RAG example

examples/RAG.ipynb

@@ -0,0 +1,211 @@
+{
+ "cells": [
+  {
+   "cell_type": "code",
+   "execution_count": 57,
+   "metadata": {},
+   "outputs": [
+    {
+     "data": {
+      "text/plain": [
+       "True"
+      ]
+     },
+     "execution_count": 57,
+     "metadata": {},
+     "output_type": "execute_result"
+    }
+   ],
+   "source": [
+    "from dotenv import load_dotenv; load_dotenv()"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 11,
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "Downloading builder script: 100%|██████████| 4.03k/4.03k [00:00<00:00, 14.0MB/s]\n",
+      "Downloading readme: 100%|██████████| 6.84k/6.84k [00:00<00:00, 4.53MB/s]\n",
+      "Downloading data: 100%|██████████| 81.4M/81.4M [00:09<00:00, 8.96MB/s]\n",
+      "Generating train split: 100%|██████████| 1600000/1600000 [00:24<00:00, 64850.86 examples/s]\n",
+      "Generating test split: 100%|██████████| 498/498 [00:00<00:00, 22408.02 examples/s]\n"
+     ]
+    }
+   ],
+   "source": [
+    "# Get Data\n",
+    "#import csv\n",
+    "\n",
+    "\n",
+    "from datasets import load_dataset\n",
+    "ds = load_dataset(\"stanfordnlp/sentiment140\")\n"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 23,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "docs = ds['train'][:2000]['text']"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 24,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "# Get our encoder to encode \n",
+    "from sentence_transformers import SentenceTransformer \n",
+    "model = SentenceTransformer('sentence-transformers/all-MiniLM-L6-v2')\n",
+    "data_emb = model.encode(docs) #16 seconds on M1 Mac 8gb\n"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 28,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "# Now set up the DB to acceept the data\n",
+    "\n",
+    "import chromadb\n",
+    "\n",
+    "chroma_client = chromadb.Client()\n",
+    "collection = chroma_client.create_collection(name=\"SampleDB\")\n",
+    "\n",
+    "collection.add(\n",
+    "    embeddings=data_emb.tolist(),\n",
+    "    documents=docs,\n",
+    "    ids=[str(idx) for idx in range(len(data_emb))]) # Doc ID's are requried\n",
+    "\n",
+    "\n"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 42,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "#prep the question in the encoding space\n",
+    "\n",
+    "question = 'What is the status of the chalkboard?'\n",
+    "question_emb = model.encode(question)\n",
+    "\n",
+    "results = collection.query(query_embeddings=question_emb.tolist(), n_results=10)\n",
+    "\n",
+    "context = ' '.join(results['documents'][0]) # Pulling out a lists of lists\n",
+    "\n",
+    "\n"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 63,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "# Call the models to derterim the answer by response\n",
+    "\n",
+    "prompt = f'Given the following data, Please answer the question:  \\n\\n ##question \\n {question}\\n\\n ##context \\n {context}'\n",
+    "\n",
+    "\n",
+    "from aimodels.client import MultiFMClient\n",
+    "client = MultiFMClient()\n",
+    "\n",
+    "messages = [\n",
+    "    {\"role\": \"system\", \"content\": \"You are a helpful agent, who answers with brevity. \"},\n",
+    "    {\"role\": \"user\", \"content\": prompt},\n",
+    "]\n"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 54,
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "The status of the chalkboard is: USELESS (because there is no chalk).\n"
+     ]
+    }
+   ],
+   "source": [
+    "#groq_llama3_8b = \"groq:llama3-8b-8192\" \n",
+    "response = client.chat.completions.create(model=\"groq:llama3-70b-8192\", messages=messages)\n",
+    "\n",
+    "print(response.choices[0].message.content)\n"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 45,
+   "metadata": {},
+   "outputs": [
+    {
+     "data": {
+      "text/plain": [
+       "\"Damn... I don't have any chalk! MY CHALKBOARD IS USELESS \""
+      ]
+     },
+     "execution_count": 45,
+     "metadata": {},
+     "output_type": "execute_result"
+    }
+   ],
+   "source": [
+    "results['documents'][0][0]"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 64,
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Based on the context provided, the status of the chalkboard is useless because the person doesn't have any chalk.\n"
+     ]
+    }
+   ],
+   "source": [
+    "client = MultiFMClient() #have to reinaiatize client or for the API key. \n",
+    "\n",
+    "response = client.chat.completions.create(model=\"anthropic:claude-3-opus-20240229\", messages=messages)\n",
+    "print(response.choices[0].message.content)"
+   ]
+  }
+ ],
+ "metadata": {
+  "kernelspec": {
+   "display_name": ".venv",
+   "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": "3.11.1"
+  }
+ },
+ "nbformat": 4,
+ "nbformat_minor": 2
+}