phsamuel

Something seems to be wrong with torch.optim.SGD. Maybe I don't really understand the usage. I tried to replace 

SGDMom_points = train_curve(lambda params: SGDMomentum(params, lr=10, momentum=0.9))

with 

SGDMom_points = train_curve(lambda params: torch.optim.SGD(params, lr=10, momentum=0.9))

on https://pytorch-lightning.readthedocs.io/en/stable/notebooks/course_UvA-DL/03-initialization-and-optimization.html

phsamuel

import sys

from PyQt5.QtWidgets import QApplication, QDialog, QMainWindow, QMessageBox
from PyQt5.uic import loadUi
from PyQt5.QtCore import pyqtSlot as pyQtSlot
from PyQt5.QtCore import QTimer,QDateTime,Qt,QRect
from PyQt5.QtGui import QPainter, QPen, QFont
from PyQt5.QtWidgets import QFileDialog

from time import sleep

phsamuel

phsamuel

{
 "cells": [
  {
   "cell_type": "markdown",
   "id": "ba091f7e",
   "metadata": {},
   "source": [
    "# CS 231n example"
   ]
  },

phsamuel

{
 "cells": [
  {
   "cell_type": "markdown",
   "id": "ae9db9ff",
   "metadata": {},
   "source": [
    "# Q1.a.  (10 points) create an network to approximate sin(x)\n",
    "\n",
    "In this problem, you will try to build a three-layer fully connected network to approximate the sin function using PyTorch. There are 100 hidden units in each layer and please use ReLU activation after the first two layers and no activation after the last layer. You should implement your network model as a subclass of [torch.nn.Module](https://pytorch.org/docs/stable/generated/torch.nn.Module.html) "

phsamuel

phsamuel

phsamuel

phsamuel

phsamuel

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Detect circles"
   ]