Removed dependency from residual flow repository

Steven-meta · Jul 26, 2022 · 314984d · 314984d
1 parent ff33d12
commit 314984d
Show file tree

Hide file tree

Showing 6 changed files with 807 additions and 31 deletions.
diff --git a/example/residual.ipynb b/example/residual.ipynb
@@ -2,15 +2,23 @@
  "cells": [
   {
    "cell_type": "markdown",
-   "metadata": {},
+   "metadata": {
+    "pycharm": {
+     "name": "#%% md\n"
+    }
+   },
    "source": [
     "# Residual Flow"
    ]
   },
   {
    "cell_type": "code",
    "execution_count": null,
-   "metadata": {},
+   "metadata": {
+    "pycharm": {
+     "name": "#%%\n"
+    }
+   },
    "outputs": [],
    "source": [
     "# Import required packages\n",
@@ -29,6 +37,9 @@
    "cell_type": "code",
    "execution_count": null,
    "metadata": {
+    "pycharm": {
+     "name": "#%%\n"
+    },
     "scrolled": false
    },
    "outputs": [],
@@ -71,6 +82,9 @@
    "cell_type": "code",
    "execution_count": null,
    "metadata": {
+    "pycharm": {
+     "name": "#%%\n"
+    },
     "scrolled": false
    },
    "outputs": [],
@@ -103,19 +117,22 @@
    "cell_type": "code",
    "execution_count": null,
    "metadata": {
+    "pycharm": {
+     "name": "#%%\n"
+    },
     "scrolled": false
    },
    "outputs": [],
    "source": [
     "# Train model\n",
-    "max_iter = 10000\n",
+    "max_iter = 20000\n",
     "num_samples = 2 ** 9\n",
     "show_iter = 500\n",
     "\n",
     "\n",
     "loss_hist = np.array([])\n",
     "\n",
-    "optimizer = torch.optim.Adam(nfm.parameters(), lr=1e-3, weight_decay=1e-5)\n",
+    "optimizer = torch.optim.Adam(nfm.parameters(), lr=3e-4, weight_decay=1e-5)\n",
     "for it in tqdm(range(max_iter)):\n",
     "    optimizer.zero_grad()\n",
     "    \n",
@@ -132,7 +149,7 @@
     "        optimizer.step()\n",
     "    \n",
     "    # Make layers Lipschitz continuous\n",
-    "    nf.utils.update_lipschitz(nfm, 5)\n",
+    "    nf.utils.update_lipschitz(nfm, 50)\n",
     "    \n",
     "    # Log loss\n",
     "    loss_hist = np.append(loss_hist, loss.to('cpu').data.numpy())\n",
@@ -160,7 +177,11 @@
   {
    "cell_type": "code",
    "execution_count": null,
-   "metadata": {},
+   "metadata": {
+    "pycharm": {
+     "name": "#%%\n"
+    }
+   },
    "outputs": [],
    "source": [
     "# Plot learned posterior distribution\n",
@@ -193,7 +214,7 @@
    "name": "python",
    "nbconvert_exporter": "python",
    "pygments_lexer": "ipython3",
-   "version": "3.8.11"
+   "version": "3.7.11"
   }
  },
  "nbformat": 4,