Merge branch 'mikecafarella:main' into main

DARPA-ASKEM · Aug 24, 2023 · 955e432 · 955e432
2 parents 0e72a65 + b7891c5
commit 955e432
Show file tree

Hide file tree

Showing 4 changed files with 133 additions and 22 deletions.
diff --git a/demos/2023-08/mit-aug-demo-data-model-cards.ipynb b/demos/2023-08/mit-aug-demo-data-model-cards.ipynb
diff --git a/...erventions in Italy-s41591-020-0883-7.pdf → demos/2023-08/sidarthe.pdf b/...erventions in Italy-s41591-020-0883-7.pdf → demos/2023-08/sidarthe.pdf
diff --git a/demos/extract_vars_example.ipynb b/demos/extract_vars_example.ipynb
@@ -12,7 +12,10 @@
      "shell.execute_reply": "2023-07-11T16:02:24.134500Z",
      "shell.execute_reply.started": "2023-07-11T16:02:24.123275Z"
     },
-    "tags": []
+    "tags": [],
+    "pycharm": {
+     "name": "#%%\n"
+    }
    },
    "outputs": [],
    "source": [
@@ -34,6 +37,9 @@
      "iopub.status.idle": "2023-07-11T16:02:25.102485Z",
      "shell.execute_reply": "2023-07-11T16:02:25.102041Z",
      "shell.execute_reply.started": "2023-07-11T16:02:24.144245Z"
+    },
+    "pycharm": {
+     "name": "#%%\n"
     }
    },
    "outputs": [],
@@ -52,6 +58,9 @@
      "iopub.status.idle": "2023-07-11T16:09:37.531739Z",
      "shell.execute_reply": "2023-07-11T16:09:37.531033Z",
      "shell.execute_reply.started": "2023-07-11T16:09:37.524656Z"
+    },
+    "pycharm": {
+     "name": "#%%\n"
     }
    },
    "outputs": [],
@@ -76,6 +85,9 @@
      "iopub.status.idle": "2023-07-11T16:02:25.352277Z",
      "shell.execute_reply": "2023-07-11T16:02:25.351864Z",
      "shell.execute_reply.started": "2023-07-11T16:02:25.348849Z"
+    },
+    "pycharm": {
+     "name": "#%%\n"
     }
    },
    "outputs": [],
@@ -95,6 +107,9 @@
      "iopub.status.idle": "2023-07-11T16:02:25.658409Z",
      "shell.execute_reply": "2023-07-11T16:02:25.657608Z",
      "shell.execute_reply.started": "2023-07-11T16:02:25.654792Z"
+    },
+    "pycharm": {
+     "name": "#%%\n"
     }
    },
    "outputs": [],
@@ -113,6 +128,9 @@
      "iopub.status.idle": "2023-07-11T16:02:32.968263Z",
      "shell.execute_reply": "2023-07-11T16:02:32.967248Z",
      "shell.execute_reply.started": "2023-07-11T16:02:32.964017Z"
+    },
+    "pycharm": {
+     "name": "#%%\n"
     }
    },
    "outputs": [],
@@ -131,6 +149,9 @@
      "iopub.status.idle": "2023-07-11T16:10:41.152175Z",
      "shell.execute_reply": "2023-07-11T16:10:41.151527Z",
      "shell.execute_reply.started": "2023-07-11T16:10:24.156244Z"
+    },
+    "pycharm": {
+     "name": "#%%\n"
     }
    },
    "outputs": [],
@@ -150,6 +171,9 @@
      "iopub.status.idle": "2023-07-11T16:10:44.415537Z",
      "shell.execute_reply": "2023-07-11T16:10:44.414974Z",
      "shell.execute_reply.started": "2023-07-11T16:10:44.409732Z"
+    },
+    "pycharm": {
+     "name": "#%%\n"
     }
    },
    "outputs": [
@@ -204,6 +228,9 @@
      "iopub.status.idle": "2023-07-11T16:04:40.391159Z",
      "shell.execute_reply": "2023-07-11T16:04:40.390371Z",
      "shell.execute_reply.started": "2023-07-11T16:04:40.387219Z"
+    },
+    "pycharm": {
+     "name": "#%%\n"
     }
    },
    "outputs": [],
@@ -222,6 +249,9 @@
      "iopub.status.idle": "2023-07-11T15:54:59.557247Z",
      "shell.execute_reply": "2023-07-11T15:54:59.556493Z",
      "shell.execute_reply.started": "2023-07-11T15:54:59.542853Z"
+    },
+    "pycharm": {
+     "name": "#%%\n"
     }
    },
    "outputs": [
@@ -251,6 +281,9 @@
      "iopub.status.idle": "2023-07-11T15:55:28.660114Z",
      "shell.execute_reply": "2023-07-11T15:55:28.659649Z",
      "shell.execute_reply.started": "2023-07-11T15:55:28.591129Z"
+    },
+    "pycharm": {
+     "name": "#%%\n"
     }
    },
    "outputs": [
@@ -280,6 +313,9 @@
      "iopub.status.idle": "2023-07-11T15:33:56.145057Z",
      "shell.execute_reply": "2023-07-11T15:33:56.144327Z",
      "shell.execute_reply.started": "2023-07-11T15:33:56.140338Z"
+    },
+    "pycharm": {
+     "name": "#%%\n"
     }
    },
    "outputs": [],
@@ -299,6 +335,9 @@
      "iopub.status.idle": "2023-07-11T15:34:00.662425Z",
      "shell.execute_reply": "2023-07-11T15:34:00.661734Z",
      "shell.execute_reply.started": "2023-07-11T15:34:00.658594Z"
+    },
+    "pycharm": {
+     "name": "#%%\n"
     }
    },
    "outputs": [
@@ -329,6 +368,9 @@
      "iopub.status.idle": "2023-07-11T15:00:13.333046Z",
      "shell.execute_reply": "2023-07-11T15:00:13.332666Z",
      "shell.execute_reply.started": "2023-07-11T15:00:13.248300Z"
+    },
+    "pycharm": {
+     "name": "#%%\n"
     }
    },
    "outputs": [
@@ -531,6 +573,9 @@
      "iopub.status.idle": "2023-07-10T21:44:09.328972Z",
      "shell.execute_reply": "2023-07-10T21:44:09.328192Z",
      "shell.execute_reply.started": "2023-07-10T21:44:09.321677Z"
+    },
+    "pycharm": {
+     "name": "#%%\n"
     }
    },
    "outputs": [],
@@ -552,6 +597,9 @@
      "iopub.status.idle": "2023-07-11T14:55:15.364057Z",
      "shell.execute_reply": "2023-07-11T14:55:15.363361Z",
      "shell.execute_reply.started": "2023-07-11T14:55:15.360527Z"
+    },
+    "pycharm": {
+     "name": "#%%\n"
     }
    },
    "outputs": [],
@@ -570,6 +618,9 @@
      "iopub.status.idle": "2023-07-11T14:56:20.099130Z",
      "shell.execute_reply": "2023-07-11T14:56:20.098256Z",
      "shell.execute_reply.started": "2023-07-11T14:56:20.095258Z"
+    },
+    "pycharm": {
+     "name": "#%%\n"
     }
    },
    "outputs": [],
@@ -589,6 +640,9 @@
      "iopub.status.idle": "2023-07-11T16:10:19.437463Z",
      "shell.execute_reply": "2023-07-11T16:10:19.436852Z",
      "shell.execute_reply.started": "2023-07-11T16:10:19.433453Z"
+    },
+    "pycharm": {
+     "name": "#%%\n"
     }
    },
    "outputs": [],
@@ -608,6 +662,9 @@
      "iopub.status.idle": "2023-07-11T14:56:25.675638Z",
      "shell.execute_reply": "2023-07-11T14:56:25.674767Z",
      "shell.execute_reply.started": "2023-07-11T14:56:25.671504Z"
+    },
+    "pycharm": {
+     "name": "#%%\n"
     }
    },
    "outputs": [
@@ -818,6 +875,9 @@
      "iopub.status.idle": "2023-07-09T03:20:21.640719Z",
      "shell.execute_reply": "2023-07-09T03:20:21.640329Z",
      "shell.execute_reply.started": "2023-07-09T03:20:21.604089Z"
+    },
+    "pycharm": {
+     "name": "#%%\n"
     }
    },
    "outputs": [],
@@ -842,6 +902,9 @@
      "iopub.status.idle": "2023-07-09T03:20:32.277515Z",
      "shell.execute_reply": "2023-07-09T03:20:32.276686Z",
      "shell.execute_reply.started": "2023-07-09T03:20:32.274315Z"
+    },
+    "pycharm": {
+     "name": "#%%\n"
     }
    },
    "outputs": [],
@@ -860,6 +923,9 @@
      "iopub.status.idle": "2023-07-10T19:44:50.800018Z",
      "shell.execute_reply": "2023-07-10T19:44:50.799200Z",
      "shell.execute_reply.started": "2023-07-10T19:44:50.794682Z"
+    },
+    "pycharm": {
+     "name": "#%%\n"
     }
    },
    "outputs": [],
@@ -878,7 +944,11 @@
    "cell_type": "code",
    "execution_count": 14,
    "id": "7d72796d",
-   "metadata": {},
+   "metadata": {
+    "pycharm": {
+     "name": "#%%\n"
+    }
+   },
    "outputs": [],
    "source": [
     "#os.environ['OPENAI_API_KEY']"
@@ -895,6 +965,9 @@
      "iopub.status.idle": "2023-07-05T15:46:08.882776Z",
      "shell.execute_reply": "2023-07-05T15:46:08.881823Z",
      "shell.execute_reply.started": "2023-07-05T15:46:08.877979Z"
+    },
+    "pycharm": {
+     "name": "#%%\n"
     }
    },
    "outputs": [],
@@ -913,6 +986,9 @@
      "iopub.status.idle": "2023-07-10T19:42:29.870784Z",
      "shell.execute_reply": "2023-07-10T19:42:29.869350Z",
      "shell.execute_reply.started": "2023-07-10T19:42:19.201975Z"
+    },
+    "pycharm": {
+     "name": "#%%\n"
     }
    },
    "outputs": [
@@ -942,6 +1018,9 @@
      "iopub.status.idle": "2023-07-10T19:20:18.951735Z",
      "shell.execute_reply": "2023-07-10T19:20:18.950827Z",
      "shell.execute_reply.started": "2023-07-10T19:20:01.423748Z"
+    },
+    "pycharm": {
+     "name": "#%%\n"
     }
    },
    "outputs": [
@@ -970,6 +1049,9 @@
      "iopub.status.idle": "2023-07-10T19:22:14.058735Z",
      "shell.execute_reply": "2023-07-10T19:22:14.057443Z",
      "shell.execute_reply.started": "2023-07-10T19:21:58.723306Z"
+    },
+    "pycharm": {
+     "name": "#%%\n"
     }
    },
    "outputs": [
@@ -997,6 +1079,9 @@
      "iopub.status.idle": "2023-07-05T15:46:30.375913Z",
      "shell.execute_reply": "2023-07-05T15:46:30.375259Z",
      "shell.execute_reply.started": "2023-07-05T15:46:30.371890Z"
+    },
+    "pycharm": {
+     "name": "#%%\n"
     }
    },
    "outputs": [],
@@ -1016,6 +1101,9 @@
      "iopub.status.idle": "2023-07-10T19:23:29.129796Z",
      "shell.execute_reply": "2023-07-10T19:23:29.128774Z",
      "shell.execute_reply.started": "2023-07-10T19:23:13.782353Z"
+    },
+    "pycharm": {
+     "name": "#%%\n"
     }
    },
    "outputs": [
@@ -1045,6 +1133,9 @@
      "iopub.status.idle": "2023-07-05T15:46:59.312211Z",
      "shell.execute_reply": "2023-07-05T15:46:59.311424Z",
      "shell.execute_reply.started": "2023-07-05T15:46:59.307950Z"
+    },
+    "pycharm": {
+     "name": "#%%\n"
     }
    },
    "outputs": [],
@@ -1070,6 +1161,9 @@
      "iopub.status.idle": "2023-07-05T15:47:12.030693Z",
      "shell.execute_reply": "2023-07-05T15:47:12.028863Z",
      "shell.execute_reply.started": "2023-07-05T15:47:12.025721Z"
+    },
+    "pycharm": {
+     "name": "#%%\n"
     }
    },
    "outputs": [],
@@ -1088,6 +1182,9 @@
      "iopub.status.idle": "2023-07-05T15:48:16.500408Z",
      "shell.execute_reply": "2023-07-05T15:48:16.498292Z",
      "shell.execute_reply.started": "2023-07-05T15:48:16.488175Z"
+    },
+    "pycharm": {
+     "name": "#%%\n"
     }
    },
    "outputs": [],
@@ -1107,6 +1204,9 @@
      "iopub.status.idle": "2023-07-05T15:48:46.336796Z",
      "shell.execute_reply": "2023-07-05T15:48:46.335206Z",
      "shell.execute_reply.started": "2023-07-05T15:48:46.329876Z"
+    },
+    "pycharm": {
+     "name": "#%%\n"
     }
    },
    "outputs": [],
@@ -1125,6 +1225,9 @@
      "iopub.status.idle": "2023-07-05T15:48:53.441866Z",
      "shell.execute_reply": "2023-07-05T15:48:53.440991Z",
      "shell.execute_reply.started": "2023-07-05T15:48:53.437089Z"
+    },
+    "pycharm": {
+     "name": "#%%\n"
     }
    },
    "outputs": [
@@ -1154,6 +1257,9 @@
      "iopub.status.idle": "2023-07-05T15:49:02.313156Z",
      "shell.execute_reply": "2023-07-05T15:49:02.312317Z",
      "shell.execute_reply.started": "2023-07-05T15:49:02.308574Z"
+    },
+    "pycharm": {
+     "name": "#%%\n"
     }
    },
    "outputs": [],
@@ -1172,6 +1278,9 @@
      "iopub.status.idle": "2023-07-05T15:49:30.957414Z",
      "shell.execute_reply": "2023-07-05T15:49:30.956622Z",
      "shell.execute_reply.started": "2023-07-05T15:49:30.951587Z"
+    },
+    "pycharm": {
+     "name": "#%%\n"
     }
    },
    "outputs": [],
@@ -1190,6 +1299,9 @@
      "iopub.status.idle": "2023-07-05T15:49:31.643917Z",
      "shell.execute_reply": "2023-07-05T15:49:31.642537Z",
      "shell.execute_reply.started": "2023-07-05T15:49:31.622891Z"
+    },
+    "pycharm": {
+     "name": "#%%\n"
     }
    },
    "outputs": [
@@ -1325,4 +1437,4 @@
  },
  "nbformat": 4,
  "nbformat_minor": 5
-}
+}
diff --git a/src/xdd/xdd_parser.py b/src/xdd/xdd_parser.py
@@ -177,7 +177,7 @@ def main(file_path, output_dir):
 
 if __name__ == "__main__":
     parser = argparse.ArgumentParser(description='Process a PDF file.')
-    parser.add_argument('--file', type=str, default="/Users/chunwei/Documents/paper/LLM/Flamingo.pdf",
+    parser.add_argument('--file', type=str, default="/Users/chunwei/Downloads/sidarthe.pdf",
                         help='The path to the PDF file to process.')
     parser.add_argument('--output_dir', type=str, default="/Users/chunwei/Downloads/",
                         help='The directory to save the extraction results.')