there is one place confused me. when you set seed on Sin function: float offset = fit01(rand( @ptnum ) * chi( "seed" ) ,0,3.14*2); do you mean float offset = fit01(rand( @ptnum * chi( "seed" ) ),0,3.14*2); becuase rand(*) * 4 is over than 1 ; that conflict with fit01. did you do it on purpose ?
I tried to use rand( @ptnum * chi( "seed" ) ) to set the seed. and I can't train the model corretly which confused me even more. can you please try that out?
okay I figured it out. there is two misstake here first is the rand function. and second is traced_cell = torch.jit.trace(seq,test_input) it should be" input " instead of " test_input". other than this it works as a charm. Thanks a lot for sharing the tutorial. great work ! I can't wait to watch the next one.
@@noodles5667 the test_input one is weird, if the issue was that you got a nan loss, you just need to run the training again. Thah kind of stuff happens sometimes with both test_input and input. Thanks, for the kind words, appreciate it !
there is one place confused me. when you set seed on Sin function: float offset = fit01(rand( @ptnum ) * chi( "seed" ) ,0,3.14*2); do you mean float offset = fit01(rand( @ptnum * chi( "seed" ) ),0,3.14*2); becuase rand(*) * 4 is over than 1 ; that conflict with fit01. did you do it on purpose ?
I tried to use rand( @ptnum * chi( "seed" ) ) to set the seed. and I can't train the model corretly which confused me even more. can you please try that out?
okay I figured it out. there is two misstake here first is the rand function. and second is traced_cell = torch.jit.trace(seq,test_input) it should be" input " instead of " test_input". other than this it works as a charm. Thanks a lot for sharing the tutorial. great work ! I can't wait to watch the next one.
@@noodles5667 Yes, thank you ! I meant to put the seed parm inside the rand like in the warmup. Pinning for visibility !
@@noodles5667 the test_input one is weird, if the issue was that you got a nan loss, you just need to run the training again. Thah kind of stuff happens sometimes with both test_input and input. Thanks, for the kind words, appreciate it !
Another great tutorial, its really fun to follow you.
Already looking forward to the next topic, keep it up :)
Amazing look 😍👀
👍Nice video, looks amazing. 😊
do u have an easier example to start with?
ML wise ? ua-cam.com/video/WNEEokEq-Fg/v-deo.html