Nguyen Nguyen Anh
Nguyen Nguyen Anh
Getting worsen results...  I will try https://github.com/l4rz/scaling-up-stylegan2 Reduce lrate, gamma=6, eliminate style mixing... Let see. Steve
I think you must triple the capacity of network to learn such perplexed dataset. Same here. Never worked for complex dataset. The GAN capability (no meter NVIDIA or Google) is...
Try this https://www.programmersought.com/article/60001511310/ Both binary, multi-class and multi-label. It seems to work for me. Steve
Why can't FastBPE owner just say that they are not implementing this for Windows !!! Annoying. Steve
Found it. Use tf.keras.applications.resnet50.preprocess_input Steve
Faces the same issue using this as backbone for Tensorflow Object Detection, using the same config file of ssd_mobilenet (changes size to 512 512 though). Anything I should have done...
You should combine your own custom dataset with UTK, CrowdHUman as well...
Use Spectral Normalization on top of CONV2D of Discriminator will stabilize the training greatly. Also, pay attention to kernel_initializer (glorot_normal etc...)
This is TOO complicated even to get a demo up and running. I gave up !
> Llama 3 has improved tokenizer based on Tiktoken v/s Llama 2 which was based on Sentencepiece. Llama 3 tokenizer expands the vocabulary size to 128k (from 32k tokens in...