Spaces:
Running
Running
File size: 6,401 Bytes
1966e2b |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 |
{
"cells": [
{
"cell_type": "markdown",
"id": "ebeba428",
"metadata": {},
"source": [
"# ✅ RAG JuJutsu PoC (Notebook with Joblib, FAISS, ChatGPT API)"
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "8bdfd3c8",
"metadata": {
"scrolled": true
},
"outputs": [],
"source": [
"\n",
"!pip install --quiet openai langchain faiss-cpu PyPDF2 sentence-transformers joblib\n",
"!pip install ipywidgets==7.7.2\n",
"!jupyter nbextension enable --py widgetsnbextension\n",
"!jupyter notebook\n"
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "49ee7721",
"metadata": {},
"outputs": [],
"source": [
"from PyPDF2 import PdfReader\n",
"from langchain.text_splitter import RecursiveCharacterTextSplitter\n",
"\n",
"def load_pdf_chunks(pdf_path):\n",
" reader = PdfReader(pdf_path)\n",
" raw_text = \"\"\n",
" for page in reader.pages:\n",
" raw_text += page.extract_text() + \"\\n\"\n",
"\n",
" splitter = RecursiveCharacterTextSplitter(chunk_size=500, chunk_overlap=50)\n",
" return splitter.split_text(raw_text)\n",
"\n",
"chunks = load_pdf_chunks(\"JuJutsu-Contexto-Significado-Conexiones-Historia.pdf\")\n",
"print(f\"Loaded {len(chunks)} chunks\")\n"
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "8109b626-0179-43e2-b924-65afe9af1e4e",
"metadata": {},
"outputs": [],
"source": [
"import openai"
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "371c637e",
"metadata": {},
"outputs": [],
"source": [
"\n",
"import faiss\n",
"import numpy as np\n",
"import joblib\n",
"\n",
"def get_openai_embeddings(texts):\n",
" embeddings = []\n",
" for text in texts:\n",
" response = openai.Embedding.create(\n",
" model=\"text-embedding-3-small\",\n",
" input=text\n",
" )\n",
" vector = response['data'][0]['embedding']\n",
" embeddings.append(vector)\n",
" return np.array(embeddings)\n",
"\n",
"embeddings = get_openai_embeddings(chunks)\n",
"index = faiss.IndexFlatL2(embeddings.shape[1])\n",
"index.add(np.array(embeddings))\n",
"\n",
"joblib.dump((chunks, index), \"rag_model.joblib\")\n",
"print(\"Chunks and index serialized to rag_model.joblib\")\n"
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "28ce4963",
"metadata": {},
"outputs": [],
"source": [
"\n",
"import joblib\n",
"chunks, index = joblib.load(\"rag_model.joblib\")\n",
"print(\"Chunks and index loaded from rag_model.joblib\")\n"
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "51a89e77",
"metadata": {},
"outputs": [],
"source": [
"\n",
"def search(query, k=3):\n",
" response = openai.Embedding.create(\n",
" model=\"text-embedding-3-small\",\n",
" input=query\n",
" )\n",
" query_vec = np.array([response['data'][0]['embedding']])\n",
" scores, indices = index.search(query_vec, k)\n",
" return [chunks[i] for i in indices[0]]\n"
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "34315775",
"metadata": {},
"outputs": [],
"source": [
"\n",
"import os\n",
"import openai\n",
"from openai import OpenAI\n",
"\n",
"os.environ[\"OPENAI_API_KEY\"] = os.getenv(\"OPENAI_API_KEY\")\n",
"client = OpenAI()\n",
"\n",
"def chat_no_rag(question):\n",
" response = client.chat.completions.create(\n",
" model=\"gpt-3.5-turbo\",\n",
" messages=[\n",
" {\"role\": \"user\", \"content\": question}\n",
" ],\n",
" temperature=0.5,\n",
" max_tokens=200,\n",
" )\n",
" return response.choices[0].message.content\n",
"\n",
"def chat_with_rag(question, retrieved_chunks):\n",
" context = \"\\n\".join(retrieved_chunks)\n",
" prompt = f\"Usa el siguiente contexto para responder la pregunta:\\n\\n{context}\\n\\nPregunta: {question}\"\n",
"\n",
" response = client.chat.completions.create(\n",
" model=\"gpt-3.5-turbo\",\n",
" messages=[\n",
" {\"role\": \"user\", \"content\": prompt}\n",
" ],\n",
" temperature=0.3,\n",
" max_tokens=200,\n",
" )\n",
" return response.choices[0].message.content\n",
"\n",
"def chat_with_rag_enhanced(question, retrieved_chunks):\n",
" context = \"\\n\".join(retrieved_chunks)\n",
" prompt = (\n",
" \"Eres un experto en historia marcial. \"\n",
" \"Usa el siguiente contexto histórico para responder con precisión y detalle.\\n\\n\"\n",
" f\"Contexto:\\n{context}\\n\\n\"\n",
" f\"Pregunta: {question}\\nRespuesta:\"\n",
" )\n",
"\n",
" response = client.chat.completions.create(\n",
" model=\"gpt-3.5-turbo\",\n",
" messages=[\n",
" {\"role\": \"user\", \"content\": prompt}\n",
" ],\n",
" temperature=0.2,\n",
" max_tokens=200,\n",
" )\n",
" return response.choices[0].message.content\n"
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "900dfdfa",
"metadata": {},
"outputs": [],
"source": [
"\n",
"# Example query\n",
"query = \"¿Cuál es el origen del JuJutsu en Japón?\"\n",
"retrieved = search(query)\n",
"\n",
"print(\"🔹 Sin RAG:\")\n",
"print(chat_no_rag(query))\n",
"\n",
"print(\"\\n🔹 Con RAG:\")\n",
"print(chat_with_rag(query, retrieved))\n",
"\n",
"print(\"\\n🔹 Con RAG + Prompt mejorado:\")\n",
"print(chat_with_rag_enhanced(query, retrieved))\n"
]
}
],
"metadata": {
"kernelspec": {
"display_name": "Python 3 (ipykernel)",
"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": "3.11.7"
}
},
"nbformat": 4,
"nbformat_minor": 5
}
|