From c4b9188077d75082ab2eaf725a877ecee8789f74 Mon Sep 17 00:00:00 2001 From: Morten Hjorth-Jensen Date: Mon, 22 Nov 2021 11:18:50 +0100 Subject: [PATCH] small change --- doc/pub/week40/ipynb/week40.ipynb | 45 ++++++++++++++++++++++++++++--- 1 file changed, 41 insertions(+), 4 deletions(-) diff --git a/doc/pub/week40/ipynb/week40.ipynb b/doc/pub/week40/ipynb/week40.ipynb index 4f2097cc1..3d2222d2b 100644 --- a/doc/pub/week40/ipynb/week40.ipynb +++ b/doc/pub/week40/ipynb/week40.ipynb @@ -1579,10 +1579,46 @@ }, { "cell_type": "code", - "execution_count": 18, + "execution_count": 2, "id": "4eb94f4f", "metadata": {}, - "outputs": [], + "outputs": [ + { + "name": "stdout", + "output_type": "stream", + "text": [ + "Own inversion\n", + "[[4.11991617]\n", + " [2.88756611]]\n", + "Eigenvalues of Hessian Matrix:[0.32686212 4.18967814]\n", + "(2, 1)\n", + "theta from own gd\n", + "[[4.11991617]\n", + " [2.88756611]]\n" + ] + }, + { + "data": { + "image/png": "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\n", + "text/plain": [ + "
" + ] + }, + "metadata": { + "needs_background": "light" + }, + "output_type": "display_data" + }, + { + "name": "stdout", + "output_type": "stream", + "text": [ + "theta from own sdg\n", + "[[4.11069364]\n", + " [2.90089392]]\n" + ] + } + ], "source": [ "# Using Autograd to calculate gradients using SGD\n", "# OLS example\n", @@ -1620,6 +1656,7 @@ "for iter in range(Niterations):\n", " gradients = (1.0/n)*training_gradient(y, X, theta)\n", " theta -= eta*gradients\n", + "print(gradients.shape)\n", "print(\"theta from own gd\")\n", "print(theta)\n", "\n", @@ -1652,7 +1689,7 @@ " random_index = M*np.random.randint(m)\n", " xi = X[random_index:random_index+M]\n", " yi = y[random_index:random_index+M]\n", - " gradients = (2.0/M)*training_gradient(yi, xi, theta)\n", + " gradients = (1.0/M)*training_gradient(yi, xi, theta)\n", " eta = learning_schedule(epoch*m+i)\n", " theta = theta - eta*gradients\n", "print(\"theta from own sdg\")\n", @@ -3396,7 +3433,7 @@ "name": "python", "nbconvert_exporter": "python", "pygments_lexer": "ipython3", - "version": "3.9.7" + "version": "3.8.12" } }, "nbformat": 4,