I discovered a machine translation for BKD-186, I cleaned it up a little and tried to interpret several of the machine dialog. However it will not be a real translation since I don't fully grasp Japanese. In any case enjoy and let me really know what you believe.
3. To combat the final problem I parsed my full map tree in a few AI to question if it observed any duplicates that have this issue. Deepseek, Chatgpt couldn't take care of it, They simply invented duplicate names. Claud.AI context window was way too modest (about 3000 actress names).
Enter the username or e-mail you utilized in your profile. A password reset website link will probably be despatched to you by e-mail.
There exists logit biasing or additional schooling facts as is possible remedies, but I haven't manufactured Substantially sense of it.
They're bare minimum amount Whisper Huge subs. VAD with an incredibly reduced threshold but an increased patience and Better of, along with a repetition penalty.
All those usually are not duplicates. They are distinctive variations of subtitles for the same movie. In the event you Examine the sizes or material, you'll detect they'll all be somewhat distinct.
mei2 said: Would everyone know a Operating technique to batch translate these documents? The scripy deeplv4.py has stopped Operating for me --I feel a improve in DeepL UI has caused it.
My new x265 rips: mkv=best quality + 10bit(not blurry like most x265 rip) and mp4=really smaller but first rate high quality(the standard blurriness of x265)
temperature: A evaluate of the amount of randomness goes in to the transription and translaton course of action. This appears to be unintuitive at first, but performing a bunch of Original conditiosn and looking at what arrives out, and evaluating each of the probabilities and variety the best, makes far better results.
Description: PRED-750 English Sub – Someone finds by themselves by itself with their brother’s husband or wife, Miu, inside a closed room. Regardless of recognizing they shouldn’t, they offer in to their desires and initiate a Bodily experience with Miu.
Also the nsps-900 I am aware is peaceful new dunno if this a single have sub in chinese or other country translated on english.
There's however a big difficulty, which happens to be aliases. one actress can have 2 distinctive directories but beneath a unique alias. I'm sure javsub an answer I could take care of it, but I would like to determine how to query javstash graphql as they've got Just about all aliases.
The downloaded Sub was, I feel, developed from tough-coded Subbed video clips by Spikespen. Simply because Spikespen only Subbed clips there have been quite a few gaps in dialog when applied to the entire Motion picture. I utilised WhisperJAV0.seven to try and fill these gaps and I also attempted to scrub it up somewhat and re-interpreted many of the meaningless/ "lewd-significantly less" dialog.
You can find continue to a giant dilemma, that's aliases. 1 actress might have 2 different directories but beneath a different alias. I'm sure a solution I could fix it, but I want to determine how to query javstash graphql as they may have Virtually all aliases.