Hello Akbar, is it possible to add an option to make your download with custom extention, for expample if i Upload Ripley_S1_E_1.srt and the outcome is RIpley_S1_E_1_Bulgarian.srt can you make it so when i export it my outcome will be shorted like Ripley_S1_E_1_BG.srt (short for bulgarian)?
Yeah, I think it’s possible. It already works this way for most of the translators except for Google and recently Gemini, which use full language name and not the short code. I will look into fixing this soon.
Yandex has been working very poorly in batch mode lately. Interrupts the translation with a 50% translation and asks if everything is OK with the Internet. Although there is no reason for the breakage. After you continue the translation, he starts it again for some reason from the first file, although he will translate.
Thanks for bringing this to my attention. The reason it breaks in the middle is mostly due to Yandex servers blocking some requests for short period of time to avoid excessive calls from a single location or misuse of their service. This one is not under my control, but I think I can improve the resume process for sure on failure. So, will work on the fix for that.
Could you help me check my Result #JAbYcbJ4CXXXXXX
Several Word in Korean doesn’t translate to Indonesian, i think this is too much word that doesnt translate with the Correct language, i already pay for the cost of translate.
Hi Maulana, these LLM based models (OpenAI, Gemini) can often get distracted during the task and it’s hard for these to be reliable. Especially for the language pairs where there is not the great deal of data. Probably best to use DeepL or Microsoft Azure for these type of languages.
One thing I noticed is that you were using a large temperature value (of 0.5). I would suggest to use small value like either 0 or 0.1 and see if that helps improve the result?
PS. I have issued a full refund for this failed translation.
Temperature controls the creativity or randomness of the text being translated.For the translation tasks, I would normally suggest to use a value between 0 to 0.4, but there is not a hard and fast rule on which value works best. As it would vary a lot by the language pair as well as the content itself. If you are getting randomness where the translator seems to be skipping the subtitles rows translation partially or fully, probably it’s time to further reduce the temperature. Here is something I found on web which explains this in more detail: https://gptforwork.com/guides/openai-gpt3-temperature
Hi, i have to Say translator with Gemini is game changer for increase what i do, but i have some issues with full translation files, some sentence are not translate and i have to pay several Time for try to fix that. Otherwhise its cheap so il glad to give you my money for youre help 😉
Hi, thanks for sharing feedback and sorry for the trouble. For these new translators based on LLM, I would suggest to always select the correct source language for the translation. My experience is that these work best when they know the source language of the subtitles. Also, if you continue to have trouble, I would suggest to try tweaking the “Temperature” parameter in the configuration (the lower value normally brings in better result).
If still continue to have trouble, please do get back to me and I will see what can be done on the server end to make it more reliable.
To me overall user experience and satisfaction is much more important than money. And most of the money is normally forwarded to the corresponding API providers and is not my source of income anyway 🙂
This is normally the actual text file line, and not the subtitle row number. Would it be possible for you to send me a copy of the subtitle file for further investigation?
Hello Akbar, is it possible to add an option to make your download with custom extention, for expample if i Upload Ripley_S1_E_1.srt and the outcome is RIpley_S1_E_1_Bulgarian.srt can you make it so when i export it my outcome will be shorted like Ripley_S1_E_1_BG.srt (short for bulgarian)?
Yeah, I think it’s possible. It already works this way for most of the translators except for Google and recently Gemini, which use full language name and not the short code. I will look into fixing this soon.
Regards, Akbar
Yandex has been working very poorly in batch mode lately. Interrupts the translation with a 50% translation and asks if everything is OK with the Internet. Although there is no reason for the breakage. After you continue the translation, he starts it again for some reason from the first file, although he will translate.
Thanks for bringing this to my attention. The reason it breaks in the middle is mostly due to Yandex servers blocking some requests for short period of time to avoid excessive calls from a single location or misuse of their service. This one is not under my control, but I think I can improve the resume process for sure on failure. So, will work on the fix for that.
Regards, Akbar
😍😍😍
very gooood
Could you help me check my Result #JAbYcbJ4CXXXXXX
Several Word in Korean doesn’t translate to Indonesian, i think this is too much word that doesnt translate with the Correct language, i already pay for the cost of translate.
Hi Maulana, these LLM based models (OpenAI, Gemini) can often get distracted during the task and it’s hard for these to be reliable. Especially for the language pairs where there is not the great deal of data. Probably best to use DeepL or Microsoft Azure for these type of languages.
One thing I noticed is that you were using a large temperature value (of 0.5). I would suggest to use small value like either 0 or 0.1 and see if that helps improve the result?
PS. I have issued a full refund for this failed translation.
Regards, Akbar
it’s good website thanks
Hi again, Can you explain me what is temperature, more in increase it more its should be close to reality?
Temperature controls the creativity or randomness of the text being translated.For the translation tasks, I would normally suggest to use a value between 0 to 0.4, but there is not a hard and fast rule on which value works best. As it would vary a lot by the language pair as well as the content itself. If you are getting randomness where the translator seems to be skipping the subtitles rows translation partially or fully, probably it’s time to further reduce the temperature. Here is something I found on web which explains this in more detail:
https://gptforwork.com/guides/openai-gpt3-temperature
I hope this helps.
Regards, Akbar
Hi, i have to Say translator with Gemini is game changer for increase what i do, but i have some issues with full translation files, some sentence are not translate and i have to pay several Time for try to fix that. Otherwhise its cheap so il glad to give you my money for youre help 😉
Hi, thanks for sharing feedback and sorry for the trouble. For these new translators based on LLM, I would suggest to always select the correct source language for the translation. My experience is that these work best when they know the source language of the subtitles. Also, if you continue to have trouble, I would suggest to try tweaking the “Temperature” parameter in the configuration (the lower value normally brings in better result).
If still continue to have trouble, please do get back to me and I will see what can be done on the server end to make it more reliable.
To me overall user experience and satisfaction is much more important than money. And most of the money is normally forwarded to the corresponding API providers and is not my source of income anyway 🙂
Regards,
Akbar
mucgas gracias por este programa es de gran utilidad
I get this error:Error: Invalid time format at line#: 2162; Line: –>
I don’t even have that much lines.
what can i do about it?
This is normally the actual text file line, and not the subtitle row number. Would it be possible for you to send me a copy of the subtitle file for further investigation?
Regards, Akbar