examples/smile_upscaling/flan-t5-base-lora16.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "import json\n",
    "import os\n",
    "from collections import defaultdict\n",
    "\n",
    "import matplotlib\n",
    "import matplotlib.pyplot as plt\n",
    "import numpy as np\n",
    "import pandas as pd\n",
    "import seaborn as sns\n",
    "from IPython.display import display\n",
    "\n",
    "\n",
    "plt.rcParams[\"font.family\"] = \"Times New Roman\"\n",
    "plt.rcParams[\"mathtext.fontset\"] = \"cm\"\n",
    "\n",
    "matplotlib.rcParams[\"pdf.fonttype\"] = 42\n",
    "matplotlib.rcParams[\"ps.fonttype\"] = 42"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<style type=\"text/css\">\n",
       "#T_5d67d_row0_col0, #T_5d67d_row0_col1, #T_5d67d_row0_col2, #T_5d67d_row0_col7, #T_5d67d_row0_col9, #T_5d67d_row1_col1, #T_5d67d_row1_col4, #T_5d67d_row1_col9, #T_5d67d_row2_col1, #T_5d67d_row3_col2, #T_5d67d_row3_col3, #T_5d67d_row4_col3, #T_5d67d_row4_col6, #T_5d67d_row5_col3, #T_5d67d_row5_col8, #T_5d67d_row5_col11, #T_5d67d_row6_col2, #T_5d67d_row6_col9, #T_5d67d_row6_col10, #T_5d67d_row7_col5, #T_5d67d_row7_col9, #T_5d67d_row8_col5, #T_5d67d_row8_col9 {\n",
       "  background-color: #a50026;\n",
       "  color: #f1f1f1;\n",
       "}\n",
       "#T_5d67d_row0_col3, #T_5d67d_row2_col0 {\n",
       "  background-color: #66bd63;\n",
       "  color: #f1f1f1;\n",
       "}\n",
       "#T_5d67d_row0_col4 {\n",
       "  background-color: #dc3b2c;\n",
       "  color: #f1f1f1;\n",
       "}\n",
       "#T_5d67d_row0_col5, #T_5d67d_row0_col8, #T_5d67d_row1_col3, #T_5d67d_row1_col8, #T_5d67d_row2_col2, #T_5d67d_row2_col3, #T_5d67d_row2_col8, #T_5d67d_row2_col11, #T_5d67d_row3_col4, #T_5d67d_row3_col5, #T_5d67d_row3_col7, #T_5d67d_row4_col4, #T_5d67d_row4_col9, #T_5d67d_row5_col2, #T_5d67d_row5_col6, #T_5d67d_row5_col9, #T_5d67d_row5_col10, #T_5d67d_row6_col1, #T_5d67d_row7_col1, #T_5d67d_row8_col0, #T_5d67d_row8_col1, #T_5d67d_row8_col2 {\n",
       "  background-color: #006837;\n",
       "  color: #f1f1f1;\n",
       "}\n",
       "#T_5d67d_row0_col6, #T_5d67d_row2_col6, #T_5d67d_row3_col6, #T_5d67d_row6_col6 {\n",
       "  background-color: #f88c51;\n",
       "  color: #f1f1f1;\n",
       "}\n",
       "#T_5d67d_row0_col10 {\n",
       "  background-color: #cbe982;\n",
       "  color: #000000;\n",
       "}\n",
       "#T_5d67d_row0_col11 {\n",
       "  background-color: #05713c;\n",
       "  color: #f1f1f1;\n",
       "}\n",
       "#T_5d67d_row1_col0 {\n",
       "  background-color: #fa9857;\n",
       "  color: #000000;\n",
       "}\n",
       "#T_5d67d_row1_col2, #T_5d67d_row3_col1, #T_5d67d_row4_col0, #T_5d67d_row4_col1, #T_5d67d_row4_col2, #T_5d67d_row5_col1, #T_5d67d_row7_col2 {\n",
       "  background-color: #fdbf6f;\n",
       "  color: #000000;\n",
       "}\n",
       "#T_5d67d_row1_col5, #T_5d67d_row2_col5, #T_5d67d_row2_col9, #T_5d67d_row3_col9, #T_5d67d_row4_col5, #T_5d67d_row5_col5, #T_5d67d_row6_col5 {\n",
       "  background-color: #feffbe;\n",
       "  color: #000000;\n",
       "}\n",
       "#T_5d67d_row1_col6, #T_5d67d_row8_col6 {\n",
       "  background-color: #87cb67;\n",
       "  color: #000000;\n",
       "}\n",
       "#T_5d67d_row1_col7, #T_5d67d_row2_col4 {\n",
       "  background-color: #f8864f;\n",
       "  color: #f1f1f1;\n",
       "}\n",
       "#T_5d67d_row1_col10 {\n",
       "  background-color: #cdea83;\n",
       "  color: #000000;\n",
       "}\n",
       "#T_5d67d_row1_col11 {\n",
       "  background-color: #16914d;\n",
       "  color: #f1f1f1;\n",
       "}\n",
       "#T_5d67d_row2_col7 {\n",
       "  background-color: #e95538;\n",
       "  color: #f1f1f1;\n",
       "}\n",
       "#T_5d67d_row2_col10 {\n",
       "  background-color: #b5df74;\n",
       "  color: #000000;\n",
       "}\n",
       "#T_5d67d_row3_col0 {\n",
       "  background-color: #c62027;\n",
       "  color: #f1f1f1;\n",
       "}\n",
       "#T_5d67d_row3_col8, #T_5d67d_row6_col0 {\n",
       "  background-color: #f46d43;\n",
       "  color: #f1f1f1;\n",
       "}\n",
       "#T_5d67d_row3_col10 {\n",
       "  background-color: #fafdb8;\n",
       "  color: #000000;\n",
       "}\n",
       "#T_5d67d_row3_col11 {\n",
       "  background-color: #e65036;\n",
       "  color: #f1f1f1;\n",
       "}\n",
       "#T_5d67d_row4_col7 {\n",
       "  background-color: #04703b;\n",
       "  color: #f1f1f1;\n",
       "}\n",
       "#T_5d67d_row4_col8, #T_5d67d_row6_col3, #T_5d67d_row7_col8 {\n",
       "  background-color: #fee08b;\n",
       "  color: #000000;\n",
       "}\n",
       "#T_5d67d_row4_col10 {\n",
       "  background-color: #118848;\n",
       "  color: #f1f1f1;\n",
       "}\n",
       "#T_5d67d_row4_col11 {\n",
       "  background-color: #fed884;\n",
       "  color: #000000;\n",
       "}\n",
       "#T_5d67d_row5_col0 {\n",
       "  background-color: #33a456;\n",
       "  color: #f1f1f1;\n",
       "}\n",
       "#T_5d67d_row5_col4 {\n",
       "  background-color: #0a7b41;\n",
       "  color: #f1f1f1;\n",
       "}\n",
       "#T_5d67d_row5_col7 {\n",
       "  background-color: #219c52;\n",
       "  color: #f1f1f1;\n",
       "}\n",
       "#T_5d67d_row6_col4, #T_5d67d_row7_col4 {\n",
       "  background-color: #279f53;\n",
       "  color: #f1f1f1;\n",
       "}\n",
       "#T_5d67d_row6_col7 {\n",
       "  background-color: #54b45f;\n",
       "  color: #f1f1f1;\n",
       "}\n",
       "#T_5d67d_row6_col8, #T_5d67d_row7_col3, #T_5d67d_row8_col3, #T_5d67d_row8_col8 {\n",
       "  background-color: #d9ef8b;\n",
       "  color: #000000;\n",
       "}\n",
       "#T_5d67d_row6_col11 {\n",
       "  background-color: #fdb96a;\n",
       "  color: #000000;\n",
       "}\n",
       "#T_5d67d_row7_col0 {\n",
       "  background-color: #fff5ae;\n",
       "  color: #000000;\n",
       "}\n",
       "#T_5d67d_row7_col6 {\n",
       "  background-color: #fffebe;\n",
       "  color: #000000;\n",
       "}\n",
       "#T_5d67d_row7_col7 {\n",
       "  background-color: #48ae5c;\n",
       "  color: #f1f1f1;\n",
       "}\n",
       "#T_5d67d_row7_col10 {\n",
       "  background-color: #e14430;\n",
       "  color: #f1f1f1;\n",
       "}\n",
       "#T_5d67d_row7_col11 {\n",
       "  background-color: #c41e27;\n",
       "  color: #f1f1f1;\n",
       "}\n",
       "#T_5d67d_row8_col4 {\n",
       "  background-color: #45ad5b;\n",
       "  color: #f1f1f1;\n",
       "}\n",
       "#T_5d67d_row8_col7 {\n",
       "  background-color: #18954f;\n",
       "  color: #f1f1f1;\n",
       "}\n",
       "#T_5d67d_row8_col10 {\n",
       "  background-color: #bd1726;\n",
       "  color: #f1f1f1;\n",
       "}\n",
       "#T_5d67d_row8_col11 {\n",
       "  background-color: #fa9b58;\n",
       "  color: #000000;\n",
       "}\n",
       "</style>\n",
       "<table id=\"T_5d67d\">\n",
       "  <thead>\n",
       "    <tr>\n",
       "      <th class=\"blank level0\" >&nbsp;</th>\n",
       "      <th id=\"T_5d67d_level0_col0\" class=\"col_heading level0 col0\" >num_params</th>\n",
       "      <th id=\"T_5d67d_level0_col1\" class=\"col_heading level0 col1\" >gate_k</th>\n",
       "      <th id=\"T_5d67d_level0_col2\" class=\"col_heading level0 col2\" >k</th>\n",
       "      <th id=\"T_5d67d_level0_col3\" class=\"col_heading level0 col3\" >cola</th>\n",
       "      <th id=\"T_5d67d_level0_col4\" class=\"col_heading level0 col4\" >mnli</th>\n",
       "      <th id=\"T_5d67d_level0_col5\" class=\"col_heading level0 col5\" >mrpc</th>\n",
       "      <th id=\"T_5d67d_level0_col6\" class=\"col_heading level0 col6\" >qnli</th>\n",
       "      <th id=\"T_5d67d_level0_col7\" class=\"col_heading level0 col7\" >qqp</th>\n",
       "      <th id=\"T_5d67d_level0_col8\" class=\"col_heading level0 col8\" >rte</th>\n",
       "      <th id=\"T_5d67d_level0_col9\" class=\"col_heading level0 col9\" >sst2</th>\n",
       "      <th id=\"T_5d67d_level0_col10\" class=\"col_heading level0 col10\" >stsb</th>\n",
       "      <th id=\"T_5d67d_level0_col11\" class=\"col_heading level0 col11\" >average</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th id=\"T_5d67d_level0_row0\" class=\"row_heading level0 row0\" >0</th>\n",
       "      <td id=\"T_5d67d_row0_col0\" class=\"data row0 col0\" >252001536</td>\n",
       "      <td id=\"T_5d67d_row0_col1\" class=\"data row0 col1\" >2</td>\n",
       "      <td id=\"T_5d67d_row0_col2\" class=\"data row0 col2\" >4</td>\n",
       "      <td id=\"T_5d67d_row0_col3\" class=\"data row0 col3\" >0.693193</td>\n",
       "      <td id=\"T_5d67d_row0_col4\" class=\"data row0 col4\" >0.829037</td>\n",
       "      <td id=\"T_5d67d_row0_col5\" class=\"data row0 col5\" >0.838235</td>\n",
       "      <td id=\"T_5d67d_row0_col6\" class=\"data row0 col6\" >0.905546</td>\n",
       "      <td id=\"T_5d67d_row0_col7\" class=\"data row0 col7\" >0.838585</td>\n",
       "      <td id=\"T_5d67d_row0_col8\" class=\"data row0 col8\" >0.833935</td>\n",
       "      <td id=\"T_5d67d_row0_col9\" class=\"data row0 col9\" >0.931193</td>\n",
       "      <td id=\"T_5d67d_row0_col10\" class=\"data row0 col10\" >0.850887</td>\n",
       "      <td id=\"T_5d67d_row0_col11\" class=\"data row0 col11\" >84.007647</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_5d67d_level0_row1\" class=\"row_heading level0 row1\" >1</th>\n",
       "      <td id=\"T_5d67d_row1_col0\" class=\"data row1 col0\" >255540480</td>\n",
       "      <td id=\"T_5d67d_row1_col1\" class=\"data row1 col1\" >2</td>\n",
       "      <td id=\"T_5d67d_row1_col2\" class=\"data row1 col2\" >8</td>\n",
       "      <td id=\"T_5d67d_row1_col3\" class=\"data row1 col3\" >0.694151</td>\n",
       "      <td id=\"T_5d67d_row1_col4\" class=\"data row1 col4\" >0.828732</td>\n",
       "      <td id=\"T_5d67d_row1_col5\" class=\"data row1 col5\" >0.835784</td>\n",
       "      <td id=\"T_5d67d_row1_col6\" class=\"data row1 col6\" >0.905913</td>\n",
       "      <td id=\"T_5d67d_row1_col7\" class=\"data row1 col7\" >0.838956</td>\n",
       "      <td id=\"T_5d67d_row1_col8\" class=\"data row1 col8\" >0.833935</td>\n",
       "      <td id=\"T_5d67d_row1_col9\" class=\"data row1 col9\" >0.931193</td>\n",
       "      <td id=\"T_5d67d_row1_col10\" class=\"data row1 col10\" >0.850867</td>\n",
       "      <td id=\"T_5d67d_row1_col11\" class=\"data row1 col11\" >83.994131</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_5d67d_level0_row2\" class=\"row_heading level0 row2\" >2</th>\n",
       "      <td id=\"T_5d67d_row2_col0\" class=\"data row2 col0\" >262618368</td>\n",
       "      <td id=\"T_5d67d_row2_col1\" class=\"data row2 col1\" >2</td>\n",
       "      <td id=\"T_5d67d_row2_col2\" class=\"data row2 col2\" >16</td>\n",
       "      <td id=\"T_5d67d_row2_col3\" class=\"data row2 col3\" >0.694151</td>\n",
       "      <td id=\"T_5d67d_row2_col4\" class=\"data row2 col4\" >0.829343</td>\n",
       "      <td id=\"T_5d67d_row2_col5\" class=\"data row2 col5\" >0.835784</td>\n",
       "      <td id=\"T_5d67d_row2_col6\" class=\"data row2 col6\" >0.905546</td>\n",
       "      <td id=\"T_5d67d_row2_col7\" class=\"data row2 col7\" >0.838833</td>\n",
       "      <td id=\"T_5d67d_row2_col8\" class=\"data row2 col8\" >0.833935</td>\n",
       "      <td id=\"T_5d67d_row2_col9\" class=\"data row2 col9\" >0.932339</td>\n",
       "      <td id=\"T_5d67d_row2_col10\" class=\"data row2 col10\" >0.851063</td>\n",
       "      <td id=\"T_5d67d_row2_col11\" class=\"data row2 col11\" >84.012436</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_5d67d_level0_row3\" class=\"row_heading level0 row3\" >3</th>\n",
       "      <td id=\"T_5d67d_row3_col0\" class=\"data row3 col0\" >252886272</td>\n",
       "      <td id=\"T_5d67d_row3_col1\" class=\"data row3 col1\" >4</td>\n",
       "      <td id=\"T_5d67d_row3_col2\" class=\"data row3 col2\" >4</td>\n",
       "      <td id=\"T_5d67d_row3_col3\" class=\"data row3 col3\" >0.689358</td>\n",
       "      <td id=\"T_5d67d_row3_col4\" class=\"data row3 col4\" >0.831279</td>\n",
       "      <td id=\"T_5d67d_row3_col5\" class=\"data row3 col5\" >0.838235</td>\n",
       "      <td id=\"T_5d67d_row3_col6\" class=\"data row3 col6\" >0.905546</td>\n",
       "      <td id=\"T_5d67d_row3_col7\" class=\"data row3 col7\" >0.840119</td>\n",
       "      <td id=\"T_5d67d_row3_col8\" class=\"data row3 col8\" >0.819495</td>\n",
       "      <td id=\"T_5d67d_row3_col9\" class=\"data row3 col9\" >0.932339</td>\n",
       "      <td id=\"T_5d67d_row3_col10\" class=\"data row3 col10\" >0.850409</td>\n",
       "      <td id=\"T_5d67d_row3_col11\" class=\"data row3 col11\" >83.834748</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_5d67d_level0_row4\" class=\"row_heading level0 row4\" >4</th>\n",
       "      <td id=\"T_5d67d_row4_col0\" class=\"data row4 col0\" >256425216</td>\n",
       "      <td id=\"T_5d67d_row4_col1\" class=\"data row4 col1\" >4</td>\n",
       "      <td id=\"T_5d67d_row4_col2\" class=\"data row4 col2\" >8</td>\n",
       "      <td id=\"T_5d67d_row4_col3\" class=\"data row4 col3\" >0.689358</td>\n",
       "      <td id=\"T_5d67d_row4_col4\" class=\"data row4 col4\" >0.831279</td>\n",
       "      <td id=\"T_5d67d_row4_col5\" class=\"data row4 col5\" >0.835784</td>\n",
       "      <td id=\"T_5d67d_row4_col6\" class=\"data row4 col6\" >0.905363</td>\n",
       "      <td id=\"T_5d67d_row4_col7\" class=\"data row4 col7\" >0.840094</td>\n",
       "      <td id=\"T_5d67d_row4_col8\" class=\"data row4 col8\" >0.823105</td>\n",
       "      <td id=\"T_5d67d_row4_col9\" class=\"data row4 col9\" >0.933486</td>\n",
       "      <td id=\"T_5d67d_row4_col10\" class=\"data row4 col10\" >0.852149</td>\n",
       "      <td id=\"T_5d67d_row4_col11\" class=\"data row4 col11\" >83.882725</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_5d67d_level0_row5\" class=\"row_heading level0 row5\" >5</th>\n",
       "      <td id=\"T_5d67d_row5_col0\" class=\"data row5 col0\" >263503104</td>\n",
       "      <td id=\"T_5d67d_row5_col1\" class=\"data row5 col1\" >4</td>\n",
       "      <td id=\"T_5d67d_row5_col2\" class=\"data row5 col2\" >16</td>\n",
       "      <td id=\"T_5d67d_row5_col3\" class=\"data row5 col3\" >0.689358</td>\n",
       "      <td id=\"T_5d67d_row5_col4\" class=\"data row5 col4\" >0.831177</td>\n",
       "      <td id=\"T_5d67d_row5_col5\" class=\"data row5 col5\" >0.835784</td>\n",
       "      <td id=\"T_5d67d_row5_col6\" class=\"data row5 col6\" >0.906096</td>\n",
       "      <td id=\"T_5d67d_row5_col7\" class=\"data row5 col7\" >0.839946</td>\n",
       "      <td id=\"T_5d67d_row5_col8\" class=\"data row5 col8\" >0.815884</td>\n",
       "      <td id=\"T_5d67d_row5_col9\" class=\"data row5 col9\" >0.933486</td>\n",
       "      <td id=\"T_5d67d_row5_col10\" class=\"data row5 col10\" >0.852436</td>\n",
       "      <td id=\"T_5d67d_row5_col11\" class=\"data row5 col11\" >83.802078</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_5d67d_level0_row6\" class=\"row_heading level0 row6\" >6</th>\n",
       "      <td id=\"T_5d67d_row6_col0\" class=\"data row6 col0\" >254655744</td>\n",
       "      <td id=\"T_5d67d_row6_col1\" class=\"data row6 col1\" >8</td>\n",
       "      <td id=\"T_5d67d_row6_col2\" class=\"data row6 col2\" >4</td>\n",
       "      <td id=\"T_5d67d_row6_col3\" class=\"data row6 col3\" >0.691275</td>\n",
       "      <td id=\"T_5d67d_row6_col4\" class=\"data row6 col4\" >0.830973</td>\n",
       "      <td id=\"T_5d67d_row6_col5\" class=\"data row6 col5\" >0.835784</td>\n",
       "      <td id=\"T_5d67d_row6_col6\" class=\"data row6 col6\" >0.905546</td>\n",
       "      <td id=\"T_5d67d_row6_col7\" class=\"data row6 col7\" >0.839847</td>\n",
       "      <td id=\"T_5d67d_row6_col8\" class=\"data row6 col8\" >0.826715</td>\n",
       "      <td id=\"T_5d67d_row6_col9\" class=\"data row6 col9\" >0.931193</td>\n",
       "      <td id=\"T_5d67d_row6_col10\" class=\"data row6 col10\" >0.848279</td>\n",
       "      <td id=\"T_5d67d_row6_col11\" class=\"data row6 col11\" >83.870147</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_5d67d_level0_row7\" class=\"row_heading level0 row7\" >7</th>\n",
       "      <td id=\"T_5d67d_row7_col0\" class=\"data row7 col0\" >258194688</td>\n",
       "      <td id=\"T_5d67d_row7_col1\" class=\"data row7 col1\" >8</td>\n",
       "      <td id=\"T_5d67d_row7_col2\" class=\"data row7 col2\" >8</td>\n",
       "      <td id=\"T_5d67d_row7_col3\" class=\"data row7 col3\" >0.692234</td>\n",
       "      <td id=\"T_5d67d_row7_col4\" class=\"data row7 col4\" >0.830973</td>\n",
       "      <td id=\"T_5d67d_row7_col5\" class=\"data row7 col5\" >0.833333</td>\n",
       "      <td id=\"T_5d67d_row7_col6\" class=\"data row7 col6\" >0.905729</td>\n",
       "      <td id=\"T_5d67d_row7_col7\" class=\"data row7 col7\" >0.839871</td>\n",
       "      <td id=\"T_5d67d_row7_col8\" class=\"data row7 col8\" >0.823105</td>\n",
       "      <td id=\"T_5d67d_row7_col9\" class=\"data row7 col9\" >0.931193</td>\n",
       "      <td id=\"T_5d67d_row7_col10\" class=\"data row7 col10\" >0.848835</td>\n",
       "      <td id=\"T_5d67d_row7_col11\" class=\"data row7 col11\" >83.815917</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_5d67d_level0_row8\" class=\"row_heading level0 row8\" >8</th>\n",
       "      <td id=\"T_5d67d_row8_col0\" class=\"data row8 col0\" >265272576</td>\n",
       "      <td id=\"T_5d67d_row8_col1\" class=\"data row8 col1\" >8</td>\n",
       "      <td id=\"T_5d67d_row8_col2\" class=\"data row8 col2\" >16</td>\n",
       "      <td id=\"T_5d67d_row8_col3\" class=\"data row8 col3\" >0.692234</td>\n",
       "      <td id=\"T_5d67d_row8_col4\" class=\"data row8 col4\" >0.830871</td>\n",
       "      <td id=\"T_5d67d_row8_col5\" class=\"data row8 col5\" >0.833333</td>\n",
       "      <td id=\"T_5d67d_row8_col6\" class=\"data row8 col6\" >0.905913</td>\n",
       "      <td id=\"T_5d67d_row8_col7\" class=\"data row8 col7\" >0.839970</td>\n",
       "      <td id=\"T_5d67d_row8_col8\" class=\"data row8 col8\" >0.826715</td>\n",
       "      <td id=\"T_5d67d_row8_col9\" class=\"data row8 col9\" >0.931193</td>\n",
       "      <td id=\"T_5d67d_row8_col10\" class=\"data row8 col10\" >0.848487</td>\n",
       "      <td id=\"T_5d67d_row8_col11\" class=\"data row8 col11\" >83.858946</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n"
      ],
      "text/plain": [
       "<pandas.io.formats.style.Styler at 0x7f4bcc6b0460>"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "data_dir = \"/workspace/svd_subspace/outputs/flan-t5-base_lora16/glue_text_generation/\"\n",
    "moe_data = defaultdict(list)\n",
    "for gate_k in [2, 4, 8]:\n",
    "    for k in [4, 8, 16]:\n",
    "        file_name = f\"gate_k={gate_k}_k={k}.json\"\n",
    "        file_name = os.path.join(data_dir, file_name)\n",
    "        if not os.path.exists(file_name):\n",
    "            # print(f\"File {file_name} does not exist\")\n",
    "            continue\n",
    "        with open(file_name) as f:\n",
    "            data = json.load(f)\n",
    "        moe_data[\"num_params\"].append(data[\"model_info\"][\"all_params\"])\n",
    "        moe_data[\"gate_k\"].append(gate_k)\n",
    "        moe_data[\"k\"].append(k)\n",
    "        scores = {}\n",
    "        for key in data:\n",
    "            if key == \"model_info\":\n",
    "                continue\n",
    "            scores[key] = (\n",
    "                data[key][\"accuracy\"]\n",
    "                if \"accuracy\" in data[key]\n",
    "                else data[key][\"spearman_rho\"]\n",
    "            )\n",
    "        average_accuracy = np.mean(list(scores.values())) * 100\n",
    "        for key in scores:\n",
    "            moe_data[key].append(scores[key])\n",
    "        moe_data[\"average\"].append(average_accuracy)\n",
    "moe_data = pd.DataFrame(moe_data)\n",
    "\n",
    "# 使用 Styler 对象并应用背景颜色渐变\n",
    "styled_df = moe_data.style.background_gradient(cmap=\"RdYlGn\")\n",
    "\n",
    "# 显示结果\n",
    "styled_df"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/tmp/ipykernel_764261/1028603405.py:5: FutureWarning: Setting an item of incompatible dtype is deprecated and will raise an error in a future version of pandas. Value 'Dense' has dtype incompatible with int64, please explicitly cast to a compatible dtype first.\n",
      "  plot_data.loc[plot_data[\"k\"] == -1, \"k\"] = \"Dense\"\n",
      "'created' timestamp seems very low; regarding as unix timestamp\n",
      "'modified' timestamp seems very low; regarding as unix timestamp\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/tmp/ipykernel_764261/1028603405.py:20: FutureWarning: Setting an item of incompatible dtype is deprecated and will raise an error in a future version of pandas. Value 'Dense' has dtype incompatible with int64, please explicitly cast to a compatible dtype first.\n",
      "  plot_data.loc[plot_data[\"k\"] == -1, \"k\"] = \"Dense\"\n",
      "'created' timestamp seems very low; regarding as unix timestamp\n",
      "'modified' timestamp seems very low; regarding as unix timestamp\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "matplotlib.rcParams[\"font.size\"] = 28\n",
    "\n",
    "plot_data = moe_data.copy()\n",
    "plot_data[\"method\"] = \"Ours\"\n",
    "plot_data.loc[plot_data[\"k\"] == -1, \"k\"] = \"Dense\"\n",
    "pivot_data = (\n",
    "    plot_data[[\"gate_k\", \"k\", \"average\"]]\n",
    "    .pivot(index=\"gate_k\", columns=\"k\", values=\"average\")\n",
    "    .copy()\n",
    ")\n",
    "\n",
    "sns.heatmap(data=pivot_data, annot=True, fmt=\".1f\", cmap=\"RdYlGn\")\n",
    "plt.xlabel(\"$k$\")\n",
    "plt.ylabel(\"$k_{gate}$\")\n",
    "plt.savefig(\"flan-t5-base_lora16_hp-acc.pdf\", bbox_inches=\"tight\")\n",
    "plt.show()\n",
    "\n",
    "plot_data = moe_data.copy()\n",
    "plot_data[\"method\"] = \"Ours\"\n",
    "plot_data.loc[plot_data[\"k\"] == -1, \"k\"] = \"Dense\"\n",
    "plot_data[\"num_params\"] = plot_data[\"num_params\"] / 247577856\n",
    "pivot_data = (\n",
    "    plot_data[[\"gate_k\", \"k\", \"num_params\"]]\n",
    "    .pivot(index=\"gate_k\", columns=\"k\", values=\"num_params\")\n",
    "    .copy()\n",
    ")\n",
    "\n",
    "sns.heatmap(data=pivot_data, annot=True, fmt=\".2f\", cmap=\"RdYlGn_r\")\n",
    "plt.xlabel(\"$k$\")\n",
    "plt.ylabel(\"$k_{gate}$\")\n",
    "plt.savefig(\"flan-t5-base_lora16_hp-params.pdf\", bbox_inches=\"tight\")\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "base",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.10.14"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}