You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Hello.I think i don't know about the function add_window.There's some code that makes me wonder,“idx < len(doc) - window_size”. Actually,I know it's a critical condition,It makes the last window retain size.But I think,in the bottom of the code,there is " item.metadata[f"{window_idx_name}_lower_bound"] = min(window)
item.metadata[f"{window_idx_name}_upper_bound"] = max(window)",
we use the min and mix func to deal the bound.So i guess the condition “idx < len(doc) - window_size” does not seem to make much sense.
The text was updated successfully, but these errors were encountered:
The main reason I added this condition is that when the number of elements at the end of doc list is not enough to create a new large chunk index, I will combine those last few elements into the previous large chunk index. You can reference this graph below:
Hello.I think i don't know about the function add_window.There's some code that makes me wonder,“idx < len(doc) - window_size”. Actually,I know it's a critical condition,It makes the last window retain size.But I think,in the bottom of the code,there is " item.metadata[f"{window_idx_name}_lower_bound"] = min(window)
item.metadata[f"{window_idx_name}_upper_bound"] = max(window)",
we use the min and mix func to deal the bound.So i guess the condition “idx < len(doc) - window_size” does not seem to make much sense.
The text was updated successfully, but these errors were encountered: