move old code from notebook

landoskape · Apr 17, 2024 · 957eaf6 · 957eaf6
1 parent dc7df97
commit 957eaf6
Show file tree

Hide file tree

Showing 2 changed files with 537 additions and 60 deletions.
diff --git a/dominoes.ipynb b/dominoes.ipynb
@@ -45,15 +45,13 @@
    "outputs": [],
    "source": [
     "# TODO For refactoring\n",
-    "# check reward sequencer function\n",
-    "\n",
     "# start working on TSP dataset\n",
     "# start working on supervised learning child of parent dataset class (and make dominoeDataset an child of that also!)"
    ]
   },
   {
    "cell_type": "code",
-   "execution_count": 3,
+   "execution_count": 1,
    "id": "8404c54b",
    "metadata": {},
    "outputs": [],
@@ -70,80 +68,66 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 7,
+   "execution_count": 20,
    "id": "94b81df6",
    "metadata": {},
    "outputs": [
     {
-     "data": {
-      "text/plain": [
-       "tensor([[3, 7, 5, 6, 8, 9, 9, 9, 9],\n",
-       "        [6, 3, 4, 0, 8, 9, 9, 9, 9],\n",
-       "        [8, 9, 9, 9, 9, 9, 9, 9, 9],\n",
-       "        [0, 1, 6, 5, 8, 9, 9, 9, 9]])"
-      ]
-     },
-     "execution_count": 7,
-     "metadata": {},
-     "output_type": "execute_result"
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "tensor([[ 1.,  0.,  0.,  0.,  0.,  0.,  0.,  0.,  0.],\n",
+      "        [ 1.,  0.,  0.,  0.,  0.,  0.,  0.,  0.,  0.],\n",
+      "        [12., 10.,  4., 10., 13.,  1.,  0.,  0.,  0.],\n",
+      "        [ 8.,  1.,  0.,  0.,  0.,  0.,  0.,  0.,  0.]])\n"
+     ]
     }
    ],
    "source": [
     "highest_dominoe = 9\n",
     "dataset = DominoeDataset(\"sequencer\", highest_dominoe, hand_size=8, return_target=True)\n",
     "\n",
-    "batch = dataset.generate_batch(train=False, batch_size=4)\n",
+    "batch = dataset.generate_batch(train=False, batch_size=4, value_method=\"dominoe\")\n",
     "dominoes = dataset.get_dominoe_set(train=False)\n",
     "\n",
-    "ib = 0\n",
-    "hand = dominoes[batch[\"selection\"][ib]]\n",
     "target_as_choice = batch[\"target\"].clone()\n",
-    "target_as_choice[target_as_choice==-1] = dataset.prms[\"hand_size\"] + 1\n",
-    "# reward = dataset._measurereward_sequencer(target_as_choice, batch)\n",
-    "target_as_choice"
-   ]
-  },
-  {
-   "cell_type": "code",
-   "execution_count": 39,
-   "metadata": {},
-   "outputs": [
-    {
-     "data": {
-      "text/plain": [
-       "torch.int64"
-      ]
-     },
-     "execution_count": 39,
-     "metadata": {},
-     "output_type": "execute_result"
-    }
-   ],
-   "source": [
-    "dominoes.dtype"
+    "target_as_choice[target_as_choice==-1] = dataset.prms[\"hand_size\"]\n",
+    "\n",
+    "reward, direction = dataset._measurereward_sequencer(target_as_choice, batch, return_direction=True)\n",
+    "print(reward)"
    ]
   },
   {
    "cell_type": "code",
-   "execution_count": 55,
-   "id": "b7e87b30",
+   "execution_count": 21,
+   "id": "80eea3a8",
    "metadata": {},
    "outputs": [
     {
-     "data": {
-      "text/plain": [
-       "False"
-      ]
-     },
-     "execution_count": 55,
-     "metadata": {},
-     "output_type": "execute_result"
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "tensor(3)\n",
+      "tensor([3., 8.]) forward value = 12.0\n",
+      "tensor([1., 8.]) reverse value = 10.0\n",
+      "tensor([1., 2.]) forward value = 4.0\n",
+      "tensor([2., 7.]) forward value = 10.0\n",
+      "tensor([5., 7.]) reverse value = 13.0\n",
+      "tensor([-1., -1.]) reverse value = -1.0\n",
+      "tensor([-1., -1.]) reverse value = -1.0\n",
+      "tensor([-1., -1.]) reverse value = -1.0\n",
+      "tensor([-1., -1.]) reverse value = -1.0\n"
+     ]
     }
    ],
    "source": [
-    "ss = \"length\"\n",
+    "ib = 2\n",
+    "hand = dominoes[batch[\"selection\"][ib]]\n",
+    "hand_null = torch.cat([hand, -torch.ones((1, 2))], dim=0)\n",
     "\n",
-    "not (ss == \"dominoe\" or ss == \"length\")"
+    "print(batch[\"available\"][ib])\n",
+    "for c, d in zip(target_as_choice[ib], direction[ib]):\n",
+    "    print(hand_null[c], \"forward\" if d==0 else \"reverse\", f\"value = {torch.sum(hand_null[c])+1}\")"
    ]
   },
   {
@@ -167,13 +151,6 @@
    "outputs": [],
    "source": []
   },
-  {
-   "cell_type": "code",
-   "execution_count": null,
-   "metadata": {},
-   "outputs": [],
-   "source": []
-  },
   {
    "cell_type": "code",
    "execution_count": null,