Model training image captioning

#1

sir please tell me what i suppose to fill in number_pics_per_batch
it shows error…
http://localhost:8889/notebooks/Image_Captioning_Project.ipynb

its really important please tell asap…?

#2

Hey @Ricoman, you could use any value like 3,4 or 5. Also for future reference take help from this link

Hope this resolved your doubt.
Plz mark it as resolved in my doubts section. :blush:

#3

sir i m getting loss of 0.0015

#4

and then in second epoch it goes like 6.5873e-04…

#5

i guess there must be something wrong sir…right…?

#6

steps = len(train_descriptions)//3…

#7

i used this as my steps…ye sahi hai kya sir…?

#8

Hey @Ricoman, ha toh 6.5873e-04…, stands for 0.000658, loss decrease hi ho rha h …
and use
epochs = 20
batch_size = 3
steps = len(train_descriptions)//number_pics_per_batch

#9

and number_images_per_batch is equal 3 right…?

#10

sir i got it now…i should’ve taken 6 images per batch…but mistakenly i had take only 3 picture per batch…does it make any difference on the quality of results…mere 12 epochs train ho chuke hai kya mai dobara 6 pictures ke saath start karun or is it fine with 3 pictures…? this is my last and final doubt sir please clear…?

#11

Hey @Ricoman, no it doesn’t matter at all, you could continue model training, 6 or 3 doesn’t make much differece, the only difference will be of time computation.

Hope this resolved your doubt.
Plz mark it as resolved in my doubts section. :blush:

#12

sir…i m getting all my captions incorrect and i really feeling too bad …i followed each and every step as taught in the videos…i spent about 5 days on this sir…if you can just look on to my proct notebook once and try to help me with this then it would be really helpful from your side sir…please sir this is my first project…please help me…?

#13

http://localhost:8888/notebooks/Image_Captioning_Project.ipynb …This is my notebook please take a look once where i did mistake…?

#14

sir my doubt is clear…i’ve done a small mistake…

#15

Hey @Ricoman, okay than that’s great.

Happy Learning :sunny:
Plz mark the doubt as resolved. :blush: