Improved FileIO

Now the file is kept open to reduce the cost of opening and closing the MMap file.

examples/mvp.ipynb

@@ -580,7 +580,7 @@
    "id": "eac8d73df27f4355",
    "metadata": {},
    "source": [
-    "def decode_mrc_data_path(data_paths: DataPathFrame) -> np.array:\n",
+    "def decode_mrc_data_path(file,data_paths) -> np.array:\n",
     "    \"\"\"\n",
     "    Decode the MRC data path returning the slice of the data speficifed.\n",
     "\n",
@@ -591,10 +591,10 @@
     "        A Numpy array containing the decoded data along with an additional axis to create Height, Width, Channel.\n",
     "\n",
     "    Examples:\n",
-    "        If we have a Data path array of (\"/tmp/0.mrc\",0) with shape (5,5) this will\n",
+    "        If we have an Data path array of (\"/tmp/0.mrc\",0) with shape (5,5) this will\n",
     "        return the numpy array with the shape (5,5,1)\n",
     "    \"\"\"\n",
-    "    mrc_file = mrcfile.mmap(data_paths[0])\n",
+    "    mrc_file = file\n",
     "    frame_index = np.argmax(mrc_file.data.shape)\n",
     "    if len(mrc_file.data.shape) == 3:\n",
     "        if frame_index == 0:\n",
@@ -616,7 +616,7 @@
     "        A list of paths and frames.\n",
     "\n",
     "    Example:\n",
-    "        If we have an MRCfile shape 1,2,3 at /tmp/0.mrc then::\n",
+    "        If we have an MRC file shape 1,2,3 at /tmp/0.mrc then::\n",
     "\n",
     "          data = get_data_paths_and_frames(\"/tmp/0.mrc\")\n",
     "\n",
@@ -654,6 +654,8 @@
     "        transform: typing.Optional[torchvision.transforms.Compose] = None,\n",
     "    ):\n",
     "        self.dataset = dataset\n",
+    "       \n",
+    "        self.data_file =  mrcfile.mmap(dataset[0][0], mode=\"r\")\n",
     "        self.transform = transform\n",
     "\n",
     "    def __len__(self) -> int:\n",
@@ -674,7 +676,7 @@
     "        \"\"\"\n",
     "        if torch.is_tensor(idx):\n",
     "            idx = idx.tolist()\n",
-    "        frame = decode_mrc_data_path(self.dataset[idx])\n",
+    "        frame = decode_mrc_data_path(self.data_file, self.dataset[idx])\n",
     "        if self.transform:\n",
     "            frame = self.transform(frame)\n",
     "        return frame"
@@ -904,11 +906,11 @@
    "metadata": {},
    "source": [
     "config[\"learning_rate\"] = 1e-5\n",
-    "config[\"epochs\"] = 10\n",
+    "config[\"epochs\"] = 5\n",
     "config[\"dataset_name\"] = \"name\"\n",
     "config[\"seed\"] = 0\n",
     "config[\"padding\"] = 1\n",
-    "config[\"batch_size\"] = 32\n",
+    "config[\"batch_size\"] = 64\n",
     "config[\"levels\"] = 1\n",
     "config[\"test_split\"] = 0.2\n",
     "config[\"model_name\"] = \"ConvSRM\"\n",