i choose a 100 epochs is that overkill ?its moving at about 5-6 minutes an epoch --i don’t know if ill have the tpu to get through a hundred epochs - i think ill pass on the ssim scores after a hundred as my early looks are blurry but its super early i think i see faces - :- is there a good egpu that works with macbook m2 ?
Hello Paul,
your explanation of your issue lacks information what kind of model you are working on. Kindly give a brief detail about your model without sharing codes like your batch size or steps per epoch if any used, model compile parameters.
If your each epoch time is taking 5-6 minutes it can be surely tpu related but it can also be model algorithm related.
So to know if it is not model related, one needs to have a bit understanding about your model, to give you hints on improving the training time.
P.S. note even for me the last two assignment of course 4 took pretty long time like 2 hours as in whole but I created the simplest model by referring to different threads in discourse community (sharing one link where I had asked for help)
Also if you feel it is GPU related, refer the below link
Let me know if your issue gets resolved.
Regards
DP
well my work is very close but not identical to the ungraded lab - i went for the 3 layers and to reference batch_3.shape as suggested in criteria and used ‘mse’ as suggested but i usually use the t4 gpu -when i dont get the t4 its molasses the tpu is second best- most of the time for my experience - the model may be a little complex -but ive had larger more intricate models go faster so i dont think my code is terrible this time - though i’m sure u could streamline it - it runs - and it didn’t throw a traceback and was beginning to show new images in the glimpse area - just super slow - my cat closed my laptop mid train so that didnt help matters i had hours into that and i only get so muh tpu per night - - i also put it out there wondering if anyone had any successful experiences running an egpu with a 2022 macbook pro m2 id like to upgrade soon maybe i should have said that last part in general forum separately but this experience rekindled the idea - im hearing gans next week is going to be even slower
1 Like
Did you try the fallback runtime link I provided in the previous comment.??
if u mean the t4 gpu yes
No I talking about this 
just trying fallback now -it just occurred to me to reread instructions - sorry my mom is not well and i stayed up looking after her so i got zero sleep again - so not thinking as clearly as im used to-ill let u know if it saves a few minutes per epoch
oh much much faster thanks ill up the epochs now
this was a really interesting assignment if u tske away the runtime snafu but thats my fault instead of really digging in and making sure i understood that fall back runtime advisement i- made it much tougher on myself than this had to to be - when things work its really fun and feels promising like theres hope to a be actually be a useful member of a ml team
1 Like