search

jackielinxiao / TPM

Codes for TPM, a tree based model for watch time prediction
18 stars 1 forks source link

Is there an error in calculating the standard deviation? #2

Open enjlife opened 4 months ago

enjlife commented 4 months ago

encoded_playtime = tf.reduce_sum(encoded_playtime*encoded_prob,axis=-1,keepdims=True) Encoded_playtime is replaced with E(X)

e_x2 = tf.reduce_sum(tf.square(encoded_playtime)*encoded_prob, axis=-1,keepdims=True) Should encoded_playtime be replaced with temp_encoded_playtime?

snailma0229 commented 4 months ago

i think there is an error in calculating process too