retrain Binary-classification

linjing-lab · Oct 27, 2023 · 2b84694 · 2b84694
1 parent b998e35
commit 2b84694
Show file tree

Hide file tree

Showing 2 changed files with 51 additions and 29 deletions.
diff --git a/released_box/models/credit.ckpt b/released_box/models/credit.ckpt
diff --git a/released_box/tests/Binary-classification Task.ipynb b/released_box/tests/Binary-classification Task.ipynb
@@ -341,7 +341,7 @@
     {
      "data": {
       "text/plain": [
-       "((30000, 23), (30000,))"
+       "'1: 6636, 0: 23364'"
       ]
      },
      "execution_count": 5,
@@ -350,6 +350,27 @@
     }
    ],
    "source": [
+    "'1: {}, 0: {}'.format(numpy.sum(labels == 1), numpy.sum(labels == 0)) # nums of 1 and 0"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 6,
+   "metadata": {},
+   "outputs": [
+    {
+     "data": {
+      "text/plain": [
+       "((30000, 23), (30000,))"
+      ]
+     },
+     "execution_count": 6,
+     "metadata": {},
+     "output_type": "execute_result"
+    }
+   ],
+   "source": [
+    "# dataset are imbalanced by labels\n",
     "features.shape, labels.shape"
    ]
   },
@@ -371,7 +392,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 6,
+   "execution_count": 7,
    "metadata": {},
    "outputs": [
     {
@@ -406,7 +427,7 @@
        "             ('device', device(type='cuda'))])"
       ]
      },
-     "execution_count": 6,
+     "execution_count": 7,
      "metadata": {},
      "output_type": "execute_result"
     }
@@ -429,7 +450,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 7,
+   "execution_count": 8,
    "metadata": {},
    "outputs": [],
    "source": [
@@ -446,23 +467,24 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 8,
+   "execution_count": 9,
    "metadata": {},
    "outputs": [
     {
      "name": "stdout",
      "output_type": "stream",
      "text": [
-      "Epoch [1/2], Step [100/3000], Training Loss: 828.4427, Validation Loss: 1241.4515\n",
-      "Epoch [1/2], Step [200/3000], Training Loss: 94.2689, Validation Loss: 64.1411\n",
-      "Epoch [1/2], Step [300/3000], Training Loss: 0.3855, Validation Loss: 14.7295\n",
-      "Epoch [1/2], Step [400/3000], Training Loss: 0.3561, Validation Loss: 4.1548\n",
-      "Epoch [1/2], Step [500/3000], Training Loss: 2.6463, Validation Loss: 4.9783\n",
-      "Epoch [1/2], Step [600/3000], Training Loss: 0.4240, Validation Loss: 3.0171\n",
-      "Epoch [1/2], Step [700/3000], Training Loss: 3.8731, Validation Loss: 1.5793\n",
-      "Epoch [1/2], Step [800/3000], Training Loss: 0.2598, Validation Loss: 0.9340\n",
-      "Epoch [1/2], Step [900/3000], Training Loss: 0.5623, Validation Loss: 0.7461\n",
-      "Epoch [1/2], Step [1000/3000], Training Loss: 0.7324, Validation Loss: 0.6750\n",
+      "Epoch [1/2], Step [100/3000], Training Loss: 0.0000, Validation Loss: 751.9165\n",
+      "Epoch [1/2], Step [200/3000], Training Loss: 151.6677, Validation Loss: 147.1012\n",
+      "Epoch [1/2], Step [300/3000], Training Loss: 11.5891, Validation Loss: 11.9286\n",
+      "Epoch [1/2], Step [400/3000], Training Loss: 102.6795, Validation Loss: 6.6973\n",
+      "Epoch [1/2], Step [500/3000], Training Loss: 0.2132, Validation Loss: 4.4743\n",
+      "Epoch [1/2], Step [600/3000], Training Loss: 3.4273, Validation Loss: 3.1016\n",
+      "Epoch [1/2], Step [700/3000], Training Loss: 0.6728, Validation Loss: 1.4565\n",
+      "Epoch [1/2], Step [800/3000], Training Loss: 0.3786, Validation Loss: 8.3835\n",
+      "Epoch [1/2], Step [900/3000], Training Loss: 0.4249, Validation Loss: 1.1698\n",
+      "Epoch [1/2], Step [1000/3000], Training Loss: 0.5671, Validation Loss: 0.7559\n",
+      "Epoch [1/2], Step [1100/3000], Training Loss: 0.7306, Validation Loss: 0.9674\n",
       "Process stop at epoch [1/2] with patience 10 within tolerance 0.001\n"
      ]
     }
@@ -481,14 +503,14 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 9,
+   "execution_count": 10,
    "metadata": {},
    "outputs": [
     {
      "name": "stdout",
      "output_type": "stream",
      "text": [
-      "loss of Box on the 3000 test dataset: 1.061116337776184.\n"
+      "loss of Box on the 3000 test dataset: 1.4942820072174072.\n"
      ]
     },
     {
@@ -500,13 +522,13 @@
        "             ('column', ('label name', ('true numbers', 'total numbers'))),\n",
        "             ('labels', {0: [2282, 3000], 1: [2282, 3000]}),\n",
        "             ('loss',\n",
-       "              {'train': 0.5675961375236511,\n",
-       "               'val': 0.6755267381668091,\n",
-       "               'test': 1.061116337776184}),\n",
+       "              {'train': 0.2478877156972885,\n",
+       "               'val': 0.6973171234130859,\n",
+       "               'test': 1.4942820072174072}),\n",
        "             ('sorted', [(0, [2282, 3000]), (1, [2282, 3000])])])"
       ]
      },
-     "execution_count": 9,
+     "execution_count": 10,
      "metadata": {},
      "output_type": "execute_result"
     }
@@ -525,7 +547,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 10,
+   "execution_count": 11,
    "metadata": {},
    "outputs": [],
    "source": [
@@ -545,7 +567,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 11,
+   "execution_count": 12,
    "metadata": {},
    "outputs": [],
    "source": [
@@ -562,14 +584,14 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 12,
+   "execution_count": 13,
    "metadata": {},
    "outputs": [
     {
      "name": "stdout",
      "output_type": "stream",
      "text": [
-      "loss of Box on the 3000 test dataset: 1.0611168146133423.\n"
+      "loss of Box on the 3000 test dataset: 1.494282841682434.\n"
      ]
     },
     {
@@ -581,13 +603,13 @@
        "             ('column', ('label name', ('true numbers', 'total numbers'))),\n",
        "             ('labels', {0: [2282, 3000], 1: [2282, 3000]}),\n",
        "             ('loss',\n",
-       "              {'train': 0.5675961375236511,\n",
-       "               'val': 0.6755267381668091,\n",
-       "               'test': 1.0611168146133423}),\n",
+       "              {'train': 0.2478877156972885,\n",
+       "               'val': 0.6973171234130859,\n",
+       "               'test': 1.494282841682434}),\n",
        "             ('sorted', [(0, [2282, 3000]), (1, [2282, 3000])])])"
       ]
      },
-     "execution_count": 12,
+     "execution_count": 13,
      "metadata": {},
      "output_type": "execute_result"
     }