Fix RNN for next word prediction

orifmilod · May 13, 2024 · 1f9f72c · 1f9f72c
1 parent 49bffbd
commit 1f9f72c
Show file tree

Hide file tree

Showing 2 changed files with 242 additions and 97 deletions.
diff --git a/example/language_model/pytorch/cbow.ipynb b/example/language_model/pytorch/cbow.ipynb
@@ -116,6 +116,73 @@
     "print(f'Context: {context}\\n')\n",
     "print(f'Prediction: {index_to_word[torch.argmax(a[0]).item()]}')"
    ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 3,
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "cpu\n",
+      "mps:0\n",
+      "Total time with cpu (10000): 0.0005950000000005673\n",
+      "Total time with cpu (100000000): 0.0002990000000000492\n",
+      "available\n",
+      "Total time with gpu (10000): 0.0002899999999996794\n",
+      "Total time with gpu (100000000): 0.0003420000000007306\n"
+     ]
+    }
+   ],
+   "source": [
+    "from time import process_time\n",
+    "import torch\n",
+    "\n",
+    "print(torch.get_default_device())\n",
+    "torch.set_default_device('mps')\n",
+    "print(torch.get_default_device())\n",
+    "def testgpu():\n",
+    "    if torch.backends.mps.is_available():\n",
+    "        print(\"available\")\n",
+    "        mps_device = torch.device(\"mps\")\n",
+    "    t0 = process_time()\n",
+    "    x = torch.ones(n1, device=mps_device)\n",
+    "    y = x + torch.rand(n1, device=mps_device)\n",
+    "    t1 = process_time()\n",
+    "    print(f\"Total time with gpu ({n1}): {t1-t0}\")\n",
+    "    t0 = process_time()\n",
+    "    x = torch.ones(n2, device=mps_device)\n",
+    "    y = x + torch.rand(n2, device=mps_device)\n",
+    "    t1 = process_time()\n",
+    "    print(f\"Total time with gpu ({n2}): {t1-t0}\")\n",
+    "\n",
+    "def testcpu():\n",
+    "    t0 = process_time()\n",
+    "    x = torch.ones(n1)\n",
+    "    y = x + torch.rand(n1)\n",
+    "    t1 = process_time()\n",
+    "    print(f\"Total time with cpu ({n1}): {t1-t0}\")\n",
+    "    t0 = process_time()\n",
+    "    x = torch.ones(n2)\n",
+    "    y = x + torch.rand(n2)\n",
+    "    t1 = process_time()\n",
+    "    print(f\"Total time with cpu ({n2}): {t1-t0}\")\n",
+    "\n",
+    "if __name__ == '__main__':\n",
+    "    n1 = 10000\n",
+    "    n2 = 100000000\n",
+    "    testcpu()\n",
+    "    testgpu()"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": []
   }
  ],
  "metadata": {