Bill Ross

Here's how I've been predicting with multiple models per GPU without Triton: loading data, then forking processes, each loading keras and a model on its own. https://www.reddit.com/r/learnmachinelearning/comments/xg8ybf/optimizing_parallel_predictions_on_gpu/ It seems there...

It demonstrates that you can run multiple models predicting with their own data streams on GPU without Triton, which answers your question. Here are several procs using GPU 0, for...

Aha, I just copied the loss function from your mnist example, as I earlier copied a siamese mnist example, so I have model += GInftlyLayer( 'dfc1', w_regularizer=(c_l2, 1e-3), f_regularizer=(c_l2, f_reg),...

Great! One more problem skated over with a slight jump, then the next blocker: ValueError: You are passing a target array of shape (2048, 1) while using as loss 'categorical_crossentropy'....

to_categorical - so I'd get two weights, for yes/no? I'd rather get weights or distance than a 0/1 choice, so I can sort the matches to a photo. Rerere: The...

Nothing like 7K decisions on data to sharpen one's mind for programming. Starting from where I left off, I was going from a GINftly Dense of dense_size/2 directly to my...

I added a new view with just photos containing these keywords: http://phobrain.com/pr/home/view.html ``` Phob->Photogs/Subjects->Select (5K pics) flower tower bridge spire dome face faces sculpture graffiti downtown juxtapose_align juxtapose juxtapose_old_new juxtapose_pattern...

Revised: solved by changing my label dimension. My y_test.shape was (2048,) while p is (2048, 1). I had ``` read_file(pair_dir + '/' + pr_type + '.pos', pairs, labels, 1.) #...

> Do you use much data (& validation data)? 25K/25K pos/neg pairs to train; 2.8K/14K to test, which reflects the ratio in the wild. I could bump the width from...

> symmetric KL code Thanks! I set labels to 0.=match, 1.=no_match, switched to vertical pairs (smaller sample, but the one I'm interested in) and added Hue*Sat 48x48 to greyscale and...