The default batch size is 16 which is the root cause of taking too much time to

Thanks for informing. I guess I have tried different batch sizes

Yep, it depends on GPU performance. After digging into the source code,

suggestions and questions on training the data about improving-rnn-recommendation-model HOT 2 OPEN

kwonmha commented on June 7, 2024

suggestions and questions on training the data

from improving-rnn-recommendation-model.

Comments (2)

kwonmha commented on June 7, 2024

Thanks for informing. I guess I have tried different batch sizes but I'm not sure how large I have tried. Batch size depends on the memory size of GPU. Nevertheless, It's worth knowing.
I think you can get loss value which is written as cost in my original code.

cost = self.model.train_on_batch(batch[0], batch[1])

Improving-RNN-recommendation-model/neural_networks/rnn_base.py

Line 217 in f63ba48

cost = self.model.train_on_batch(batch[0], batch[1])

And I can't imagine how you can get accuracy but if you can have loss and accuracy after performing each iteration, it seems easy to save weight file by adding simple condition.

if acc > 0.99 or loss < 0.04: 
    SAVE_MODEL

Related code is here :

Improving-RNN-recommendation-model/neural_networks/rnn_base.py

Line 260 in f63ba48

elif autosave == 'Best':

Maybe modifying codes in that block would work.

from improving-rnn-recommendation-model.

shwangdev commented on June 7, 2024

Yep, it depends on GPU performance.
After digging into the source code, I found there was 2 parameter from the command line to force save a weight file in iterations. --progress and --min-iter. I have set the two parameters as 500. the weight file can be generated out quickly.

from improving-rnn-recommendation-model.

Recommend Projects