Skip to content

Instantly share code, notes, and snippets.

View ImTheCurse's full-sized avatar

Rani ImTheCurse

View GitHub Profile
@ImTheCurse
ImTheCurse / greedy_backprop.py
Last active April 28, 2026 21:51
custom Tensorflow implementation of a Greedy Layer-Wise backprop with global backprop on the MNIST dataset. utilizes online auxiliary training for deep supervision. skips waiting for a global loss signal to propagate through the entire network, each hidden block is explicitly trained to preserve local information.
# Disclamer: This program is a proof-of-concept, and I won't be liable for any damagaes caused by this program.
# by: github.com/imthecurse
import tensorflow as tf
class GreedyAutoencoderBlock(tf.keras.layers.Layer):
def __init__(self, units, use_bn=True, **kwargs):
super().__init__(**kwargs)
self.units = units
self.use_bn = use_bn
@ImTheCurse
ImTheCurse / https-file-server.go
Created December 29, 2023 11:10
simple static file server in go
package main
import (
"fmt"
"log"
"net/http"
"os"
)
func main(){