You can usually reduce the size of translation memories exported from Studio quite easily using Olifant. Edit > Attributes (Ctrl+T)
What's more, certain invalid characters are often also removed, simply by opening the TMX in Olifant and resaving it. This works particularly well in Xbench, which will almost always export a TMX that opens fine in all other tools.
With 16 GB of RAM, you can try as follows:
1. Make sure CafeTran runs on 64-bit Java and increase Java memory in MB to 8192 as described here: https://cafetran.freshdesk.com/solution/articles/6000160241-handling-large-projects-or-translation-memories
2. Try opening your Trados TM and when you see the error, click "Remove Invalid XML characters" button. It may take a while but there is a chance that this amount of RAM will be enough to parse and remove those characters from your huge TM.
I use 16 GB Ram (for virtual machines), so this should be enough. Is it maybe possible to export the TM from Studio without the invalid XML characters? I don't have any problems in Trados using this TM.
Your TM contains some invalid XML characters . CafeTran can remove them during the loading but in your case the TM is huge so the program might have not enough RAM memory to handle it. Please provide your system info by attaching the screenshot of CafeTran's Help > About panel. Also how much RAM memory do you have in your system? If you have enough RAM (8 GB at at least) and run CafeTran on 64-bit version of Java, the program might be able to handle such a big TM with invalid XML characters.
I also should mention that I exported the TM two times to TMX, so it shouldn't be a problem in this process.
I am very new to Cafetrans, and recently I wanted to work with a large translation memory (over 1,5 GB) that I imported from Trados.
Everytime I want to open it Espresso (v 2018), I get the error "reference "&#". After that the TM is empty.
I can open other imported TMs, only this one displays this error.