Hints
LM-C15-C-001
Use one recurrent step per token.
LM-C15-C-002
Fixed hidden size names the vector length.
LM-C15-C-003
Count the positions named.
LM-C15-C-004
Find the largest weight first.
LM-C15-C-005
Recall softmax over attention scores.
Use one recurrent step per token.
Fixed hidden size names the vector length.
Count the positions named.
Find the largest weight first.
Recall softmax over attention scores.