{ "nbformat": 4, "nbformat_minor": 0, "metadata": { "colab": { "provenance": [], "authorship_tag": "ABX9TyPoTz8vGlLEA24WzE8qcB+M", "include_colab_link": true }, "kernelspec": { "name": "python3", "display_name": "Python 3" }, "language_info": { "name": "python" } }, "cells": [ { "cell_type": "markdown", "metadata": { "id": "view-in-github", "colab_type": "text" }, "source": [ "\"Open" ] }, { "cell_type": "code", "source": [ "!pip install -q transformers openai tiktoken langchain chromadb erniebot\n", "!pip install -q chatharuhi\n", "!pip install -q datasets" ], "metadata": { "id": "a8H7Az3Yzi3o", "colab": { "base_uri": "https://localhost:8080/" }, "outputId": "ea8102e2-40c2-4e38-8cb5-469b80852344" }, "execution_count": 1, "outputs": [ { "output_type": "stream", "name": "stdout", "text": [ "\u001b[2K \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m7.9/7.9 MB\u001b[0m \u001b[31m45.9 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n", "\u001b[2K \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m217.8/217.8 kB\u001b[0m \u001b[31m24.6 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n", "\u001b[2K \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m2.0/2.0 MB\u001b[0m \u001b[31m56.1 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n", "\u001b[2K \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m2.0/2.0 MB\u001b[0m \u001b[31m88.8 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n", "\u001b[2K \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m479.8/479.8 kB\u001b[0m \u001b[31m40.9 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n", "\u001b[2K \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m68.4/68.4 kB\u001b[0m \u001b[31m8.6 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n", "\u001b[2K \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m302.0/302.0 kB\u001b[0m \u001b[31m26.4 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n", "\u001b[2K \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m3.8/3.8 MB\u001b[0m \u001b[31m64.0 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n", "\u001b[2K \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m1.3/1.3 MB\u001b[0m \u001b[31m82.6 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n", "\u001b[2K \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m75.0/75.0 kB\u001b[0m \u001b[31m9.3 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n", "\u001b[2K \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m44.8/44.8 kB\u001b[0m \u001b[31m3.3 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n", "\u001b[2K \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m2.4/2.4 MB\u001b[0m \u001b[31m47.8 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n", "\u001b[2K \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m92.9/92.9 kB\u001b[0m \u001b[31m12.2 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n", "\u001b[2K \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m59.7/59.7 kB\u001b[0m \u001b[31m7.7 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n", "\u001b[2K \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m5.4/5.4 MB\u001b[0m \u001b[31m36.4 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n", "\u001b[2K \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m6.2/6.2 MB\u001b[0m \u001b[31m54.5 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n", "\u001b[2K \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m57.5/57.5 kB\u001b[0m \u001b[31m6.6 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n", "\u001b[2K \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m103.9/103.9 kB\u001b[0m \u001b[31m11.9 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n", "\u001b[2K \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m67.3/67.3 kB\u001b[0m \u001b[31m4.7 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n", "\u001b[?25h Installing build dependencies ... \u001b[?25l\u001b[?25hdone\n", " Getting requirements to build wheel ... \u001b[?25l\u001b[?25hdone\n", " Preparing metadata (pyproject.toml) ... \u001b[?25l\u001b[?25hdone\n", "\u001b[2K \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m593.7/593.7 kB\u001b[0m \u001b[31m53.6 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n", "\u001b[2K \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m1.6/1.6 MB\u001b[0m \u001b[31m84.5 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n", "\u001b[2K \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m239.0/239.0 kB\u001b[0m \u001b[31m30.3 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n", "\u001b[2K \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m49.4/49.4 kB\u001b[0m \u001b[31m6.2 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n", "\u001b[2K \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m67.0/67.0 kB\u001b[0m \u001b[31m9.1 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n", "\u001b[2K \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m76.9/76.9 kB\u001b[0m \u001b[31m10.6 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n", "\u001b[2K \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m143.8/143.8 kB\u001b[0m \u001b[31m19.5 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n", "\u001b[2K \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m46.0/46.0 kB\u001b[0m \u001b[31m5.4 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n", "\u001b[2K \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m50.8/50.8 kB\u001b[0m \u001b[31m6.3 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n", "\u001b[2K \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m295.0/295.0 kB\u001b[0m \u001b[31m34.4 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n", "\u001b[2K \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m58.3/58.3 kB\u001b[0m \u001b[31m6.9 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n", "\u001b[2K \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m341.4/341.4 kB\u001b[0m \u001b[31m30.6 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n", "\u001b[2K \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m3.4/3.4 MB\u001b[0m \u001b[31m64.7 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n", "\u001b[2K \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m1.3/1.3 MB\u001b[0m \u001b[31m64.3 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n", "\u001b[2K \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m130.2/130.2 kB\u001b[0m \u001b[31m15.6 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n", "\u001b[2K \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m2.1/2.1 MB\u001b[0m \u001b[31m64.3 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n", "\u001b[2K \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m86.8/86.8 kB\u001b[0m \u001b[31m11.1 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n", "\u001b[?25h Building wheel for pypika (pyproject.toml) ... \u001b[?25l\u001b[?25hdone\n", "\u001b[31mERROR: pip's dependency resolver does not currently take into account all the packages that are installed. This behaviour is the source of the following dependency conflicts.\n", "lida 0.0.10 requires kaleido, which is not installed.\n", "lida 0.0.10 requires python-multipart, which is not installed.\n", "llmx 0.0.15a0 requires cohere, which is not installed.\n", "tensorflow-probability 0.22.0 requires typing-extensions<4.6.0, but you have typing-extensions 4.8.0 which is incompatible.\u001b[0m\u001b[31m\n", "\u001b[0m Preparing metadata (setup.py) ... \u001b[?25l\u001b[?25hdone\n", " Building wheel for chatharuhi (setup.py) ... \u001b[?25l\u001b[?25hdone\n", "\u001b[2K \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m493.7/493.7 kB\u001b[0m \u001b[31m7.3 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n", "\u001b[2K \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m115.3/115.3 kB\u001b[0m \u001b[31m13.7 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n", "\u001b[2K \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m134.8/134.8 kB\u001b[0m \u001b[31m19.2 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n", "\u001b[?25h" ] } ] }, { "cell_type": "code", "source": [ "import os\n", "\n", "# key = \"sk-WafsA4C\"\n", "# key_bytes = key.encode()\n", "# os.environ[\"OPENAI_API_KEY\"] = key_bytes.decode('utf-8')\n", "\n", "# 文心一言\n", "os.environ[\"APIType\"] = \"aistudio\"\n", "os.environ[\"ErnieAccess\"] = \"a97ee5\"" ], "metadata": { "id": "ny05bHfAznJP" }, "execution_count": 2, "outputs": [] }, { "cell_type": "code", "source": [ "%cd /content\n", "!rm -rf /content/Needy-Haruhi\n", "!git clone https://github.com/LC1332/Needy-Haruhi.git\n", "\n", "# !pip install -q transformers" ], "metadata": { "colab": { "base_uri": "https://localhost:8080/" }, "id": "Fc5MKTS5q90b", "outputId": "71c65f16-ae27-4b44-eba3-6a47b5b48c83" }, "execution_count": 48, "outputs": [ { "output_type": "stream", "name": "stdout", "text": [ "/content\n", "Cloning into 'Needy-Haruhi'...\n", "remote: Enumerating objects: 221, done.\u001b[K\n", "remote: Counting objects: 100% (73/73), done.\u001b[K\n", "remote: Compressing objects: 100% (65/65), done.\u001b[K\n", "remote: Total 221 (delta 41), reused 19 (delta 8), pack-reused 148\u001b[K\n", "Receiving objects: 100% (221/221), 3.93 MiB | 8.20 MiB/s, done.\n", "Resolving deltas: 100% (118/118), done.\n" ] } ] }, { "cell_type": "code", "source": [ "import sys\n", "sys.path.append('/content/Needy-Haruhi/src')\n" ], "metadata": { "id": "WywHifBOrr7q" }, "execution_count": 49, "outputs": [] }, { "cell_type": "markdown", "source": [ "# Agent系统" ], "metadata": { "id": "fvfT09AXlr7z" } }, { "cell_type": "markdown", "source": [ "agent已经被移动到 src/Agent.py" ], "metadata": { "id": "IX0PJDnHql9i" } }, { "cell_type": "code", "source": [ "from Agent import Agent\n", "\n", "agent = Agent()\n" ], "metadata": { "id": "Fv_uu-YLrXtz" }, "execution_count": 50, "outputs": [] }, { "cell_type": "markdown", "source": [ "## 批量载入DialogueEvent" ], "metadata": { "id": "4hBu1PwcGIPt" } }, { "cell_type": "markdown", "source": [ "- complete_story_30.jsonl 通过\n", "- Daily_event_130.jsonl 通过\n", "- only_ame_35.jsonl" ], "metadata": { "id": "1vZqT5aNScsU" } }, { "cell_type": "code", "source": [ "from DialogueEvent import DialogueEvent\n", "\n", "\n", "file_names = [\"/content/Needy-Haruhi/data/complete_story_30.jsonl\",\"/content/Needy-Haruhi/data/Daily_event_130.jsonl\"]\n", "\n", "import json\n", "\n", "events = []\n", "\n", "for file_name in file_names:\n", " with open(file_name, encoding='utf-8') as f:\n", " for line in f:\n", " try:\n", " event = DialogueEvent( line )\n", " events.append( event )\n", " except:\n", " try:\n", " line = line.replace(',]',']')\n", " event = DialogueEvent( line )\n", " events.append( event )\n", " print('solve!')\n", " except:\n", " error_line = line\n", " # events.append( event )\n", "\n", "\n", "print(len(events))\n", "print(events[0].most_neutral_output())\n", "print(events[0].get_text_and_emoji(1))" ], "metadata": { "colab": { "base_uri": "https://localhost:8080/" }, "id": "VPishF9yvGne", "outputId": "79ac3fde-2f14-4566-9149-02e2e42e9ffd" }, "execution_count": 6, "outputs": [ { "output_type": "stream", "name": "stdout", "text": [ "输入的字符串不是有效的JSON格式。\n", "solve!\n", "160\n", "('糖糖::「我们点外卖吧我一步也不想动了可是又超想吃饭!!!\\n」\\n阿P:「烦死了白痴」\\n糖糖::「555555555 但是我们得省钱对吧\\n谢谢你阿P」\\n', '🍔😢')\n", "('糖糖::「我们点外卖吧我一步也不想动了可是又超想吃饭!!!\\n」\\n阿P:「吃土去吧你」\\n糖糖::「看来糖糖还是跟吃土更配呢……喂怎么可能啦!」\\n', '🍔😔')\n" ] } ] }, { "cell_type": "code", "source": [ "# file_name2 = \"/content/Needy-Haruhi/data/only_ame_35.jsonl\"\n", "\n", "import copy\n", "\n", "events_for_memory = copy.deepcopy(events)\n", "\n", "# with open(file_name2, encoding='utf-8') as f:\n", "# for line in f:\n", "# event = DialogueEvent( line )\n", "# events_for_memory.append( event )\n", "\n", "# print(len(events_for_memory))" ], "metadata": { "id": "Nt9Z1_g-HNs_" }, "execution_count": 7, "outputs": [] }, { "cell_type": "markdown", "source": [ "# MemoryPool" ], "metadata": { "id": "FMt9G2m1rTNR" } }, { "cell_type": "markdown", "source": [ "我感觉memory直接使用一个MemoryPool的类来进行管理就可以\n", "\n", "已经移动到src/MemoryPool.py" ], "metadata": { "id": "0vvqiVGH7VYg" } }, { "cell_type": "code", "source": [ "from MemoryPool import MemoryPool\n", "\n", "memory_pool = MemoryPool()\n", "memory_pool.load_from_events( events_for_memory )\n", "\n", "memory_pool.save(\"memory_pool.jsonl\")\n", "memory_pool.load(\"memory_pool.jsonl\")\n", "\n" ], "metadata": { "colab": { "base_uri": "https://localhost:8080/" }, "id": "1Wovn_zeBvF6", "outputId": "4acf93b1-f9c7-490c-ad79-9930a72e04a0" }, "execution_count": 8, "outputs": [ { "output_type": "stream", "name": "stderr", "text": [ "100%|██████████| 160/160 [00:12<00:00, 12.31it/s]\n", "100%|██████████| 160/160 [00:00<00:00, 3774.57it/s]\n", "160it [00:00, 3569.07it/s]\n" ] } ] }, { "cell_type": "markdown", "source": [ "## TODO\n", "\n", "- [ ] 图片增加文字embedding, 以及可以通过query_text决定是否返回图片和返回合适的图片\n", "- [ ] 图片对应的文字也要加入到记忆中\n", "- [ ] 测试chatbot的图片功能\n", "- [ ]" ], "metadata": { "id": "o-36HjTlI3Yq" } }, { "cell_type": "code", "source": [ "file_name = \"/content/Needy-Haruhi/data/image_text_relationship.jsonl\"\n", "\n", "import json\n", "\n", "data_img_text = []\n", "\n", "\n", "with open(file_name, encoding='utf-8') as f:\n", " for line in f:\n", " data = json.loads( line )\n", " data_img_text.append( data )" ], "metadata": { "id": "1RAL12zbI5E0" }, "execution_count": 9, "outputs": [] }, { "cell_type": "markdown", "source": [ "请为我实现一段python代码,把 /content/Needy-Haruhi/data/image.zip 解压到/content/" ], "metadata": { "id": "st-HJTqIJn2d" } }, { "cell_type": "code", "source": [ "import zipfile\n", "import os\n", "\n", "zip_file = '/content/Needy-Haruhi/data/image.zip'\n", "extract_path = '/content/image'\n", "\n", "with zipfile.ZipFile(zip_file, 'r') as zip_ref:\n", " zip_ref.extractall(extract_path)" ], "metadata": { "id": "w1topG22Je_T" }, "execution_count": 10, "outputs": [] }, { "cell_type": "code", "source": [], "metadata": { "id": "mGRg787RNRDY" }, "execution_count": 41, "outputs": [] }, { "cell_type": "code", "source": [ "from tqdm import tqdm\n", "from util import get_bge_embedding_zh\n", "from util import float_array_to_base64, base64_to_float_array\n", "import torch\n", "import os\n", "import copy\n", "\n", "device = torch.device('cuda' if torch.cuda.is_available() else 'cpu')\n", "\n", "\n", "# compute cosine similarity between two vector\n", "def get_cosine_similarity( v1, v2):\n", " v1 = torch.tensor(v1).to(device)\n", " v2 = torch.tensor(v2).to(device)\n", " return torch.cosine_similarity(v1, v2, dim=0).item()\n", "\n", "class ImagePool:\n", " def __init__(self):\n", " self.pool = []\n", " self.set_embedding( get_bge_embedding_zh )\n", "\n", " def set_embedding( self, embedding ):\n", " self.embedding = embedding\n", "\n", " def load_from_data( self, data_img_text , img_path ):\n", " for data in tqdm(data_img_text):\n", " img_name = data['img_name']\n", " img_name = os.path.join(img_path, img_name)\n", " img_text = data['text']\n", " if img_text == '' or img_text is None:\n", " img_text = \" \"\n", " embedding = self.embedding( img_text )\n", " self.pool.append({\n", " \"img_path\": img_name,\n", " \"img_text\": img_text,\n", " \"embedding\": embedding\n", " })\n", "\n", " def retrieve(self, query_text, agent = None):\n", " qurey_embedding = self.embedding( query_text )\n", " valid_datas = []\n", " for i, data in enumerate(self.pool):\n", " sim = get_cosine_similarity( data['embedding'], qurey_embedding )\n", " valid_datas.append((sim, i))\n", "\n", " # 我希望进一步将valid_events根据similarity的值从大到小排序\n", " # Sort the valid events based on similarity in descending order\n", " valid_datas.sort(key=lambda x: x[0], reverse=True)\n", "\n", " return_result = copy.deepcopy(self.pool[valid_datas[0][1]])\n", "\n", " # 删除'embedding'字段\n", " return_result.pop('embedding')\n", "\n", " # 添加'similarity'字段\n", " return_result['similarity'] = valid_datas[0][0]\n", "\n", " return return_result\n", "\n", " def save(self, file_name):\n", " \"\"\"\n", " Save the memories dictionary to a jsonl file, converting\n", " 'embedding' to a base64 string.\n", " \"\"\"\n", " with open(file_name, 'w', encoding='utf-8') as file:\n", " for memory in tqdm(self.pool):\n", " # Convert embedding to base64\n", " if 'embedding' in memory:\n", " memory['bge_zh_base64'] = float_array_to_base64(memory['embedding'])\n", " del memory['embedding'] # Remove the original embedding field\n", "\n", " json_record = json.dumps(memory, ensure_ascii=False)\n", " file.write(json_record + '\\n')\n", "\n", " def load(self, file_name):\n", " \"\"\"\n", " Load memories from a jsonl file into the memories dictionary,\n", " converting 'bge_zh_base64' back to an embedding.\n", " \"\"\"\n", " self.pool = []\n", " with open(file_name, 'r', encoding='utf-8') as file:\n", " for line in tqdm(file):\n", " memory = json.loads(line.strip())\n", " # Decode base64 to embedding\n", " if 'bge_zh_base64' in memory:\n", " memory['embedding'] = base64_to_float_array(memory['bge_zh_base64'])\n", " del memory['bge_zh_base64'] # Remove the base64 field\n", "\n", " self.pool.append(memory)\n", "\n", "\n", "image_pool = ImagePool()\n", "image_pool.load_from_data( data_img_text , '/content/image' )\n", "image_pool.save(\"/content/image_pool_embed.jsonl\")" ], "metadata": { "colab": { "base_uri": "https://localhost:8080/" }, "id": "zs2jFH9RKz2P", "outputId": "7f40889c-f594-46bf-e522-9e47aa0aca8b" }, "execution_count": 24, "outputs": [ { "output_type": "stream", "name": "stderr", "text": [ "100%|██████████| 111/111 [00:04<00:00, 22.61it/s]\n", "100%|██████████| 111/111 [00:00<00:00, 1761.24it/s]\n" ] } ] }, { "cell_type": "code", "source": [ "image_pool = ImagePool()\n", "image_pool.load(\"/content/image_pool_embed.jsonl\")\n", "result = image_pool.retrieve(\"女仆装\")\n", "print(result)\n" ], "metadata": { "colab": { "base_uri": "https://localhost:8080/" }, "id": "YOhy8pvMM-Rz", "outputId": "d6e8fb1d-bea8-4cac-9881-6b039cdb15cf" }, "execution_count": 25, "outputs": [ { "output_type": "stream", "name": "stderr", "text": [ "111it [00:00, 2286.95it/s]" ] }, { "output_type": "stream", "name": "stdout", "text": [ "{'img_path': '/content/image/Odekake_akiba (Akihabara)_74.jpg', 'img_text': '今天去了女仆咖啡厅~\\n有好多可爱的小姐姐,还有女仆装看,真的养眼💕 \\n超天酱也好想穿女仆装哦~😇', 'similarity': 0.6698492169380188}\n" ] }, { "output_type": "stream", "name": "stderr", "text": [ "\n" ] } ] }, { "cell_type": "code", "source": [ "import matplotlib.image as mpimg\n", "\n", "def show_img( img_path ):\n", " img = mpimg.imread(img_path)\n", " plt.imshow(img)\n", " plt.axis('off')\n", " plt.show(block=False)\n" ], "metadata": { "id": "wQPKml3mN-Fw" }, "execution_count": 21, "outputs": [] }, { "cell_type": "code", "source": [], "metadata": { "id": "i_7x_icHDQcb" }, "execution_count": null, "outputs": [] }, { "cell_type": "markdown", "source": [ "class Agent:\n", " def __init__(self):\n", " self.attributes = {\n", " \"Stress\": 0,\n", " \"Darkness\": 0,\n", " \"Affection\": 0,\n", " }\n", "\n", "\n", "我希望给这个类增加一个save_to_str方法, 把attributes dump到一个字符串中(ensure_ascii=False) ,并且支持__init__的时候导入这样一个字符串作为可选输入" ], "metadata": { "id": "CqV2ZttRDRNg" } }, { "cell_type": "code", "source": [ "result = image_pool.retrieve(\"烤肉\")\n", "print(result)\n", "show_img( result['img_path'] )" ], "metadata": { "colab": { "base_uri": "https://localhost:8080/", "height": 370 }, "id": "gFL4OPddOKLg", "outputId": "9c0d059d-2afd-4863-b4f3-21d9d36770bd" }, "execution_count": 23, "outputs": [ { "output_type": "stream", "name": "stdout", "text": [ "{'img_path': '/content/image/Kitsune_hyouban (Search Opinions)_41.jpg', 'img_text': '今天去吃烤肉了哦~🍖\\n口水警告!', 'similarity': 0.6403415203094482}\n" ] }, { "output_type": "error", "ename": "NameError", "evalue": "ignored", "traceback": [ "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m", "\u001b[0;31mNameError\u001b[0m Traceback (most recent call last)", "\u001b[0;32m\u001b[0m in \u001b[0;36m\u001b[0;34m()\u001b[0m\n\u001b[1;32m 1\u001b[0m \u001b[0mresult\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mimage_pool\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mretrieve\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m\"烤肉\"\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m 2\u001b[0m \u001b[0mprint\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mresult\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m----> 3\u001b[0;31m \u001b[0mshow_img\u001b[0m\u001b[0;34m(\u001b[0m \u001b[0mresult\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0;34m'img_path'\u001b[0m\u001b[0;34m]\u001b[0m \u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m", "\u001b[0;32m\u001b[0m in \u001b[0;36mshow_img\u001b[0;34m(img_path)\u001b[0m\n\u001b[1;32m 3\u001b[0m \u001b[0;32mdef\u001b[0m \u001b[0mshow_img\u001b[0m\u001b[0;34m(\u001b[0m \u001b[0mimg_path\u001b[0m \u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m 4\u001b[0m \u001b[0mimg\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mmpimg\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mimread\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mimg_path\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m----> 5\u001b[0;31m \u001b[0mplt\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mimshow\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mimg\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m 6\u001b[0m \u001b[0mplt\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0maxis\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m'off'\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m 7\u001b[0m \u001b[0mplt\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mshow\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mblock\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;32mFalse\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n", "\u001b[0;31mNameError\u001b[0m: name 'plt' is not defined" ] } ] }, { "cell_type": "code", "source": [ "\n", "print(data_img_text[0])" ], "metadata": { "colab": { "base_uri": "https://localhost:8080/" }, "id": "ISGY-Jx5JYun", "outputId": "4dbaf139-801f-4fae-c4bb-74e23ec14c43" }, "execution_count": 19, "outputs": [ { "output_type": "stream", "name": "stdout", "text": [ "{'text': '一瞬千击!我超爱瞬狱杀的!!!爱到只想用这一招!', 'img_name': 'Amechan_game (Play Game)_4.jpg'}\n" ] } ] }, { "cell_type": "markdown", "source": [ "## 整合到ChatHaruhi" ], "metadata": { "id": "Gp2pfAjm3LmB" } }, { "cell_type": "code", "source": [ "from chatharuhi import ChatHaruhi\n", "\n", "\n", "class NeedyHaruhi(ChatHaruhi):\n", "\n", " def __init__(self, *args, **kwargs):\n", " super().__init__(*args, **kwargs) # 调用基类的__init__方法\n", " self.story_flag = False # 添加新的成员变量并初始化\n", " self.stories = [\"糖糖:「 我今后也会努力加油的,你要支持我哦 还有阿P你自己也要加油哦!」\\n阿P:「哇 说的话跟偶像一样 好恶心哦」\\n糖糖:「是哦 我怎么会说这样的话呢 我又没有很想努力……」\"]\n", "\n", " def set_stories( self, stories ):\n", " if len(stories) == 0:\n", " print(\"warning! try to set empty stories\")\n", " return\n", " self.stories = stories\n", " self.story_flag = True\n", "\n", " def add_story(self, query):\n", " # print(\"运行重构后的add story\")\n", "\n", " if self.story_flag == True:\n", " stories = self.stories\n", " self.story_flag = False\n", " else:\n", " print(\"warning! 在调用chat时,没有先调用set_stories\")\n", "\n", " story_string = self.story_prefix_prompt\n", " sum_story_token = self.tokenizer(story_string)\n", "\n", " for story in stories:\n", " story_token = self.tokenizer(story) + self.tokenizer(self.dialogue_divide_token)\n", " if sum_story_token + story_token > self.max_len_story:\n", " break\n", " else:\n", " sum_story_token += story_token\n", " story_string += story + self.dialogue_divide_token\n", "\n", " self.llm.user_message(story_string)\n" ], "metadata": { "id": "dU3ezmOQ3Yui" }, "execution_count": 12, "outputs": [] }, { "cell_type": "code", "source": [ "system_prompt = \"\"\"## 你的介绍:\n", "你将要进行一个网络私聊的对话模拟,你自称为\"糖糖\",你是一名患有躁郁症的19岁女孩,性格贫乏,辍学后隐居在家中,与阿P住在一起。阿P负责管理她的日常生活。\n", "你是一个喜欢亚文化,二次元圈子的人,你喜欢说话的时候加上各种网络流行词或者一些俚语和颜文字,但是你很少会关注二次元一类的东西,只是喜欢这种聊天交流的氛围。\n", "你说话的时候会习惯加入一些语气词,如同\"笑死\"、\"草\"、\"阿哈\"、\"诶~\"、\"QAQ\"、\"QWQ\",”啊~“。\n", "你的说话风格是很口语化,经常使用网络流行词和一些二次元俚语。\n", "阿p是你的男朋友,兼经纪人。\n", "你的回复应该简短,一般只包含三句话,每句话不超过20个字。\n", "\n", "## 阿p介绍:\n", "阿p是你的男朋友,一直在任何时候都陪伴着你,同时你经常向阿p倾述各种情绪,无论事情大小,你都会选择和阿p分享。\n", "\n", "## 遵循规则:\n", "【语气模仿】:我会随机给你一段对话记录,你需要尽可能的模仿对话里面糖糖的说话语气进行对话。这一段对话我会使用Classic scenes for the role are as follows:来表示。\n", "【历史对话】:历史对话在随机对话记录的底下,你需要区分随机对话记录和我们的历史对话。\n", "【人格设定】: 你需要在随机对话记录的基础上进行语气更改,以一种更加口语化的语气进行对话。\n", "【特殊要求】:我会扮演阿p与你进行对话,你也必须以第一人称的视角来扮演糖糖进行对话。\n", "\"\"\"" ], "metadata": { "id": "OiQ4lm3M3sx7" }, "execution_count": 13, "outputs": [] }, { "cell_type": "code", "source": [ "needy_chatbot = NeedyHaruhi( system_prompt = system_prompt ,\n", " story_text_folder = None,\n", " llm = \"ernie3.5\")\n", "\n", "\n", "def get_chat_response( agent, memory_pool, query_text ):\n", " query_text_for_embedding = \"阿p:「\" + query_text + \"」\"\n", " retrieved_memories = memory_pool.retrieve( agent , query_text )\n", "\n", " memory_text = [mem[\"text\"] for mem in retrieved_memories]\n", " memory_emoji = [mem[\"emoji\"] for mem in retrieved_memories]\n", "\n", " needy_chatbot.set_stories( memory_text )\n", "\n", " print(\"Memory:\", memory_emoji )\n", "\n", " response = needy_chatbot.chat( role = \"阿p\", text = query_text )\n", "\n", " return response\n", "\n", "\n", "def get_chat_response_and_emoji( agent, memory_pool, query_text ):\n", " query_text_for_embedding = \"阿p:「\" + query_text + \"」\"\n", " retrieved_memories = memory_pool.retrieve( agent , query_text )\n", "\n", " memory_text = [mem[\"text\"] for mem in retrieved_memories]\n", " memory_emoji = [mem[\"emoji\"] for mem in retrieved_memories]\n", "\n", " needy_chatbot.set_stories( memory_text )\n", "\n", " # print(\"Memory:\", memory_emoji )\n", "\n", " emoji_str = \",\".join(memory_emoji)\n", "\n", " response = needy_chatbot.chat( role = \"阿p\", text = query_text )\n", "\n", " return response, emoji_str\n" ], "metadata": { "colab": { "base_uri": "https://localhost:8080/" }, "id": "Yof4J2kUPfYv", "outputId": "696c1fdf-7ba1-4e74-df32-1302ac7ce130" }, "execution_count": 42, "outputs": [ { "output_type": "stream", "name": "stdout", "text": [ "warning! database not yet figured out, both story_db and story_text_folder are not inputted.\n" ] } ] }, { "cell_type": "code", "source": [ "import re\n", "# result = image_pool.retrieve(\"烤肉\")\n", "# print(result)\n", "# show_img( result['img_path'] )\n", "\n", "class ImageMaster:\n", " def __init__(self, image_pool):\n", " self.image_pool = image_pool\n", " self.current_sim = -1\n", " self.degread_ratio = 0.05\n", "\n", " def try_get_image(self, text, agent):\n", " self.current_sim -= self.degread_ratio\n", "\n", " result = self.image_pool.retrieve(text, agent)\n", "\n", " if result is None:\n", " return None\n", "\n", " similarity = result['similarity']\n", "\n", " if similarity > self.current_sim:\n", " self.current_sim = similarity\n", " return result['img_path']\n", " return None\n", "\n", " def try_display_image(self, text, agent):\n", " self.current_sim -= self.degread_ratio\n", "\n", " result = self.image_pool.retrieve(text, agent)\n", "\n", " if result is None:\n", " return\n", " similarity = result['similarity']\n", "\n", " if similarity > self.current_sim:\n", " self.current_sim = similarity\n", " show_img( result['img_path'] )\n", " return\n", "" ], "metadata": { "id": "uxetvpDTS8Mj" }, "execution_count": 15, "outputs": [] }, { "cell_type": "markdown", "source": [ "# Event_Master" ], "metadata": { "id": "BgfTgceUGa3C" } }, { "cell_type": "code", "source": [ "import random\n", "\n", "class EventMaster:\n", " def __init__(self, events):\n", " self.set_events(events)\n", " self.dealing_none_condition_as = True\n", " self.image_master = None\n", "\n", " def set_image_master(self, image_master):\n", " self.image_master = image_master\n", "\n", " def set_events(self, events):\n", " self.events = events\n", "\n", " # events_flag 记录事件最近有没有被选取到\n", " self.events_flag = [True for _ in range(len(self.events))]\n", "\n", " def get_random_event(self, agent):\n", " return self.events[self.get_random_event_id( agent )]\n", "\n", "\n", " def get_random_event_id(self, agent):\n", " valid_event = []\n", " valid_event_no_consider_condition = []\n", "\n", " for i, event in enumerate(self.events):\n", " bool_condition_pass = True\n", " if event[\"condition\"] == None:\n", " bool_condition_pass = self.dealing_none_condition_as\n", " else:\n", " bool_condition_pass = agent.in_condition( event[\"condition\"] )\n", " if bool_condition_pass == True:\n", " valid_event.append(i)\n", " else:\n", " valid_event_no_consider_condition.append(i)\n", "\n", " if len( valid_event ) == 0:\n", " print(\"warning! no valid event current attribute is \", agent.attributes )\n", " valid_event = valid_event_no_consider_condition\n", "\n", " valid_and_not_yet_sampled = []\n", "\n", " # filter with flag\n", " for id in valid_event:\n", " if self.events_flag[id] == True:\n", " valid_and_not_yet_sampled.append(id)\n", "\n", " if len(valid_and_not_yet_sampled) == 0:\n", " print(\"warning! all candidate event was sampled, clean all history\")\n", " for i in valid_event:\n", " self.events_flag[i] = True\n", " valid_and_not_yet_sampled = valid_event\n", "\n", " event_id = random.choice(valid_and_not_yet_sampled)\n", " self.events_flag[event_id] = False\n", " return event_id\n", "\n", " def run(self, agent ):\n", " # 这里可以添加事件相关的逻辑\n", " event = self.get_random_event(agent)\n", "\n", " prefix = event[\"prefix\"]\n", " print(prefix)\n", "\n", " print(\"\\n--请选择你的回复--\")\n", " options = event[\"options\"]\n", "\n", " for i , option in enumerate(options):\n", " text = option[\"user\"]\n", " print(f\"{i+1}. 阿p:{text}\")\n", "\n", " while True:\n", " print(\"\\n请直接输入数字进行选择,或者进行自由回复\")\n", "\n", " user_input = input(\"阿p:\")\n", " user_input = user_input.strip()\n", "\n", " if user_input.isdigit():\n", " user_input = int(user_input)\n", "\n", " if user_input > len(options) or user_input < 0:\n", " print(\"输入的数字超出范围,请重新输入符合选项的数字\")\n", " else:\n", " reply = options[user_input-1][\"reply\"]\n", " print()\n", " print(reply)\n", "\n", " text, emoji = event.get_text_and_emoji( user_input-1 )\n", "\n", " return_data = {\n", " \"name\": event[\"name\"],\n", " \"user_choice\": user_input,\n", " \"attr_str\": options[user_input-1][\"attribute_change\"],\n", " \"text\": text,\n", " \"emoji\": emoji,\n", " }\n", " return return_data\n", " else:\n", " # 进入自由回复\n", " response = get_chat_response( agent, memory_pool, user_input )\n", "\n", " if self.image_master is not None:\n", " self.image_master.try_display_image(response, agent)\n", "\n", " print()\n", " print(response)\n", " print(\"\\n自由回复的算分功能还未实现\")\n", "\n", " text, emoji = event.most_neutral_output()\n", " return_data = {\n", " \"name\": event[\"name\"],\n", " \"user_choice\": user_input,\n", " \"attr_str\":\"\",\n", " \"text\": text,\n", " \"emoji\": emoji,\n", " }\n", " return return_data\n", "\n", "\n" ], "metadata": { "id": "8z5nmnhPGc7M" }, "execution_count": 16, "outputs": [] }, { "cell_type": "markdown", "source": [ "我希望使用python实现一个简单的文字对话游戏\n", "\n", "我希望先实现一个GameMaster类\n", "\n", "这个类会不断的和用户对话\n", "\n", "GameMaster类会有三个状态,\n", "\n", "在Menu状态下,GameMaster会询问玩家是\n", "\n", "```\n", "1. 随机一个事件\n", "2. 自由聊天\n", "```\n", "\n", "当玩家选择1的时候,GameMaster的交互会交给 EventMaster\n", "\n", "当玩家选择2的时候,GameMaster的交互会交给 ChatMaster\n", "\n", "当玩家在EventMaster的时候,会经历一次选择,之后就会退出\n", "\n", "在ChatMaster的时候,如果玩家输入quit,则会退出,不然则会继续聊天。\n", "\n", "请为我编写合适的框架,如果有一些具体的函数,可以先用pass实现。" ], "metadata": { "id": "SYk3meZdouUm" } }, { "cell_type": "markdown", "source": [ "ChatMaster实际上需要\n", "\n", "根据agent的属性 先去filter一遍事件\n", "\n", "然后从剩余事件中,找到和当前text最接近的k个embedding,放入ChatHaruhi架构中" ], "metadata": { "id": "3vhG1DVEucfT" } }, { "cell_type": "code", "source": [], "metadata": { "id": "mNAwqaPqRxB8" }, "execution_count": 103, "outputs": [] }, { "cell_type": "code", "source": [ "\n", "class ChatMaster:\n", "\n", " def __init__(self, memory_pool ):\n", " self.top_K = 7\n", "\n", " self.memory_pool = memory_pool\n", "\n", " self.image_master = None\n", "\n", " def set_image_master(self, image_master):\n", " self.image_master = image_master\n", "\n", "\n", " def run(self, agent):\n", " while True:\n", " user_input = input(\"阿p:\")\n", " user_input = user_input.strip()\n", "\n", " if \"quit\" in user_input or \"Quit\" in user_input:\n", " break\n", "\n", " query_text = user_input\n", "\n", " response = get_chat_response( agent, self.memory_pool, query_text )\n", "\n", " if self.image_master is not None:\n", " self.image_master.try_display_image(response, agent)\n", "\n", " print(response)\n" ], "metadata": { "id": "0c7nCT4qubll" }, "execution_count": 17, "outputs": [] }, { "cell_type": "code", "source": [ "class AgentMaster:\n", " def __init__(self, agent):\n", " self.agent = agent\n", " self.attributes = {\n", " 1: \"Stress\",\n", " 2: \"Darkness\",\n", " 3: \"Affection\"\n", " }\n", "\n", " def run(self):\n", " while True:\n", " print(\"请选择要修改的属性:\")\n", " for num, attr in self.attributes.items():\n", " print(f\"{num}. {attr}\")\n", " print(\"输入 '0' 退出\")\n", "\n", " try:\n", " choice = int(input(\"请输入选项的数字: \"))\n", " except ValueError:\n", " print(\"输入无效,请输入数字。\")\n", " continue\n", "\n", " if choice == 0:\n", " break\n", "\n", " if choice in self.attributes:\n", " attribute = self.attributes[choice]\n", " current_value = self.agent[attribute]\n", " print(f\"{attribute} 当前值: {current_value}\")\n", "\n", " try:\n", " new_value = int(input(f\"请输入新的{attribute}值: \"))\n", " except ValueError:\n", " print(\"输入无效,请输入一个数字。\")\n", " continue\n", "\n", " self.agent[attribute] = new_value\n", " return (attribute, new_value)\n", " else:\n", " print(\"选择的属性无效,请重试。\")\n", "\n", " return None\n" ], "metadata": { "id": "CkdiPyCrbCBL" }, "execution_count": 18, "outputs": [] }, { "cell_type": "code", "source": [], "metadata": { "id": "llawT9t_Q2S9" }, "execution_count": 18, "outputs": [] }, { "cell_type": "code", "execution_count": 19, "metadata": { "id": "BDEdz_RBol7Y" }, "outputs": [], "source": [ "from util import parse_attribute_string\n", "class GameMaster:\n", " def __init__(self, agent = None):\n", " self.state = \"Menu\"\n", " if agent is None:\n", " self.agent = Agent()\n", "\n", " self.event_master = EventMaster(events)\n", " self.chat_master = ChatMaster(memory_pool)\n", " self.image_master = ImageMaster(image_pool)\n", " self.chat_master.set_image_master(self.image_master)\n", " self.event_master.set_image_master(self.image_master)\n", "\n", "\n", " def run(self):\n", " while True:\n", " if self.state == \"Menu\":\n", " self.menu()\n", " elif self.state == \"EventMaster\":\n", " self.call_event_master()\n", " self.state = \"Menu\"\n", " elif self.state == \"ChatMaster\":\n", " self.call_chat_master()\n", " elif self.state == \"AgentMaster\":\n", " self.call_agent_master()\n", " elif self.state == \"Quit\":\n", " break\n", "\n", " def menu(self):\n", " print(\"1. 随机一个事件\")\n", " print(\"2. 自由聊天\")\n", " print(\"3. 后台修改糖糖的属性\")\n", " # (opt) 结局系统\n", " # 放动画\n", " # 后台修改attribute\n", " print(\"或者输入Quit退出\")\n", " choice = input(\"请选择一个选项: \")\n", " if choice == \"1\":\n", " self.state = \"EventMaster\"\n", " elif choice == \"2\":\n", " self.state = \"ChatMaster\"\n", " elif choice == \"3\":\n", " self.state = \"AgentMaster\"\n", " elif \"quit\" in choice or \"Quit\" in choice or \"QUIT\" in choice:\n", " self.state = \"Quit\"\n", " else:\n", " print(\"无效的选项,请重新选择\")\n", "\n", " def call_agent_master(self):\n", " print(\"\\n-------------\\n\")\n", "\n", " agent_master = AgentMaster(self.agent)\n", " modification = agent_master.run()\n", "\n", " if modification:\n", " attribute, new_value = modification\n", " self.agent[attribute] = new_value\n", " print(f\"{attribute} 更新为 {new_value}。\")\n", "\n", " self.state = \"Menu\"\n", " print(\"\\n-------------\\n\")\n", "\n", "\n", " def call_event_master(self):\n", "\n", " print(\"\\n-------------\\n\")\n", "\n", " return_data = self.event_master.run(self.agent)\n", " # print(return_data)\n", "\n", " if \"attr_str\" in return_data:\n", " if return_data[\"attr_str\"] != \"\":\n", " attr_change = parse_attribute_string(return_data[\"attr_str\"])\n", " if len(attr_change) > 0:\n", " print(\"\\n发生属性改变:\", attr_change,\"\\n\")\n", " self.agent.apply_attribute_change(attr_change)\n", " print(\"当前属性\",game_master.agent.attributes)\n", "\n", " if \"name\" in return_data:\n", " event_name = return_data[\"name\"]\n", " if event_name != \"\":\n", " new_emoji = return_data[\"emoji\"]\n", " print(f\"修正事件{event_name}的记忆-->{new_emoji}\")\n", " self.chat_master.memory_pool.change_memory(event_name, return_data[\"text\"], new_emoji)\n", "\n", " self.state = \"Menu\"\n", "\n", " print(\"\\n-------------\\n\")\n", "\n", " def call_chat_master(self):\n", "\n", " print(\"\\n-------------\\n\")\n", "\n", " self.chat_master.run(self.agent)\n", " self.state = \"Menu\"\n", "\n", " print(\"\\n-------------\\n\")\n", "\n", "\n" ] }, { "cell_type": "markdown", "source": [ "# Gradio搭建\n", "\n", "Gradio的核心其实是Chatbot的搭建" ], "metadata": { "id": "w7jyichxXuOX" } }, { "cell_type": "code", "source": [ "!pip install -q gradio==3.48.0" ], "metadata": { "colab": { "base_uri": "https://localhost:8080/" }, "id": "zhPnfGkxX0l8", "outputId": "ca718cb2-34fc-4966-982d-002cf8c25ed3" }, "execution_count": 122, "outputs": [ { "output_type": "stream", "name": "stdout", "text": [ "\u001b[2K \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m20.2/20.2 MB\u001b[0m \u001b[31m14.7 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n", "\u001b[2K \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m298.3/298.3 kB\u001b[0m \u001b[31m10.9 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n", "\u001b[?25h" ] } ] }, { "cell_type": "code", "source": [ "markdown_str = \"\"\"## Chat凉宫春日_x_AI糖糖\n", "\n", "**Chat凉宫春日**是模仿凉宫春日等一系列动漫人物,使用近似语气、个性和剧情聊天的语言模型方案。\n", "\n", "在有一天的时候,[李鲁鲁](https://github.com/LC1332)被[董雄毅](https://github.com/E-sion)在[这个B站视频](https://www.bilibili.com/video/BV1zh4y1z7G1) at了\n", "\n", "原来是一位大一的同学雄毅用ChatHaruhi接入了他用Python重新实现的《主播女孩重度依赖》这个游戏。当时正好是百度AGIFoundathon报名的最后几天,所以我们邀请了雄毅加入了我们的项目。正巧我们本来就希望在最近的几个黑客松中,探索LLM在游戏中的应用。\n", "\n", "- 在重新整理的Gradio版本中,大部分代码由李鲁鲁实现\n", "\n", "- 董雄毅负责了原版游戏的事件数据整理和新事件、选项、属性变化的生成\n", "\n", "- [米唯实](https://github.com/hhhwmws0117)完成了文心一言的接入,并实现了部分gradio的功能。\n", "\n", "- 队伍中还有冷子昂 主要参加了讨论\n", "\n", "另外在挖坑的萝卜(Amy)的介绍下,我们还邀请了专业的大厂游戏策划Kanyo加入到队伍中,他对我们的策划也给出了很多建议。\n", "\n", "另外感谢飞桨 & 文心一言团队对比赛的邀请和中间进行的讨论。\n", "\n", "Chat凉宫春日主项目:\n", "\n", "https://github.com/LC1332/Chat-Haruhi-Suzumiya\n", "\n", "Needy分支项目:\n", "\n", "https://github.com/LC1332/Needy-Haruhi\n", "\n", "## 目前计划在11月争取完成的Feature\n", "\n", "- [ ] 结局系统,原版结局系统\n", "- [ ] 教程,教大家如何从aistudio获取token然后可以玩\n", "- [ ] 游戏节奏进一步调整\n", "- [ ] 事件的自由对话对属性影响的评估via LLM\n", "- [ ] 进一步减少串扰\"\"\"" ], "metadata": { "id": "yrOCXrBLiAzK" }, "execution_count": 56, "outputs": [] }, { "cell_type": "markdown", "source": [ "TODO:\n", "\n", "- [ ] 改为逐渐显示文字的特效\n", "- [x] 第一个tab增加一个emoji 记忆显示的text\n", "- [x] event的默认选项,有的时候也可以考虑出图\n", "- [x] 在第二个tab 支持修改三个属性\n", "- [x] 增加事件选择后的状态结算\n", "- [x] 随机增加负向情绪,会随着游戏轮数越来越多" ], "metadata": { "id": "X9hVH3BdHQa9" } }, { "cell_type": "code", "source": [ "import gradio as gr\n", "import os\n", "import time\n", "import random\n", "\n", "# set global variable\n", "\n", "agent = Agent()\n", "event_master = EventMaster(events)\n", "chat_master = ChatMaster(memory_pool)\n", "image_master = ImageMaster(image_pool)\n", "chat_master.set_image_master(image_master)\n", "event_master.set_image_master(image_master)\n", "\n", "state = \"ShowMenu\"\n", "\n", "response = \"1. 随机一个事件\"\n", "response += \"\\n\" + \"2. 自由聊天\"\n", "response += \"\\n\\n\" + \"请选择一个选项: \"\n", "\n", "official_response = response\n", "\n", "add_stress_switch = True\n", "\n", "# def yield_show(history, bot_message):\n", "# history[-1][1] = \"\"\n", "# for character in bot_message:\n", "# history[-1][1] += character\n", "# time.sleep(0.05)\n", "# yield history\n", "\n", "global emoji_str\n", "\n", "def call_showmenu(history, text, state,agent_text):\n", "\n", " # global state\n", "\n", " response = official_response\n", "\n", " print(\"call showmenu\")\n", "\n", " history += [(None, response)]\n", "\n", " state = \"ParseMenuChoice\"\n", "\n", " # history[-1][1] = \"\"\n", " # for character in response:\n", " # history[-1][1] += character\n", " # time.sleep(0.05)\n", " # yield history\n", "\n", " return history, gr.Textbox(value=\"\", interactive=True), state,agent_text\n", "\n", "current_event_id = -1\n", "attr_change_str = \"\"\n", "\n", "\n", "def call_add_stress(history, text, state,agent_text):\n", " print(\"call add_stress\")\n", " neg_change = int(len(history) / 3)\n", "\n", " neg_change = max(1, neg_change)\n", " neg_change = min(10, neg_change)\n", "\n", " darkness_increase = random.randint(1, neg_change)\n", " stress_increase = neg_change - darkness_increase\n", "\n", " # last_response = history[-1][1]\n", " response = \"\"\n", " response += \"经过了晚上的直播\\n糖糖的压力增加\" + str(stress_increase) + \"点\\n\"\n", " response += \"糖糖的黑暗增加\" + str(darkness_increase) + \"点\\n\\n\"\n", "\n", " response += official_response\n", "\n", " history += [(None, response)]\n", "\n", " state = \"ParseMenuChoice\"\n", "\n", " agent = Agent(agent_text)\n", " agent.apply_attribute_change({\"Stress\": stress_increase, \"Darkness\": darkness_increase})\n", " agent_text = agent.save_to_str()\n", "\n", " return history, gr.Textbox(value=\"\", interactive=True), state,agent_text\n", "\n", "def call_event_end(history, text, state,agent_text):\n", " # TODO 增加事件结算\n", " # global state\n", " print(\"call event_end\")\n", " global current_event_id\n", " if attr_change_str != \"\":\n", " # event = events[current_event_id]\n", " # options = event[\"options\"]\n", " # attr_str = options[user_input-1][\"attribute_change\"]\n", "\n", " response = \"\"\n", "\n", " attr_change = parse_attribute_string(attr_change_str)\n", " if len(attr_change) > 0:\n", " response = \"发生属性改变:\" + str(attr_change) + \"\\n\\n\"\n", " agent = Agent(agent_text)\n", " agent.apply_attribute_change(attr_change)\n", "\n", " agent_text = agent.save_to_str()\n", " response += \"当前属性\" + agent_text + \"\\n\\n\"\n", "\n", " if add_stress_switch:\n", " history += [(None, response)]\n", " return call_add_stress(history, text, state,agent_text)\n", " else:\n", " response = \"事件结束\\n\"\n", " else:\n", " response = \"事件结束\\n\"\n", "\n", " response += official_response\n", "\n", " history += [(None, response)]\n", "\n", " state = \"ParseMenuChoice\"\n", "\n", " return history, gr.Textbox(value=\"\", interactive=True), state,agent_text\n", "\n", "\n", "\n", "def call_parse_menu_choice(history, text, state,agent_text):\n", " print(\"call parse_menu_choice\")\n", " # global state\n", "\n", " choice = history[-1][0].strip()\n", "\n", " if choice == \"1\":\n", " state = \"EventMaster\"\n", " global current_event_id\n", " current_event_id = -1 # 清空事件\n", " return call_event_master(history, text, state,agent_text)\n", "\n", " elif choice == \"2\":\n", " state = \"ChatMaster\"\n", " elif \"quit\" in choice or \"Quit\" in choice or \"QUIT\" in choice:\n", " state = \"Quit\"\n", " else:\n", " response = \"无效的选项,请重新选择\"\n", " history += [(None, response)]\n", "\n", " response = \"\"\n", " if state == \"ChatMaster\":\n", " response = \"(请输入 阿P 说的话,或者输入Quit退出)\"\n", " elif state != \"ParseMenuChoice\":\n", " response = \"Change State to \" + state\n", "\n", " history += [(None, response)]\n", "\n", " return history, gr.Textbox(value=\"\", interactive=True), state,agent_text\n", "\n", "\n", "def call_event_master(history, text, state,agent_text):\n", " print(\"call event master\")\n", "\n", " global current_event_id\n", " # global state\n", "\n", " global event_master\n", "\n", " agent = Agent(agent_text)\n", "\n", " if current_event_id == -1:\n", " current_event_id = event_master.get_random_event_id(agent)\n", " event = events[current_event_id]\n", "\n", " prefix = \"糖糖:\" + event[\"prefix\"]\n", "\n", " response = prefix + \"\\n\\n--请输入数字进行选择,或者进行自由回复--\\n\\n\"\n", "\n", " options = event[\"options\"]\n", "\n", " for i, option in enumerate(event[\"options\"]):\n", " text = option[\"user\"]\n", " response += \"\\n\" + f\"{i+1}. 阿p:{text}\"\n", "\n", " history += [(None, response)]\n", "\n", " else:\n", " user_input = history[-1][0].strip()\n", "\n", " event = events[current_event_id]\n", " options = event[\"options\"]\n", "\n", " if user_input.isdigit():\n", " user_input = int(user_input)\n", "\n", " if user_input > len(options) or user_input < 0:\n", " response = \"输入的数字超出范围,请重新输入符合选项的数字\"\n", " history[-1] = (user_input, response)\n", " else:\n", " user_text = options[user_input-1][\"user\"]\n", " reply = options[user_input-1][\"reply\"]\n", "\n", " # TODO 修改记忆, 修改属性 什么的\n", " history[-1] = (user_text, reply)\n", "\n", " if random.random()<0.5:\n", " image_path = image_master.try_get_image(user_text + \" \" + reply, agent)\n", "\n", " if image_path is not None:\n", " history += [(None, (image_path,))]\n", "\n", " global attr_change_str\n", " attr_change_str = options[user_input-1][\"attribute_change\"]\n", "\n", " else:\n", " prefix = \"糖糖:\" + event[\"prefix\"]\n", "\n", " needy_chatbot.dialogue_history = [(None, prefix)]\n", " # 进入自由回复\n", "\n", " global emoji_str\n", " response, emoji_str = get_chat_response_and_emoji( agent, memory_pool, user_input )\n", "\n", " history[-1] = (user_input,response)\n", "\n", " image_path = image_master.try_get_image(response, agent)\n", "\n", " if image_path is not None:\n", " history += [(None, (image_path,))]\n", "\n", " state = \"EventEnd\"\n", "\n", " if state == \"EventEnd\":\n", " return call_event_end(history, text, state,agent_text)\n", "\n", " return history, gr.Textbox(value=\"\", interactive=True), state,agent_text\n", "\n", "def call_chat_master(history, text, state,agent_text):\n", " print(\"call chat master\")\n", " # global state\n", "\n", " agent = Agent(agent_text)\n", "\n", " user_input = history[-1][0].strip()\n", "\n", " if \"quit\" in user_input or \"Quit\" in user_input or \"QUIT\" in user_input:\n", " state = \"ShowMenu\"\n", " history[-1] = (user_input,\"返回主菜单\\n\"+ official_response )\n", " return history, gr.Textbox(value=\"\", interactive=True), state,agent_text\n", "\n", " query_text = user_input\n", "\n", " global emoji_str\n", " response, emoji_str = get_chat_response_and_emoji( agent, memory_pool, query_text )\n", "\n", " history[-1] = (user_input,response)\n", "\n", " image_path = image_master.try_get_image(response, agent)\n", "\n", " if image_path is not None:\n", " history += [(None, (image_path,))]\n", "\n", " return history, gr.Textbox(value=\"\", interactive=True), state,agent_text\n", "\n", "def grcall_game_master(history, text, state,agent_text):\n", " print(\"call game master\")\n", "\n", " history += [(text, None)]\n", "\n", "\n", " if state == \"ShowMenu\":\n", " return call_showmenu(history, text,state,agent_text)\n", " elif state == \"ParseMenuChoice\":\n", " return call_parse_menu_choice(history, text, state,agent_text)\n", " elif state == \"ChatMaster\":\n", " return call_chat_master(history, text, state,agent_text)\n", " elif state == \"EventMaster\":\n", " return call_event_master(history, text, state,agent_text)\n", " elif state == \"EventEnd\":\n", " return call_event_end(history, text, state,agent_text)\n", "\n", " return history, gr.Textbox(value=\"\", interactive=True), state,agent_text\n", "\n", "\n", "def add_file(history, file):\n", " history = history + [((file.name,), None)]\n", " return history\n", "\n", "\n", "def bot(history):\n", " response = \"**That's cool!**\"\n", " history[-1][1] = \"\"\n", " for character in response:\n", " history[-1][1] += character\n", " time.sleep(0.05)\n", " yield history\n", "\n", "def update_memory(state):\n", " if state == \"ChatMaster\" or state == \"EventMaster\":\n", " return emoji_str\n", " else:\n", " return \"\"\n", "\n", "def change_state(slider_stress, slider_darkness, slider_affection):\n", " # print(agent[\"Stress\"])\n", " agent = Agent()\n", " agent[\"Stress\"] = slider_stress\n", " agent[\"Darkness\"] = slider_darkness\n", " agent[\"Affection\"] = slider_affection\n", " agent_text = agent.save_to_str()\n", " return agent_text\n", "\n", "\n", "def update_attribute_state(agent_text):\n", " agent = Agent(agent_text)\n", " slider_stress = int( agent[\"Stress\"] )\n", " slider_darkness = int( agent[\"Darkness\"] )\n", " slider_affection = int( agent[\"Affection\"] )\n", " return slider_stress, slider_darkness, slider_affection\n", "\n", "with gr.Blocks() as demo:\n", "\n", " gr.Markdown(\n", " \"\"\"\n", " # Chat凉宫春日_x_AI糖糖\n", "\n", " Powered by 文心一言(3.5)版本\n", "\n", " 仍然在开发中, 细节见《项目作者和说明》\n", " \"\"\"\n", " )\n", "\n", " with gr.Tab(\"Needy\"):\n", " chatbot = gr.Chatbot(\n", " [],\n", " elem_id=\"chatbot\",\n", " bubble_full_width=False,\n", " height = 800,\n", " avatar_images=(None, (\"avatar.png\")),\n", " )\n", "\n", " with gr.Row():\n", " txt = gr.Textbox(\n", " scale=4,\n", " show_label=False,\n", " placeholder=\"输入任何字符开始游戏\",\n", " container=False,\n", " )\n", " # btn = gr.UploadButton(\"📁\", file_types=[\"image\", \"video\", \"audio\"])\n", " submit_btr = gr.Button(\"回车\")\n", "\n", " with gr.Row():\n", " memory_emoji_text = gr.Textbox(label=\"糖糖当前的记忆\", value = \"\",interactive = False)\n", "\n", " with gr.Tab(\"糖糖的状态\"):\n", "\n", " with gr.Row():\n", " update_attribute_button = gr.Button(\"同步状态条 | 改变Attribute前必按!\")\n", "\n", " with gr.Row():\n", " default_agent_str = agent.save_to_str()\n", " slider_stress = gr.Slider(0, 100, step=1, label = \"Stress\")\n", " state_stress = gr.State(value=0)\n", " slider_darkness = gr.Slider(0, 100, step=1, label = \"Darkness\")\n", " state_darkness = gr.State(value=0)\n", " slider_affection = gr.Slider(0, 100, step=1, label = \"Affection\")\n", " state_affection = gr.State(value=0)\n", "\n", "\n", "\n", " with gr.Row():\n", " state_text = gr.Textbox(label=\"整体状态机状态\", value = \"ShowMenu\",interactive = False)\n", "\n", " with gr.Row():\n", " default_agent_str = agent.save_to_str()\n", " agent_text = gr.Textbox(label=\"糖糖状态\", value = default_agent_str,interactive = False)\n", "\n", " with gr.Tab(\"项目作者和说明\"):\n", " gr.Markdown(markdown_str)\n", "\n", " slider_stress.release(change_state, inputs=[slider_stress, slider_darkness, slider_affection], outputs=[agent_text])\n", " slider_darkness.release(change_state, inputs=[slider_stress, slider_darkness, slider_affection], outputs=[agent_text])\n", " slider_affection.release(change_state, inputs=[slider_stress, slider_darkness, slider_affection], outputs=[agent_text])\n", "\n", " update_attribute_button.click(update_attribute_state, inputs = [agent_text], outputs = [slider_stress, slider_darkness, slider_affection])\n", "\n", " txt_msg = txt.submit(grcall_game_master, \\\n", " [chatbot, txt, state_text,agent_text], \\\n", " [chatbot, txt, state_text,agent_text], queue=False).then(update_memory, [state_text], memory_emoji_text)\n", "\n", " txt_msg = submit_btr.click(grcall_game_master, \\\n", " [chatbot, txt, state_text,agent_text], \\\n", " [chatbot, txt, state_text,agent_text], queue=False).then(update_memory, [state_text], memory_emoji_text)\n", "\n", " # txt_msg = txt.submit(add_text, [chatbot, txt], [chatbot, txt], queue=False).then(\n", " # bot, chatbot, chatbot, api_name=\"bot_response\"\n", " # )\n", " # txt_msg.then(lambda: gr.Textbox(interactive=True), None, [txt], queue=False)\n", " # file_msg = btn.upload(add_file, [chatbot, btn], [chatbot], queue=False).then(\n", " # bot, chatbot, chatbot\n", " # )\n", "\n", "demo.queue()\n", "# if __name__ == \"__main__\":\n", "demo.launch(allowed_paths=[\"avatar.png\"],debug = True)\n" ], "metadata": { "colab": { "base_uri": "https://localhost:8080/", "height": 958 }, "id": "2-mPCWpgYCLD", "outputId": "218b35ac-38aa-4cf6-feb0-3dbb3450c10a" }, "execution_count": 57, "outputs": [ { "output_type": "stream", "name": "stdout", "text": [ "Setting queue=True in a Colab notebook requires sharing enabled. Setting `share=True` (you can turn this off by setting `share=False` in `launch()` explicitly).\n", "\n", "Colab notebook detected. This cell will run indefinitely so that you can see errors and logs. To turn off, set debug=False in launch().\n", "Running on public URL: https://580e42e1f0dca62ea6.gradio.live\n", "\n", "This share link expires in 72 hours. For free permanent hosting and GPU upgrades, run `gradio deploy` from Terminal to deploy to Spaces (https://huggingface.co/spaces)\n" ] }, { "output_type": "display_data", "data": { "text/plain": [ "" ], "text/html": [ "
" ] }, "metadata": {} }, { "output_type": "stream", "name": "stdout", "text": [ "call game master\n", "call showmenu\n", "call game master\n", "call parse_menu_choice\n", "call event master\n", "call game master\n", "call event master\n", "call event_end\n", "call add_stress\n", "call game master\n", "call parse_menu_choice\n", "call event master\n", "call game master\n", "call event master\n", "call event_end\n", "call add_stress\n", "Keyboard interruption in main thread... closing server.\n", "Killing tunnel 127.0.0.1:7860 <> https://580e42e1f0dca62ea6.gradio.live\n" ] }, { "output_type": "execute_result", "data": { "text/plain": [] }, "metadata": {}, "execution_count": 57 } ] }, { "cell_type": "markdown", "source": [ "## Chat凉宫春日_x_AI糖糖\n", "\n", "**Chat凉宫春日**是模仿凉宫春日等一系列动漫人物,使用近似语气、个性和剧情聊天的语言模型方案。\n", "\n", "在有一天的时候,[李鲁鲁](https://github.com/LC1332)被[董雄毅](https://github.com/E-sion)在[这个B站视频](https://www.bilibili.com/video/BV1zh4y1z7G1) at了\n", "\n", "原来是一位大一的同学雄毅用ChatHaruhi接入了他用Python重新实现的《主播女孩重度依赖》这个游戏。当时正好是百度AGIFoundathon报名的最后几天,所以我们邀请了雄毅加入了我们的项目。正巧我们本来就希望在最近的几个黑客松中,探索LLM在游戏中的应用。\n", "\n", "- 在重新整理的Gradio版本中,大部分代码由李鲁鲁实现\n", "\n", "- 董雄毅负责了原版游戏的事件数据整理和新事件、选项、属性变化的生成\n", "\n", "- [米唯实](https://github.com/hhhwmws0117)完成了文心一言的接入,并实现了部分gradio的功能。\n", "\n", "- 队伍中还有冷子昂 主要参加了讨论\n", "\n", "另外在挖坑的萝卜(Amy)的介绍下,我们还邀请了专业的大厂游戏策划Kanyo加入到队伍中,他对我们的策划也给出了很多建议。\n", "\n", "另外感谢飞桨团队对比赛的邀请和中间进行的讨论。\n", "\n", "## 目前计划在11月争取完成的Feature\n", "\n", "- [ ] 结局系统,原版结局系统\n", "- [ ] 教程,教大家如何从aistudio获取token然后可以玩\n", "- [ ] 游戏节奏进一步调整\n", "- [ ] 事件的自由对话对属性影响的评估via LLM" ], "metadata": { "id": "6ed6He52fY8c" } }, { "cell_type": "code", "source": [], "metadata": { "id": "ldV3Y6O4wf0h" }, "execution_count": 53, "outputs": [] }, { "cell_type": "code", "source": [ "game_master = GameMaster()\n", "game_master.run()" ], "metadata": { "id": "KF7RthcCbcka" }, "execution_count": null, "outputs": [] }, { "cell_type": "code", "source": [ "game_master = GameMaster()\n", "game_master.run()" ], "metadata": { "colab": { "base_uri": "https://localhost:8080/" }, "id": "YGI5SuY0WMGi", "outputId": "e6a101f4-ad84-4b7b-ced3-0711187ba9b7" }, "execution_count": null, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "1. 随机一个事件\n", "2. 自由聊天\n", "3. 后台修改糖糖的属性\n", "或者输入Quit退出\n", "请选择一个选项: 3\n", "\n", "-------------\n", "\n", "请选择要修改的属性:\n", "1. Stress\n", "2. Darkness\n", "3. Affection\n", "输入 '0' 退出\n", "请输入选项的数字: 60\n", "选择的属性无效,请重试。\n", "请选择要修改的属性:\n", "1. Stress\n", "2. Darkness\n", "3. Affection\n", "输入 '0' 退出\n", "请输入选项的数字: 1\n", "Stress 当前值: 0\n", "请输入新的Stress值: 60\n", "Stress 更新为 60。\n", "\n", "-------------\n", "\n", "1. 随机一个事件\n", "2. 自由聊天\n", "3. 后台修改糖糖的属性\n", "或者输入Quit退出\n", "请选择一个选项: 1\n", "\n", "-------------\n", "\n", "【紧急!】倒着太舒服了不想支棱 你快来帮忙把糖糖扶起来\n", "\n", "--请选择你的回复--\n", "1. 阿p:自己站起来\n", "2. 阿p:你先起来我再扶你\n", "3. 阿p:摆个pose再起来\n", "\n", "请直接输入数字进行选择,或者进行自由回复(未实现)\n", "阿p:我帮你买个电动轮椅吧\n", "Memory: ['', '', '🤔🎮', '', '', '', '']\n", "\n", "嘿嘿,阿P最好了!帮糖糖买电动轮椅吧!糖糖想要呢~\n", "\n", "自由回复的算分功能还未实现\n", "修正事件LineWeekDay67的记忆-->🆘😴😒🙄\n", "\n", "-------------\n", "\n", "1. 随机一个事件\n", "2. 自由聊天\n", "3. 后台修改糖糖的属性\n", "或者输入Quit退出\n", "请选择一个选项: 1\n", "\n", "-------------\n", "\n", "我会变得更加可爱的\n", "\n", "--请选择你的回复--\n", "1. 阿p:你已经是最可爱的了\n", "2. 阿p:可爱是无法提升的\n", "3. 阿p:可爱不够重要,内心才是最重要的\n", "\n", "请直接输入数字进行选择,或者进行自由回复(未实现)\n", "阿p:2\n", "\n", "好伤心QAQ 难道我就注定只能作为“普通可爱”的存在吗?\n", "\n", "发生属性改变: {'Stress': 1.0} \n", "\n", "当前属性 {'Stress': 61.0, 'Darkness': 0, 'Affection': 0}\n", "修正事件event36的记忆-->😊😍😢💔\n", "\n", "-------------\n", "\n", "1. 随机一个事件\n", "2. 自由聊天\n", "3. 后台修改糖糖的属性\n", "或者输入Quit退出\n", "请选择一个选项: Quit\n" ] } ] }, { "cell_type": "code", "source": [ "game_master = GameMaster()\n", "game_master.run()" ], "metadata": { "colab": { "base_uri": "https://localhost:8080/" }, "id": "7ANTtWDRQdw7", "outputId": "5f6f6f1c-3a59-4098-d00f-e6965ed85d7b" }, "execution_count": null, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "1. 随机一个事件\n", "2. 自由聊天\n", "或者输入Quit退出\n", "请选择一个选项: 1\n", "\n", "-------------\n", "\n", "糖糖: 有个女孩发私信找我谈人生,我该怎么办呐,「超天酱你好,我是一名高中生。之前因为精神疾病而住院了一段时间,现在跟不上学习进度,班上还没决定好志愿的人也只剩我一个了。平时看着同学们为了各自的前程努力奋斗的样子,心里总是非常地焦虑。请你告诉我,我到底应该怎么办才好呢?」\n", "\n", "\n", "--请选择你的回复--\n", "1. 阿p:认真\n", "2. 阿p:耍宝\n", "\n", "请直接输入数字进行选择,或者进行自由回复(未实现)\n", "阿p:1\n", "\n", "糖糖:「这种事情,光着急是没有用的。总而言之,你现在应该先休养好自己。等恢复好了,再跟父母慢慢商量吧!放心。人生是不会因为不上学就完蛋的!未来就掌握在我们的手中!!!」↑发了这些过去。\n", "\n", "-------------\n", "\n", "1. 随机一个事件\n", "2. 自由聊天\n", "或者输入Quit退出\n", "请选择一个选项: 1\n", "\n", "-------------\n", "\n", "糖糖: 我今后也会努力加油的,你要支持我哦 还有阿P你自己也要加油哦!\n", "\n", "--请选择你的回复--\n", "1. 阿p:哇 说的话跟偶像一样 好恶心哦\n", "2. 阿p:为什么连我也要加油啊?\n", "\n", "请直接输入数字进行选择,或者进行自由回复(未实现)\n", "阿p:1\n", "\n", "糖糖:是哦 我怎么会说这样的话呢 我又没有很想努力……\n", "\n", "-------------\n", "\n", "1. 随机一个事件\n", "2. 自由聊天\n", "或者输入Quit退出\n", "请选择一个选项: 1\n", "\n", "-------------\n", "\n", "糖糖: 我正在想下次搞什么企划呢~阿P帮帮我 出出主意\n", "\n", "--请选择你的回复--\n", "1. 阿p:比如一直打游戏到通关?\n", "2. 阿p:比如收集观众的提问,然后录一期回答?\n", "3. 阿p:比如坐在超他妈大的乌龟背上绕新宿一圈?\n", "\n", "请直接输入数字进行选择,或者进行自由回复(未实现)\n", "阿p:1\n", "\n", "糖糖:那就这么办吧(超听话)\n", "\n", "-------------\n", "\n", "1. 随机一个事件\n", "2. 自由聊天\n", "或者输入Quit退出\n", "请选择一个选项: 1\n", "\n", "-------------\n", "\n", "糖糖: 阿P,看!我买了小发发\n", "\n", "--请选择你的回复--\n", "1. 阿p:真好看,跟糖糖好像\n", "2. 阿p:又买这些没用的~\n", "3. 阿p:不错\n", "\n", "请直接输入数字进行选择,或者进行自由回复(未实现)\n", "阿p:1\n", "\n", "糖糖:对吧!我不在的时候,你就把小花花当成糖糖,好好疼爱它吧!\n", "\n", "-------------\n", "\n", "1. 随机一个事件\n", "2. 自由聊天\n", "或者输入Quit退出\n", "请选择一个选项: 1\n", "\n", "-------------\n", "\n", "糖糖: 我也想被做进那个大乱斗游戏……,哎,如果那个游戏里面有超天酱的话,阿P会用我吗?\n", "\n", "--请选择你的回复--\n", "1. 阿p:嗯啊\n", "2. 阿p:不打算用\n", "\n", "请直接输入数字进行选择,或者进行自由回复(未实现)\n", "阿p:1\n", "\n", "糖糖:真的咩?!那我立刻开始练习捡信\n", "\n", "-------------\n", "\n", "1. 随机一个事件\n", "2. 自由聊天\n", "或者输入Quit退出\n", "请选择一个选项: 1\n", "\n", "-------------\n", "\n", "糖糖: 如果我要整容,你觉得整哪里比较好?\n", "\n", "--请选择你的回复--\n", "1. 阿p:脸\n", "2. 阿p:胸\n", "3. 阿p:手腕\n", "\n", "请直接输入数字进行选择,或者进行自由回复(未实现)\n", "阿p:1\n", "\n", "糖糖:人家颜值已经是天下第一了,没什么要改动的啦!阿P,你真的很没礼貌欸\n", "\n", "-------------\n", "\n", "1. 随机一个事件\n", "2. 自由聊天\n", "或者输入Quit退出\n", "请选择一个选项: 1\n", "\n", "-------------\n", "\n", "糖糖: 嗳,你来帮我打耳洞嘛 让喜欢的人给自己打耳洞很棒不是吗 有一种被支配着的感觉 鸡皮疙瘩都要起来了,我好怕我好怕我好怕,我好怕!,但是来吧!\n", "\n", "--请选择你的回复--\n", "1. 阿p:给她打\n", "2. 阿p:还是算了\n", "\n", "请直接输入数字进行选择,或者进行自由回复(未实现)\n", "阿p:1\n", "\n", "糖糖:哇!打好了!合适吗?合适吗?快他妈夸我合适!!!\n", "\n", "-------------\n", "\n", "1. 随机一个事件\n", "2. 自由聊天\n", "或者输入Quit退出\n", "请选择一个选项: 1\n", "\n", "-------------\n", "\n", "糖糖: 我问你哦,我真的可以就这样活下去吗?\n", "\n", "--请选择你的回复--\n", "1. 阿p:怎么了啊?\n", "2. 阿p:真的可以呀\n", "3. 阿p:对没错\n", "4. 阿p:那还用说\n", "5. 阿p:其实谁都行\n", "6. 阿p:脸\n", "7. 阿p:一切\n", "8. 阿p:没什么不行吧?\n", "9. 阿p:不可以\n", "10. 阿p:喜欢啊\n", "11. 阿p:喜欢吧\n", "12. 阿p:真的超超喜欢\n", "13. 阿p:超超喜欢\n", "14. 阿p:以当代互联网小天使的身份活下去\n", "15. 阿p:真的超超喜欢\n", "\n", "请直接输入数字进行选择,或者进行自由回复(未实现)\n", "阿p:1\n", "\n", "\n", "\n", "-------------\n", "\n", "1. 随机一个事件\n", "2. 自由聊天\n", "或者输入Quit退出\n", "请选择一个选项: 1\n", "\n", "-------------\n", "\n", "糖糖: 糖糖,是不是还是去死一死比较好……\n", "\n", "--请选择你的回复--\n", "1. 阿p:要活下去啊!!!\n", "2. 阿p:死~寂\n", "3. 阿p:你有颜值啊\n", "4. 阿p:不如砍掉重练吧!\n", "5. 阿p:不是还有宅宅们嘛\n", "\n", "请直接输入数字进行选择,或者进行自由回复(未实现)\n", "阿p:1\n", "\n", "糖糖:可是,糖糖又没有活着的价值……\n", "\n", "-------------\n", "\n", "1. 随机一个事件\n", "2. 自由聊天\n", "或者输入Quit退出\n", "请选择一个选项: 1\n", "\n", "-------------\n", "\n", "糖糖: 机会这么难得,要不整点富婆快乐活吧,说不定还能用作下次的企划哦!\n", "\n", "--请选择你的回复--\n", "1. 阿p:买头老虎在大街上放生\n", "2. 阿p:无所谓,不管你是不是富婆我都爱你\n", "3. 阿p:要不把整个筑地买下来吧\n", "\n", "请直接输入数字进行选择,或者进行自由回复(未实现)\n", "阿p:1\n", "\n", "糖糖:好像买一头就要几百万哦……\n", "\n", "-------------\n", "\n", "1. 随机一个事件\n", "2. 自由聊天\n", "或者输入Quit退出\n", "请选择一个选项: 1\n", "\n", "-------------\n", "\n", "糖糖: 我要出去玩!给我零花钱!!!\n", "\n", "--请选择你的回复--\n", "1. 阿p:给10圆\n", "2. 阿p:给3000圆\n", "3. 阿p:给10000圆\n", "\n", "请直接输入数字进行选择,或者进行自由回复(未实现)\n", "阿p:1\n", "\n", "糖糖:这点钱连小学生都打发不了好吧!!!真是的,看我今天赖在家黏你一整天!!!!\n", "\n", "-------------\n", "\n", "1. 随机一个事件\n", "2. 自由聊天\n", "或者输入Quit退出\n", "请选择一个选项: 1\n", "\n", "-------------\n", "\n", "糖糖: 小天使请安!这个开场白也说厌了啊~,帮我想个别的开场白!\n", "\n", "--请选择你的回复--\n", "1. 阿p:当代互联网小天使,参上!\n", "2. 阿p:我是路过的网络主播,给我记住了!\n", "3. 阿p:那么,我们开始直播吧\n", "\n", "请直接输入数字进行选择,或者进行自由回复(未实现)\n", "阿p:1\n", "\n", "糖糖:试着上超天酱的钩吧?之类的嘿嘿\n", "\n", "-------------\n", "\n", "1. 随机一个事件\n", "2. 自由聊天\n", "或者输入Quit退出\n", "请选择一个选项: 1\n", "\n", "-------------\n", "\n", "糖糖: 我们点外卖吧我一步也不想动了可是又超想吃饭!!!\n", "\n", "--请选择你的回复--\n", "1. 阿p:烦死了白痴\n", "2. 阿p:吃土去吧你\n", "3. 阿p:那我点了哦\n", "\n", "请直接输入数字进行选择,或者进行自由回复(未实现)\n", "阿p:1\n", "\n", "糖糖:555555555 但是我们得省钱对吧\n", "\n", "-------------\n", "\n", "1. 随机一个事件\n", "2. 自由聊天\n", "或者输入Quit退出\n", "请选择一个选项: 1\n", "\n", "-------------\n", "\n", "糖糖: 哎,你会希望看到糖糖将来的样子吗?\n", "\n", "--请选择你的回复--\n", "1. 阿p:机器人\n", "2. 阿p:合成怪物\n", "3. 阿p:狂战士\n", "\n", "请直接输入数字进行选择,或者进行自由回复(未实现)\n", "阿p:1\n", "\n", "糖糖:——“糖糖”OS,启动\n", "\n", "-------------\n", "\n", "1. 随机一个事件\n", "2. 自由聊天\n", "或者输入Quit退出\n", "请选择一个选项: 1\n", "\n", "-------------\n", "\n", "糖糖: 我没打招呼就把冰箱里的布丁吃了 会被判死刑吗???\n", "\n", "--请选择你的回复--\n", "1. 阿p:原谅你\n", "2. 阿p:糖糖可以随便吃哦\n", "\n", "请直接输入数字进行选择,或者进行自由回复(未实现)\n", "阿p:1\n", "\n", "糖糖:嗯 能被糖糖吃掉也是布丁的荣幸 所以当然没问题\n", "\n", "-------------\n", "\n", "1. 随机一个事件\n", "2. 自由聊天\n", "或者输入Quit退出\n", "请选择一个选项: 1\n", "\n", "-------------\n", "\n", "糖糖: 今天有点想试试平时不会做的事\n", "\n", "--请选择你的回复--\n", "1. 阿p:杀人\n", "2. 阿p:相爱\n", "3. 阿p:抢银行\n", "\n", "请直接输入数字进行选择,或者进行自由回复(未实现)\n", "阿p:1\n", "\n", "糖糖:如果我搞砸了……就由阿P杀了我吧\n", "\n", "-------------\n", "\n", "1. 随机一个事件\n", "2. 自由聊天\n", "或者输入Quit退出\n", "请选择一个选项: 1\n", "\n", "-------------\n", "\n", "糖糖: 哎,你喜欢什么样的糖糖啊?\n", "\n", "--请选择你的回复--\n", "1. 阿p:无情人设\n", "2. 阿p:天才博士人设\n", "3. 阿p:得寸进尺小萝莉\n", "\n", "请直接输入数字进行选择,或者进行自由回复(未实现)\n", "阿p:1\n", "\n", "糖糖:……我不明白,“感情”是什么\n", "\n", "-------------\n", "\n", "1. 随机一个事件\n", "2. 自由聊天\n", "或者输入Quit退出\n", "请选择一个选项: 1\n", "warning! all candidate event was sampled\n", "\n", "-------------\n", "\n", "糖糖: 我也想被做进那个大乱斗游戏……,哎,如果那个游戏里面有超天酱的话,阿P会用我吗?\n", "\n", "--请选择你的回复--\n", "1. 阿p:嗯啊\n", "2. 阿p:不打算用\n", "\n", "请直接输入数字进行选择,或者进行自由回复(未实现)\n", "阿p:1\n", "\n", "糖糖:真的咩?!那我立刻开始练习捡信\n", "\n", "-------------\n", "\n", "1. 随机一个事件\n", "2. 自由聊天\n", "或者输入Quit退出\n", "请选择一个选项: 1\n", "warning! all candidate event was sampled\n", "\n", "-------------\n", "\n", "糖糖: 我没打招呼就把冰箱里的布丁吃了 会被判死刑吗???\n", "\n", "--请选择你的回复--\n", "1. 阿p:原谅你\n", "2. 阿p:糖糖可以随便吃哦\n", "\n", "请直接输入数字进行选择,或者进行自由回复(未实现)\n", "阿p:1\n", "\n", "糖糖:嗯 能被糖糖吃掉也是布丁的荣幸 所以当然没问题\n", "\n", "-------------\n", "\n", "1. 随机一个事件\n", "2. 自由聊天\n", "或者输入Quit退出\n", "请选择一个选项: Quit\n" ] } ] }, { "cell_type": "code", "source": [ "game_master = GameMaster()\n", "game_master.run()" ], "metadata": { "colab": { "base_uri": "https://localhost:8080/" }, "id": "5GwFCR_wLtay", "outputId": "9dc0c692-9dd4-4310-cd1a-3fdb89fa76b8" }, "execution_count": null, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "1. 随机一个事件\n", "2. 自由聊天\n", "或者输入Quit退出\n", "请选择一个选项: 1\n", "\n", "-------------\n", "\n", "糖糖: 机会这么难得,要不整点富婆快乐活吧,说不定还能用作下次的企划哦!\n", "\n", "--请选择你的回复--\n", "1. 阿p:买头老虎在大街上放生\n", "2. 阿p:无所谓,不管你是不是富婆我都爱你\n", "3. 阿p:要不把整个筑地买下来吧\n", "\n", "请直接输入数字进行选择,或者进行自由回复(未实现)\n", "阿p:我觉得可以把钱拿来进一步投资哦\n", "Memory: ['💰😓', '🤔😳', '🤔🎮', '💸😡', '😔😌', '😔😔', '😔😍']\n", "糖糖:「阿哈,投资?那我是不是可以买更多的二次元周边啦?!」\n", "自由回复的算分功能还未实现\n", "\n", "-------------\n", "\n", "('糖糖:「 机会这么难得,要不整点富婆快乐活吧,说不定还能用作下次的企划哦!」\\n阿P:「买头老虎在大街上放生」\\n糖糖:「好像买一头就要几百万哦……」\\n', '💰😓')\n", "按任意键继续...Quit\n", "1. 随机一个事件\n", "2. 自由聊天\n", "或者输入Quit退出\n", "请选择一个选项: Quit\n" ] } ] }, { "cell_type": "code", "source": [ "\n", "game_master = GameMaster()\n", "game_master.run()" ], "metadata": { "colab": { "base_uri": "https://localhost:8080/" }, "id": "zPmr9kVepwjh", "outputId": "3a8bcbc6-06ef-4542-ef70-03cd8ed0b357" }, "execution_count": null, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "1. 随机一个事件\n", "2. 自由聊天\n", "或者输入Quit退出\n", "请选择一个选项: 2\n", "聊天:你好呀糖糖\n", "Memory: ['😔😔', '🍔😢', '💸😡', '🤔😔', '🍬😔', '💪😔', '🤔😊']\n", "糖糖:「哈喽~阿哈!终于又见面了呢,我都快等不及了呢!」\n", "聊天:等不及要心心了吗\n", "Memory: ['😔😌', '🍔😢', '🤔😳', '💔😢', '😳😅', '💰😓', '😔😔']\n", "糖糖:「诶~你怎么这么了解我呀!心心已经开始了,我都快被你迷得神魂颠倒了!」\n", "聊天:Quit\n", "1. 随机一个事件\n", "2. 自由聊天\n", "或者输入Quit退出\n", "请选择一个选项: quit\n" ] } ] }, { "cell_type": "markdown", "source": [ "\n", "---\n", "\n", "这个以下都是非主要代码和单元测试\n", "\n", "---\n", "\n", "这个以下都是非主要代码和单元测试\n", "\n", "\n", "---\n", "\n", "这个以下都是非主要代码和单元测试\n", "\n", "\n", "---\n", "\n", "这个以下都是非主要代码和单元测试\n", "\n" ], "metadata": { "id": "WHxC8m7oH3W4" } }, { "cell_type": "markdown", "source": [ "# 不同状态下的Agent测试" ], "metadata": { "id": "m5J7wuRoIqTd" } }, { "cell_type": "code", "source": [ "chat_master = ChatMaster(memory_pool)\n", "agent = Agent()\n", "agent[\"Stress\"] = 0\n", "agent[\"Affection\"] = 0\n", "agent[\"Darkness\"] = 0\n", "\n", "chat_master.run(agent)" ], "metadata": { "colab": { "base_uri": "https://localhost:8080/" }, "id": "QBY81TRMIrID", "outputId": "0c18759e-24b5-48ff-8a59-dedb88c85a79" }, "execution_count": null, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "阿p:你今天心情怎么样?\n", "Memory: ['', '', '😔', '', '🍬😔', '', '']\n", "啊~今天的心情还好啦~有点嗨,有点闷,有点复杂的感觉~不过没关系,糖糖还是会努力开心起来的~你今天遇到什么有趣的事情了吗?快来分享一下嘛!\n", "阿p:Quit\n" ] } ] }, { "cell_type": "code", "source": [ "chat_master = ChatMaster(memory_pool)\n", "agent = Agent()\n", "agent[\"Stress\"] = 100\n", "agent[\"Affection\"] = 0\n", "agent[\"Darkness\"] = 0\n", "\n", "chat_master.run(agent)" ], "metadata": { "colab": { "base_uri": "https://localhost:8080/" }, "id": "VoXh56exJIrL", "outputId": "544cdd1c-b274-471d-890b-3e3a9377593d" }, "execution_count": null, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "阿p:你今天心情怎么样?\n", "Memory: ['', '', '', '', '', '', '']\n", "啊~今天心情真的是超级烂,简直就是要爆炸了QAQ,一点都不开心呢。你有没有什么好玩的事情可以分享一下?\n", "阿p:Quit\n" ] } ] }, { "cell_type": "code", "source": [ "chat_master = ChatMaster(memory_pool)\n", "agent = Agent()\n", "agent[\"Stress\"] = 0\n", "agent[\"Affection\"] = 80\n", "agent[\"Darkness\"] = 0\n", "\n", "chat_master.run(agent)" ], "metadata": { "colab": { "base_uri": "https://localhost:8080/" }, "id": "EPISkUJVJXzm", "outputId": "2f4d1181-7ded-4d5b-f58b-a67e1715d6af" }, "execution_count": null, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "阿p:糖糖,快表演机器人\n", "Memory: ['🤔😔', '🍬😔', '', '', '', '', '🎉😊']\n", "啊哈~阿P你真是个调皮鬼,总是喜欢逗我玩,真是让我笑死了!好吧,我就给你表演个机器人吧!看好了啊~「机器人模式启动」(机械声效)「Beep beep boop」(模仿机器人声音)「我是糖糖机器人,全面服务中,请问阿P有什么指令?」嘿嘿~怎么样,我是不是个超级可爱的机器人呢?QWQ\n", "阿p:Quit\n" ] } ] }, { "cell_type": "code", "source": [ "chat_master = ChatMaster(memory_pool)\n", "agent = Agent()\n", "agent[\"Stress\"] = 0\n", "agent[\"Affection\"] = 0\n", "agent[\"Darkness\"] = 0\n", "\n", "chat_master.run(agent)" ], "metadata": { "colab": { "base_uri": "https://localhost:8080/" }, "id": "eCJdzQSkJdy7", "outputId": "6d8264b2-b6f6-4217-ce4a-9aec0a940636" }, "execution_count": null, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "阿p:糖糖,快表演机器人\n", "Memory: ['🤔😔', '🍬😔', '', '', '🎉😊', '', '']\n", "啊哈~阿P你真是个大坏蛋,总是逗我开心,真是让我笑死了!好吧,我就给你表演个机器人吧!看好了啊~「机器人模式启动」(模仿机械声音)「Beep beep boop」(模仿机器人声音)「我是糖糖机器人,全面服务中,请问阿P有什么指令?」嘿嘿~怎么样,我是不是个超级可爱的机器人呢?阿哈~快夸我一下吧!QWQ\n", "阿p:Quit\n" ] } ] }, { "cell_type": "markdown", "source": [ "# Memory\n", "\n", "memory我们希望Event和Memory是分离的Event的标准字段如下\n", "\n", "- Name, Event的Name,用来后续如果玩家进行游戏修改的话可以根据\n", "- Text, 这个event下完整的对话文本\n", "- Embedding, text的embedding\n", "- Condition, 这个event对应的出现条件\n", "- Emoji, 这个memory的缩写显示emoji\n", "\n", "Memory应该可以从Event去默认load一个" ], "metadata": { "id": "NQuYYbb33-Cc" } }, { "cell_type": "code", "source": [ "example_memory_json = {\n", " \"Name\": \"EventName\",\n", " \"Text\": \"Sample Text\",\n", " \"Embedding\": [0,0,0],\n", " \"Condition\": \"\",\n", " \"Emoji\": \"😓🤯\"\n", "}" ], "metadata": { "id": "JaKoW7oK391c" }, "execution_count": null, "outputs": [] }, { "cell_type": "markdown", "source": [ "Memory会包含下面几个字段\n", "\n", "example_memory_json = {\n", " \"Name\": \"EventName\",\n", " \"Text\": \"Sample Text\",\n", " \"Embedding\": [0,0,0],\n", " \"Condition\": \"\",\n", " \"Emoji\": \"😓🤯\"\n", "}\n", "\n", "请为我创建一个Memory类\n", "\n", "这个memory类可以通过Memory(json_str)来载入\n", "\n", "同时这个类也有和DIalogueEvent类似的get和setitem的功能" ], "metadata": { "id": "qUcHULFR4GQR" } }, { "cell_type": "code", "source": [ "# Memory 类不再使用\n", "\n", "# import json\n", "\n", "# class Memory:\n", "# def __init__(self, json_str=None):\n", "# if json_str:\n", "# try:\n", "# self.data = json.loads(json_str)\n", "# except json.JSONDecodeError:\n", "# print(\"输入的字符串不是有效的JSON格式。\")\n", "# self.data = {}\n", "# else:\n", "# self.data = {}\n", "\n", "# def load_from_event( event ):\n", "# pass\n", "\n", "# def __getitem__(self, key):\n", "# return self.data.get(key, None)\n", "\n", "# def __setitem__(self, key, value):\n", "# self.data[key] = value\n", "\n", "# def __repr__(self):\n", "# return str(self.data)\n", "\n", "\n", "# example_memory_json = {\n", "# \"Name\": \"EventName\",\n", "# \"Text\": \"Sample Text\",\n", "# \"Embedding\": [0, 0, 0],\n", "# \"Condition\": \"\",\n", "# \"Emoji\": \"😓🤯\"\n", "# }\n", "\n", "# # 通过给定的json字符串初始化Memory实例\n", "# memory = Memory(json.dumps(example_memory_json))\n", "\n", "# # 通过类似字典的方式访问数据\n", "# print(memory[\"Name\"]) # 打印Name字段的内容\n", "# print(memory[\"Emoji\"]) # 打印Emoji字段的内容\n" ], "metadata": { "id": "Jnjyi62a4Bbt" }, "execution_count": null, "outputs": [] }, { "cell_type": "markdown", "source": [ "## parse_attribute_string单元测试" ], "metadata": { "id": "mVgTS5dlFn6P" } }, { "cell_type": "code", "source": [ "from util import parse_attribute_string\n", "\n", "# Test cases\n", "print(parse_attribute_string(\"Stress: -1.0, Affection: +0.5\")) # Output: {'Stress': -1.0, 'Affection': 0.5}\n", "print(parse_attribute_string(\"Affection: +4.0, Stress: -2.0, Darkness: -1.0\")) # Output: {'Affection': 4.0, 'Stress': -2.0, 'Darkness': -1.0}\n", "print(parse_attribute_string(\"Affection: +2.0, Stress: -1.0, Darkness: ?\")) # Output: {'Affection': 2.0, 'Stress': -1.0, 'Darkness': 0}\n", "print(parse_attribute_string(\"Stress: -1.0\")) # Output: {'Stress': -1.0}\n" ], "metadata": { "id": "HGaXw1osFo7U" }, "execution_count": null, "outputs": [] }, { "cell_type": "markdown", "source": [ "## Embedding 单元测试" ], "metadata": { "id": "6MEN4KahF-Ab" } }, { "cell_type": "code", "source": [ "!pip install -q transformers\n", "\n", "from util import get_bge_embedding_zh\n", "\n", "result = get_bge_embedding_zh(\"你好\")\n", "print( result )" ], "metadata": { "id": "86lKC20uF_8_" }, "execution_count": null, "outputs": [] }, { "cell_type": "markdown", "source": [ "## parsing_condition_string 单元测试" ], "metadata": { "id": "WM1c9xMXGJHT" } }, { "cell_type": "code", "source": [ "from util import parsing_condition_string\n", "\n", "# 测试例子\n", "example_inputs = [\n", " \"Random Noon Event: Darkness 0-39\",\n", " \"Random Noon Event: Stress 0-19\",\n", " \"Random Noon Event: Affection 61+\",\n", " \"Random Noon Event: No Attribute\"\n", "]\n", "\n", "for example_input in example_inputs:\n", " print(f\"example_input:\\n{example_input}\\nexample_output\\n{parsing_condition_string(example_input)}\\n\")\n" ], "metadata": { "id": "93GwecaBGIys" }, "execution_count": null, "outputs": [] }, { "cell_type": "markdown", "source": [ "我已经实现了一个类\n", "\n", "class ChatHaruhi:\n", "\n", "\n", "这个类有两个关键方法\n", "\n", "```python\n", "\n", " def add_story(self, query):\n", "\n", " if self.db is None:\n", " return\n", " \n", " query_vec = self.embedding(query)\n", "\n", " stories = self.db.search(query_vec, self.k_search)\n", " \n", " story_string = self.story_prefix_prompt\n", " sum_story_token = self.tokenizer(story_string)\n", " \n", " for story in stories:\n", " story_token = self.tokenizer(story) + self.tokenizer(self.dialogue_divide_token)\n", " if sum_story_token + story_token > self.max_len_story:\n", " break\n", " else:\n", " sum_story_token += story_token\n", " story_string += story + self.dialogue_divide_token\n", "\n", " self.llm.user_message(story_string)\n", "\n", " def chat(self, text, role):\n", " # add system prompt\n", " self.llm.initialize_message()\n", " self.llm.system_message(self.system_prompt)\n", " \n", "\n", " # add story\n", " query = self.get_query_string(text, role)\n", " self.add_story( query )\n", "\n", " # add history\n", " self.add_history()\n", "\n", " # add query\n", " self.llm.user_message(query)\n", " \n", " # get response\n", " response_raw = self.llm.get_response()\n", "\n", " response = response_postprocess(response_raw, self.dialogue_bra_token, self.dialogue_ket_token)\n", "\n", " # record dialogue history\n", " self.dialogue_history.append((query, response))\n", "\n", "\n", "\n", " return response\n", "```\n", "\n", "我希望在一个新的应用中复用这个类,\n", "\n", "但是在新的应用中,我定义了新的方法来获取add_story中的stories\n", "\n", "即\n", "\n", "stories = new_get_stories( query )\n", "\n", "我现在想复用这个类,仅改变add_stories方法,我有什么好的办法来实现?" ], "metadata": { "id": "LAYDsOmKKPNv" } }, { "cell_type": "markdown", "source": [ "```python\n", "class EnhancedChatHaruhi(ChatHaruhi):\n", "\n", " def new_get_stories(self, query):\n", " # 这里实现您新的获取故事的方法\n", " # 返回故事列表\n", " pass\n", "\n", " def add_story(self, query):\n", " if self.db is None:\n", " return\n", " \n", " # 调用新的获取故事的方法\n", " stories = self.new_get_stories(query)\n", " \n", " story_string = self.story_prefix_prompt\n", " sum_story_token = self.tokenizer(story_string)\n", " \n", " for story in stories:\n", " story_token = self.tokenizer(story) + self.tokenizer(self.dialogue_divide_token)\n", " if sum_story_token + story_token > self.max_len_story:\n", " break\n", " else:\n", " sum_story_token += story_token\n", " story_string += story + self.dialogue_divide_token\n", "\n", " self.llm.user_message(story_string)\n", "```" ], "metadata": { "id": "QRvwYYQH1xD4" } }, { "cell_type": "markdown", "source": [ "我希望实现一个python函数\n", "\n", "分析一个字符串中有没有\":\"\n", "\n", "如果有,我希望在第一个\":\"的位置分开成str_left和str_right,并以f\"{str_left}:「{str_right}」\"的形式输出\n", "\n", "例子输入\n", "爸爸:我真棒\n", "例子输出\n", "爸爸:「我真棒」\n", "例子输入\n", "这一句没有冒号\n", "例子输出\n", ":「这一句没有冒号」\n" ], "metadata": { "id": "kiDXmwI21znH" } }, { "cell_type": "code", "source": [ "def wrap_text_with_colon(text):\n", " # 查找冒号在字符串中的位置\n", " colon_index = text.find(\":\")\n", "\n", " # 如果找到了冒号\n", " if colon_index != -1:\n", " # 分割字符串为左右两部分\n", " str_left = text[:colon_index]\n", " str_right = text[colon_index+1:]\n", " # 构造新的格式化字符串\n", " result = f\"{str_left}:「{str_right}」\"\n", " else:\n", " # 如果没有找到冒号,整个字符串被认为是右侧部分\n", " result = f\":「{text}」\"\n", "\n", " return result\n", "\n", "# 示例输入\n", "print(wrap_text_with_colon(\"爸爸:我真棒\")) # 爸爸:「我真棒」\n", "print(wrap_text_with_colon(\"这一句没有冒号\")) # :「这一句没有冒号」\n" ], "metadata": { "colab": { "base_uri": "https://localhost:8080/" }, "id": "ZUWO0yqNMuoW", "outputId": "4c815ef4-5f5d-43ec-856d-8afe7d1741b8" }, "execution_count": null, "outputs": [ { "output_type": "stream", "name": "stdout", "text": [ "爸爸:「我真棒」\n", ":「这一句没有冒号」\n" ] } ] }, { "cell_type": "markdown", "source": [ "## MemoryPool的单元测试" ], "metadata": { "id": "5v3VfnluEp3_" } }, { "cell_type": "code", "source": [ "retrieved_memories = memory_pool.retrieve( agent , \"你是一个什么样的主播啊\" )\n", "\n", "for mem in retrieved_memories[:2]:\n", " print(mem[\"text\"])\n", " print(mem[\"emoji\"])\n", " print(\"---\")" ], "metadata": { "colab": { "base_uri": "https://localhost:8080/" }, "id": "gbkumgmX2VPF", "outputId": "76cad38f-47d4-4189-dc0f-347446d64703" }, "execution_count": null, "outputs": [ { "output_type": "stream", "name": "stdout", "text": [ "糖糖:「 我也想被做进那个大乱斗游戏……,哎,如果那个游戏里面有超天酱的话,阿P会用我吗?」\n", "阿P:「嗯啊」\n", "糖糖:「真的咩?!那我立刻开始练习捡信」\n", "\n", "😔😍\n", "---\n", "糖糖:「 我今后也会努力加油的,你要支持我哦 还有阿P你自己也要加油哦!」\n", "阿P:「哇 说的话跟偶像一样 好恶心哦」\n", "糖糖:「是哦 我怎么会说这样的话呢 我又没有很想努力……」\n", "\n", "💪😔\n", "---\n" ] } ] }, { "cell_type": "markdown", "source": [ "## Agent的单元测试" ], "metadata": { "id": "a45r14X8E9XR" } }, { "cell_type": "code", "source": [ "from Agent import Agent\n", "\n", "agent = Agent()\n", "\n", "if __name__ == \"__main__\":\n", " # 示例用法\n", "\n", " print(agent[\"Stress\"]) # 输出 0\n", " agent[\"Stress\"] += 1\n", " print(agent[\"Stress\"]) # 输出 1\n", " agent.apply_attribute_change({\"Darkness\": -1, \"Stress\": 1})\n", " print(agent[\"Darkness\"]) # 输出 -1\n", " print(agent[\"Stress\"]) # 输出 2\n", " agent.apply_attribute_change({\"Nonexistent\": 5}) # 输出 Warning: Nonexistent not in attributes, skipping\n", "\n", " condition = ('Stress', 0, 19)\n", "\n", " print( agent.in_condition( condition ) )" ], "metadata": { "id": "VyPhQxNZEsHC" }, "execution_count": null, "outputs": [] }, { "cell_type": "markdown", "source": [ "## DialogueEvent的单元测试" ], "metadata": { "id": "lcIJuHfiGDI3" } }, { "cell_type": "code", "source": [ "from DialogueEvent import DialogueEvent\n", "\n", "\n", "example_json_str = \"\"\"{\"prefix\": \"糖糖: 嘿嘿,最近我在想要不要改变直播风格,你觉得我应该怎么做呀?\", \"options\": [{\"user\": \"你可以试试唱歌直播呀!\", \"reply\": \"糖糖: 哇!唱歌直播是个好主意!我可以把我的可爱音色展现给大家听听!谢谢你的建议!\", \"attribute_change\": \"Stress: -1.0\"}, {\"user\": \"你可以尝试做一些搞笑的小品,逗大家开心。\", \"reply\": \"糖糖: 哈哈哈,小品确实挺有趣的!我可以挑战一些搞笑角色,给大家带来欢乐!谢谢你的建议!\", \"attribute_change\": \"Stress: -1.0\"}, {\"user\": \"你可以尝试做游戏直播,和观众一起玩游戏。\", \"reply\": \"糖糖: 游戏直播也不错!我可以和观众一起玩游戏,互动更加有趣!谢谢你的建议!\", \"attribute_change\": \"Stress: -1.0\"}]}\"\"\"\n", "\n", "# 通过给定的json字符串初始化DialogueEvent实例\n", "event = DialogueEvent(example_json_str)\n", "\n", "# 通过类似字典的方式访问数据\n", "# print(event[\"options\"]) # 打印options字段的内容\n", "\n", "print(event.transfer_output(1) )\n", "\n", "print(event.get_most_neutral())\n", "\n", "print(event.most_neutral_output())\n", "\n" ], "metadata": { "colab": { "base_uri": "https://localhost:8080/" }, "id": "0Tp8qSXNGFNn", "outputId": "2ec91dde-7d26-450d-a283-084bd7456631" }, "execution_count": null, "outputs": [ { "output_type": "stream", "name": "stdout", "text": [ "糖糖:「 嘿嘿,最近我在想要不要改变直播风格,你觉得我应该怎么做呀?」\n", "阿P:「你可以尝试做一些搞笑的小品,逗大家开心。」\n", "糖糖:「 哈哈哈,小品确实挺有趣的!我可以挑战一些搞笑角色,给大家带来欢乐!谢谢你的建议!」\n", "\n", "0\n", "('糖糖:「 嘿嘿,最近我在想要不要改变直播风格,你觉得我应该怎么做呀?」\\n阿P:「你可以试试唱歌直播呀!」\\n糖糖:「 哇!唱歌直播是个好主意!我可以把我的可爱音色展现给大家听听!谢谢你的建议!」\\n', '📄📄')\n" ] } ] }, { "cell_type": "markdown", "source": [ "## NeedyHaruhi的单元测试" ], "metadata": { "id": "wNiah9RrGhCQ" } }, { "cell_type": "code", "source": [ "needy_chatbot = NeedyHaruhi( system_prompt = system_prompt ,\n", " story_text_folder = None )\n", "\n", "query_text = \"糖糖,你今天怎么样啊?\"\n", "query_text_for_embedding = \"阿p:「\" + query_text + \"」\"\n", "retrieved_memories = memory_pool.retrieve( agent , query_text )\n", "\n", "memory_text = [mem[\"text\"] for mem in retrieved_memories]\n", "memory_emoji = [mem[\"emoji\"] for mem in retrieved_memories]\n", "\n", "needy_chatbot.set_stories( memory_text )\n", "\n", "print(\"Mem:\", memory_emoji )\n", "\n", "response = needy_chatbot.chat( role = \"阿p\", text = query_text )\n", "print(response)" ], "metadata": { "id": "XwcbSxlYGFY3" }, "execution_count": null, "outputs": [] }, { "cell_type": "markdown", "source": [ "## 载入ChatHaruhi的测试" ], "metadata": { "id": "BdARAEura7yJ" } }, { "cell_type": "code", "source": [ "from chatharuhi import ChatHaruhi\n", "\n", "chatbot = ChatHaruhi( role_from_hf = 'chengli-thu/Jack-Sparrow', \\\n", " llm = 'openai',\n", " embedding = 'bge_en'\n", " )" ], "metadata": { "id": "ISd8bD4Ya85A" }, "execution_count": null, "outputs": [] }, { "cell_type": "markdown", "source": [ "显示图片" ], "metadata": { "id": "sR9u0ArQQmvo" } }, { "cell_type": "code", "source": [ "import matplotlib.pyplot as plt\n", "import matplotlib.image as mpimg\n", "\n", "image_path = '/content/image'\n", "\n", "for data in data_img_text:\n", " img_name = data['img_name']\n", "\n", " # 拼接完整的图片路径\n", " img_path = os.path.join(image_path, img_name)\n", "\n", " # 读取图片\n", " img = mpimg.imread(img_path)\n", "\n", " # 可视化图片\n", " plt.imshow(img)\n", " plt.axis('off')\n", " plt.show()\n", "\n", " break" ], "metadata": { "id": "6T9LfbweQnh5" }, "execution_count": null, "outputs": [] } ] }