@TripolarFandom26
Okay, I appreciate that you picked this up, but the translation is extremely rough. But I'm not the type to criticize without suggesting fixes.
Mainly, you seem to be having issues keeping track of who is saying what, as well as properly wrangling the MTL.
Here, look at this attempt that I made at translating another manga...
See what I did there? The page I'm working on side-by-side with a table arranged by panels (Koma 1, Koma 2, etc...) and then the elements in those panels (#1, #2, #3, etc..). Each element has their type classified (Box, Bubble, SFX, etc...). I not only have the original text from each element, but also the Romanji transliteration.
I also put notes in the sidebar (using Google Docs here) for various things like odd verbal tics, dialects, slang, special meanings - pretty much anything useful I need to remember about what I'm translating.
I set this up in Google Docs by setting the page to landscape mode with two columns for text, pasting the image for the page I'm working on in the first column and sizing it to fit lengthwise, and then putting the table on the second column.
Also note that I DO NOT break up two-page spreads. Instead, I stitch the two pages together into a single image before putting it in the Google Doc. This isn't hard to do. You can do it by just copying and pasting the pages into MS Paint, and then lining the two up so they fit together. Sometimes they seem like they don't line-up exactly. This is because a "whitespace" was inserted between the pages for the print-version of the manga. You'll just have to guess at how big the whitespace is and position the pages accordingly.
While it seems like a lot of effort, it will pay off when you're translating because you can quickly refer to what's going on in the panel and have that context ready at-hand. And in Japanese? Context is of crucial importance. It will also make it a lot easier to figure out who is talking about what and with whom, so you know what pronouns to use.
Speaking of context, you also seem to have issues of maintaining context through dialogues. There is a way of fixing this with MTL, fortunately.
You see, most MTL like Google Translate and DeepL will only use context on a line-by-line basis. Meaning, if you put a line break in the text, it will regard that new line without using the previous line for any context at all.
Only use line breaks when another person speaks, or when the scene changes, or when the topic changes. This will vastly improve the translation.
Lastly, Tools of the Trade.
I highly recommend Sugoi Toolkit. It's an all-in-one solution for OCR text reading and translation of Japanese. It has DeepL integrated into it, as well as JPasser, which is essential for sussing out the meaning of individual parts of a sentence.
Here's an example of an older version of the tool kit at-work:
You can download the latest version by contributing to MingShiba's Patreon (he's working on this thing full-time, so he needs the support!) or waiting until the 15th-16th of every month for the free link. He does it like this because it helps cut down on the number of bug reports. Otherwise, he gets absolutely inundated with reports for the same bug over and over again.
You'll find the free link when it's available here:
https://www.patreon.com/mingshiba/about
Hope this helps.