Site Li

how to tune the parameters in margin_inner_product_layer

1

 I think that the lambda is used as a tradeoff between the original softmax and A-softmax. But how could I adjust the base, power, lambda_min according to different task?...

Error when unzipping dataset

1

 I met this error when unzipping the provided dataset, could you please tell me how to solve this?

Does setting "SEMI_HARD = True" have a obvious improvement on the model? (converging speed, accuracy, etc.) I'm setting it to FALSE, and it seems that the loss couldn't decrease after...

Gradient Modification

1

``` gradients_vars = optimizer.compute_gradients(loss, LAYERS_WIEGHTS) grads = [grad for grad, var in gradients_vars] train_step = optimizer.apply_gradients(gradients_vars) ``` Hi, in this code, I see that the final gradients is **gradients_vars**, not...