
To do so use this command :
Last edited:
Hey run into an issue in my translation, the player name is being placed weirdly ( e.g \N[1]"dialogue") I don't know if this is due to my Vocab.txt (\N[1] (\N[1]) - Male) or if it is something else?
- Jul 24, 2017
- 2,531
- 28,586
Do yourself a favor and replace the n variables with the actual actor names in question. GPT doesnt always recognize the variable.
You can replace them back after if needed.
So i should put in a name that i know i can replace then? i'l give it a go.
Last edited:
- Jul 24, 2017
- 2,531
- 28,586
Two new models I wanna talk about!
First things first, the newest Meta release brought us another 70b... with much better EVERYTHING! Translations I give it that the old 70b model would flounder on, managed to at least be somewhat competent and while it's not nearly as good as the 400b model, the fact that I could say it's pretty close for a 70b model is downright fantastic! That, PLUS META is working on a something called "BLT" or "
2nd one really isn't a upgrade for translations, but Google's newest flash model is REALLY good for OCR, even better than their pro and experimental models, including the one that was release earlier this month. And, unlike those models, it seems like it can handle more than one picture at a time (Depending on the complexity of the pictures) and it's not a bad translator either.
Honestly, everything that's been releasing for the last 3 months have truly been amazing! I genuinely hope to see more releases like these during Christmas.
First things first, the newest Meta release brought us another 70b... with much better EVERYTHING! Translations I give it that the old 70b model would flounder on, managed to at least be somewhat competent and while it's not nearly as good as the 400b model, the fact that I could say it's pretty close for a 70b model is downright fantastic! That, PLUS META is working on a something called "BLT" or "
You must be registered to see the links
" which, from what I understand, is not reliant on tokenizer. Have no idea what this really means, but supposedly it could mean much more organic or "natural" responses from the AI built on this concept rather than just predicting a response. The model they benched versus their other non BLT llama models scored HIGH on a bunch of different benchmarks.
2nd one really isn't a upgrade for translations, but Google's newest flash model is REALLY good for OCR, even better than their pro and experimental models, including the one that was release earlier this month. And, unlike those models, it seems like it can handle more than one picture at a time (Depending on the complexity of the pictures) and it's not a bad translator either.
Honestly, everything that's been releasing for the last 3 months have truly been amazing! I genuinely hope to see more releases like these during Christmas.
- Jan 19, 2023
- 83
- 137
Say dazedanon have you encountered this error?
I'm translating this
I'm getting this "RecursionError: maximum recursion depth exceeded" whenever it encounters the end of the file or such when translating a map###.json file and for other files like system.json I'm getting "✗ list index out of range" errors which causes them to not be translated at all.
Using the last version of the script I was using fixes the "✗ list index out of range" error, however the "RecursionError" does still happen from the time to time when translating the map###.json files but not as much...I think.
The error appears like this:
I wonder if it's because when using the module, the number of lines that the script is suppose to report is instead shown as a question mark?
In both versions, I can confirm that I encountered a bug where the lines containing "「……。」" were for some reason skipped and instead adding the "..." translated line in the next line, causing the lines to be out-of-order and worse, losing translated lines when reaching the end of the lines. Thankfully the translation history has the lines that are lost so I just manually fixed the placement of lines...
This is the script I was using last that worked (besides the recursion error):
I don't know which version since that one when the issues appear again...
Ah I'm using Python 3.13.1 with g4f like usual with fully up-to-date requirements too.
I'm translating this
You must be registered to see the links
for fun and noticed I'm getting this strange error, even after updating to the newest version of your script you made around 2/3 hours ago! This one:
You must be registered to see the links
I'm getting this "RecursionError: maximum recursion depth exceeded" whenever it encounters the end of the file or such when translating a map###.json file and for other files like system.json I'm getting "✗ list index out of range" errors which causes them to not be translated at all.
Using the last version of the script I was using fixes the "✗ list index out of range" error, however the "RecursionError" does still happen from the time to time when translating the map###.json files but not as much...I think.
The error appears like this:
You don't have permission to view the spoiler content.
Log in or register now.
I wonder if it's because when using the module, the number of lines that the script is suppose to report is instead shown as a question mark?
Python:
CommonEvents.json: | | 64/?This is the script I was using last that worked (besides the recursion error):
You must be registered to see the links
I don't know which version since that one when the issues appear again...
Ah I'm using Python 3.13.1 with g4f like usual with fully up-to-date requirements too.
- Jul 24, 2017
- 2,531
- 28,586
Hmm I just TLed an MV game without issues not sure why that's happening. Can you give me the system.json that's giving you issues and I'll take a look.
- Jan 19, 2023
- 83
- 137
Here you go! I'll give you not just the system.json but one of the Map###.json files that has the "「………。」" text too.
Be sure to check the json to see if that text is still there and another "..." appears right after it!
Ah I remember you, you're the one who wrote in a previous post if this supported 18+!
Curious to try DazedMTL eh? Well to answer your question if there's a free to use this...
Yes yes! There's a free way to use this, I've written a couple posts about so check this post first!
https://f95zone.to/threads/dazedmtl...anslations-using-chatgpt.155635/post-15225135
Make sure to follow my instructions and everything should work A-okay and if not, just tell me and I'll try my best to help ya out!
- Jul 24, 2017
- 2,531
- 28,586
This looks to be an issue with the files themselves. The encoding seems to be all messed up and there are a bunch of undefined symbols that are likely throwing off GPT. That or the files you gave me are out of wack.
Edit:
Actually despite all that it still worked for me. Maybe try the latest prompt example I have and see if that fixes it.
Last edited:
- Jan 19, 2023
- 83
- 137
Okay I'll test it, thank you! I didn't know that the .json files for that game had such strange encoding and symbols in it!
Hopefully this new version of the RPGMaker module and prompt example works!
AH I encountered an error when starting up your newest version of DazedMTL!
...Ah I see the new prompt.example, when changed to prompt.txt is set to shift-jis, causing this error:
You don't have permission to view the spoiler content.
Log in or register now.
Simply converting the encoding of that prompt.txt back to utf-8 fixes the problem, now to test!
...AH I just noticed you changed the csv module to now use Shift-JIS encoding, interesting!
If it works than I'll keep it as is, though I'm aware that Translator++ .csv files are encoded in utf-8 so hopefully
it doesn't cause any issue with that, otherwise I'll manually change the encoding back to utf-8-sig heh.
Okay now to test the RPGMAKERMVMZ mapt####.json file and such!
- Jul 24, 2017
- 2,531
- 28,586
Oops yeah need to fix the prompt. And yeah Im translating a game that uses csvs right now so I needed to switch the encoding but should be easy to switch back.
- Jan 19, 2023
- 83
- 137
Tested it out and it worked! The skipped line is no more and the translations are properly placed!
What's more is that the names are also fixed and properly done too compared to before when they looked like this:
JSON:
"Line1": "[Yabuotoko]: “Is it already this late...?”",
"Line2": "[{ \"Name\": \"Lin\" }]: “Ah! I have to go home now...”",
"Line3": "[Yabuotoko]: “Well then, if you feel like it, come over to see the cats next time, okay?”",
"Line4": "[{ \"Name\": \"Lin\" }]: “Yeah ♡”",
"Line5": "[Yabuotoko]: “See you, Lin-chan!!”",
"Line6": "[{ \"Name\": \"Lin\" }]: “Onii-chan!!”",In which I had to manually adjust the prompt to this in the rpgmaker module:
"... translation of the NPC name and just the name only"
Now I don't have to anymore hopefully!
I'll test and see if it still works with my old prompt and if not, I'll add the new prompt guidelines to my current one! ^^
...OH, about the "IndexError: list index out of range" error, that still happens but the files process well and I think it still translates them despite the issue. I'll post this error here incase you figure something out.
You don't have permission to view the spoiler content.
Log in or register now.
This might have to do with the new python 13 version I guess...
EDIT: Okay I tested it!
The skipped "..." issue is fixed no matter which prompt I use!
The new prompt.example you shared actually fixed the name issue so I'll intergrate the changes into my current prompt!
Thanks for helping thus far DazedAnon! ^_^
Last edited: