Can I obtain finer grained information when encoding images? For example, the attention score for each patch?
#4 opened about 1 hour ago
by
1728-zxy
why do we need + "<|endoftext|>" when processing query and doccument
2
#3 opened about 2 months ago
by
Mengyao00
Finetuning for a different language
6
#2 opened 2 months ago
by
jdchawla
Any explanation for this bug and what it causes ?
2
#1 opened 3 months ago
by
manu