cookielee77

icon indicating an email [email protected]

icon company University of Washington icon location Seattle icon location Ph.D. candidate at University of Washington

Results 4 comments of cookielee77

"past" is not used in training

OK, I see. Thanks! I'm think the connection between GPT2 and TransformerXL. I feel like if the GPT2 uses "past" during training, then they are almost the same (except the...

cannot install language-check & language-check's version's too old, and their solution

Hi, Thanks for your valuable suggestions. I will update the README refer to your post. Unfortunately, It's difficult for me to update the dependencies as I already lost access to...

gather() got an unexpected keyword argument 'axis'

Got the same problem. Did you already solve it?

Possible Memory leak for RAM MEM

Nope. I didn't debug too many details. I'm not familiar with the author's dataloader type. Maybe the author could fix this at some time.