@arizeai/phoenix-mcp

{ "cells": [ { "cell_type": "markdown", "metadata": {}, "source": [ "# Wrangle HaluEval\n", "\n", "Wrangles [HaluEval](https://github.com/RUCAIBox/HaluEval) QA data." ] }, { "cell_type": "code", "execution_count": null, "metadata": {}, "outputs": [], "source": [ "import json\n", "import os\n", "from urllib.request import urlopen\n", "\n", "import pandas as pd\n", "\n", "pd.set_option(\"display.max_colwidth\", None)\n", "\n", "data_url = \"https://raw.githubusercontent.com/RUCAIBox/HaluEval/main/data/qa_data.json\"\n", "records = []\n", "with urlopen(data_url) as url:\n", " for line in url.readlines():\n", " records.append(json.loads(line))\n", "df = pd.DataFrame(records)\n", "df" ] }, { "cell_type": "code", "execution_count": null, "metadata": {}, "outputs": [], "source": [ "df = df.melt(\n", " id_vars=[\"knowledge\", \"question\"],\n", " value_vars=[\"right_answer\", \"hallucinated_answer\"],\n", " var_name=\"answer_type\",\n", " value_name=\"answer\",\n", ")\n", "df[\"is_hallucination\"] = df[\"answer_type\"] == \"hallucinated_answer\"\n", "df = df.drop(\"answer_type\", axis=1)\n", "df = df.sort_values([\"knowledge\", \"question\"]).reset_index(drop=True)\n", "df = df.rename(\n", " columns={\"knowledge\": \"reference\", \"question\": \"query\", \"answer\": \"response\"},\n", ")\n", "df" ] }, { "cell_type": "code", "execution_count": null, "metadata": {}, "outputs": [], "source": [ "output_path = \"halueval_qa_data.jsonl\"\n", "\n", "try:\n", " os.remove(output_path)\n", "except OSError:\n", " pass\n", "\n", "with open(output_path, \"a\") as f:\n", " for record in df.to_dict(orient=\"records\"):\n", " f.write(json.dumps(record) + \"\\n\")" ] } ], "metadata": { "language_info": { "name": "python" } }, "nbformat": 4, "nbformat_minor": 2 }