radi-cho

const trainSave = async (model, querySnapshot) => {
  if (!querySnapshot.docs.length) return false;

  const xs_data = querySnapshot.docs.map(doc => fitData(doc.data().text));
  const ys_data = querySnapshot.docs.map(
    doc => (doc.data().y === "positive" ? [1] : [0])
  );

  const xs = tf.tensor2d(xs_data);
  const ys = tf.tensor2d(ys_data);

radi-cho

// TFjs and gcs are using a lot of memory, so increasing it will speed up the functions
exports.train = functions
  .runWith({ memory: "2GB" })
  .https.onRequest(async (request, response) => {
    const existJSON = await bucket.file("model.json").exists().then(ex => ex[0]);
    const existBIN = await bucket.file("weights.bin").exists().then(ex => ex[0]);

    if (!existJSON || !existBIN) {
      const model = tf.sequential();
      model.add(tf.layers.dense({ units: 2, inputShape: [vocabulary.length] }));

radi-cho

-keep class com.google.unity.** {
   *;
}
-keep public class com.google.android.gms.ads.**{
   public *;
}
-keep public class com.google.ads.**{
   public *;
}
-keepattributes *Annotation*

radi-cho

JSON.safeStringify = (obj, indent = 2) => {
  let cache = [];
  const retVal = JSON.stringify(
    obj,
    (key, value) =>
      typeof value === "object" && value !== null
        ? cache.includes(value)
          ? undefined // Duplicate reference found, discard key
          : cache.push(value) && value // Store value in our collection
        : value,

radi-cho

temperature = 10
dist = tfp.distributions.RelaxedOneHotCategorical(temperature, probs=p)
dist.sample()

<tf.Tensor: shape=(3,), dtype=float32, numpy=array([0.31916314, 0.34642866, 0.33440822], dtype=float32)>

radi-cho

temperature = 0.01
dist = tfp.distributions.RelaxedOneHotCategorical(temperature, probs=p)
dist.sample()

<tf.Tensor: shape=(3,), dtype=float32, numpy=array([0., 1., 0.], dtype=float32)>

radi-cho

tf.one_hot(tf.argmax(p), depth = len(p))

<tf.Tensor: shape=(3,), dtype=float32, numpy=array([0., 1., 0.], dtype=float32)>

radi-cho

self.normalize1 = layers.LayerNormalization(epsilon=1e-6)
self.normalize2 = layers.LayerNormalization(epsilon=1e-6)

self.channel_projection1 = keras.Sequential(
    [
        layers.Dense(units=embedding_dim * 2),
        layers.ReLU(),
        layers.Dropout(rate=dropout_rate),
    ]
)

radi-cho

import tensorflow as tf
from tensorflow import keras
from tensorflow.keras import layers

class gMLPLayer(layers.Layer):
    def __init__(self, num_patches, embedding_dim, dropout_rate, *args, **kwargs):
        super(gMLPLayer, self).__init__(*args, **kwargs)

radi-cho

def spatial_gating_unit(self, x):
    # u and v shape: [batch_size, num_patchs, embedding_dim]
    u, v = tf.split(x, num_or_size_splits=2, axis=2)
    v = self.normalize2(v)