Model Weight Can't Be Loaded due to Git LFS load limit

[takeimhk]

when i do python main.py error appear

[HRNPH]

It seems that the error caused the checkpoint loading of the model which seems to be caused by an incorrect model weight make sure you have GITL-FS installed on your computer to ensure that when you git clone the weight will be downloaded correctly so you need to install git LFS then git clone again

[takeimhk]

it still the same, already install git lfs and error when doing git clone :

[HRNPH]

ohhh it's my fault because I'm out of git LFS quota you need to download the model via this link and placed it in the folder manually then https://huggingface.co/spaces/kouenYoung/anime-tts/resolve/main/1374_epochs.pth

[HRNPH]

The new version of the code now fixed the model download problem using git lfs to use wget instead you should try to delete the folder and clone it again refer

[HRNPH]

Thx for noticing the problem please notice me if this issue was fixed so I can close this issue!

[takeimhk]

awesomeeee, it works.

you are awsome! Keeper going!

so i am just asking, what database the the AI use? the AI answer my question with veeery long text and repeating, while i use different question or text

cheers!

[HRNPH]

currently, the model is the pygmalion6B which was trained on something like RPG conversation or novel stuff conversation there's no database that model relies on to answer except the trained weight(knowledge) on the datasets that she has seen the context of how the model will reply will pretty much depend on your history of conversation (currently capped at 10 lines to save memory in the test so the result is she'll not remember the old conversations at all) I'l increase the memory to 30-40 lines, It should give a better answer further information on how she's forgetful can be read here #22