Why do gpt-2 model response so different in computer run and chaquopy android run?
#56
by
gamingjam60
- opened
When I run gpt-2 124M model in python interpreter of windows machine it's response is more relevant than when I run the same model in Chaquopy android it is responding more randomly which doesn't have any meaning.Why? In case of Chaquopy android tokenizer version is 4.30.0 and I am using local data folder for gpt-2 model which includes all binary along with other file
