agentops/examples/anthropic/anthropic-example-async.ipynb

298 lines
10 KiB
Plaintext

{
"cells": [
{
"cell_type": "markdown",
"metadata": {},
"source": [
"# Anthropic Async Example\n",
"\n",
"Anthropic supports both sync and async! This is great because we can wait for functions to finish before we use them! \n",
"\n",
"In this example, we will make a program called \"Titan Support Protocol.\" In this example, we will assign our mech a personality type and have a message generated based on our Titan's health (Which we randomly choose). We also send four generated UUIDs which are generated while the LLM runs"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"First, we start by importing Agentops and Anthropic"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {
"execution": {
"iopub.execute_input": "2024-11-09T19:24:21.051231Z",
"iopub.status.busy": "2024-11-09T19:24:21.050842Z",
"iopub.status.idle": "2024-11-09T19:24:46.728962Z",
"shell.execute_reply": "2024-11-09T19:24:46.727711Z",
"shell.execute_reply.started": "2024-11-09T19:24:21.051179Z"
},
"trusted": true
},
"outputs": [],
"source": [
"%pip install agentops\n",
"%pip install anthropic"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"Setup our generic default statements"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {
"execution": {
"iopub.execute_input": "2024-11-09T19:24:46.731735Z",
"iopub.status.busy": "2024-11-09T19:24:46.731341Z",
"iopub.status.idle": "2024-11-09T19:24:47.550169Z",
"shell.execute_reply": "2024-11-09T19:24:47.549415Z",
"shell.execute_reply.started": "2024-11-09T19:24:46.731687Z"
},
"trusted": true
},
"outputs": [],
"source": [
"from anthropic import Anthropic\n",
"import agentops\n",
"from dotenv import load_dotenv\n",
"import os\n",
"import random\n",
"import asyncio\n",
"import uuid"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"And set our API keys."
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {
"execution": {
"iopub.execute_input": "2024-11-09T19:48:37.019670Z",
"iopub.status.busy": "2024-11-09T19:48:37.018784Z",
"iopub.status.idle": "2024-11-09T19:48:37.024482Z",
"shell.execute_reply": "2024-11-09T19:48:37.023495Z",
"shell.execute_reply.started": "2024-11-09T19:48:37.019626Z"
},
"trusted": true
},
"outputs": [],
"source": [
"load_dotenv()\n",
"os.environ[\"AGENTOPS_API_KEY\"] = os.getenv(\"AGENTOPS_API_KEY\", \"your_api_key_here\")\n",
"os.environ[\"ANTHROPIC_API_KEY\"] = os.getenv(\"ANTHROPIC_API_KEY\", \"your_anthropic_api_key_here\")"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"\n",
"Now let's set the client as Anthropic and open an agentops trace!"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {
"execution": {
"iopub.execute_input": "2024-11-09T19:48:26.615366Z",
"iopub.status.busy": "2024-11-09T19:48:26.614702Z",
"iopub.status.idle": "2024-11-09T19:48:26.630956Z",
"shell.execute_reply": "2024-11-09T19:48:26.630026Z",
"shell.execute_reply.started": "2024-11-09T19:48:26.615326Z"
},
"trusted": true
},
"outputs": [],
"source": [
"client = Anthropic()"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {
"trusted": true
},
"outputs": [],
"source": [
"agentops.init(tags=[\"anthropic-async\", \"agentops-example\"])"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"Now we create three personality presets; \n",
"\n",
"Legion is a relentless and heavy-hitting Titan that embodies brute strength and defensive firepower, Northstar is a precise and agile sniper that excels in long-range combat and flight, while Ronin is a swift and aggressive melee specialist who thrives on close-quarters hit-and-run tactics."
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {
"execution": {
"iopub.execute_input": "2024-11-09T19:48:45.831654Z",
"iopub.status.busy": "2024-11-09T19:48:45.830897Z",
"iopub.status.idle": "2024-11-09T19:48:45.835837Z",
"shell.execute_reply": "2024-11-09T19:48:45.835037Z",
"shell.execute_reply.started": "2024-11-09T19:48:45.831616Z"
},
"trusted": true
},
"outputs": [],
"source": [
"TitanPersonality = [\n",
" \"Legion is a relentless and heavy-hitting Titan that embodies brute strength and defensive firepower. He speaks bluntly.,\",\n",
" \"Northstar is a precise and agile sniper that excels in long-range combat and flight. He speaks with an edge of coolness to him\",\n",
" \"Ronin is a swift and aggressive melee specialist who thrives on close-quarters hit-and-run tactics. He talks like a Samurai might.\",\n",
"]"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"And our comabt log generator! We select from four health presets!"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {
"execution": {
"iopub.execute_input": "2024-11-09T19:48:47.703344Z",
"iopub.status.busy": "2024-11-09T19:48:47.702974Z",
"iopub.status.idle": "2024-11-09T19:48:47.707915Z",
"shell.execute_reply": "2024-11-09T19:48:47.706767Z",
"shell.execute_reply.started": "2024-11-09T19:48:47.703308Z"
},
"trusted": true
},
"outputs": [],
"source": [
"TitanHealth = [\n",
" \"Fully functional\",\n",
" \"Slightly Damaged\",\n",
" \"Moderate Damage\",\n",
" \"Considerable Damage\",\n",
" \"Near Destruction\",\n",
"]"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"Now to the real core of this; making our message stream! We create this as a function we can call later! I create examples since the LLM's context size can handle it!"
]
},
{
"cell_type": "code",
"metadata": {
"execution": {
"iopub.execute_input": "2024-11-09T19:49:04.543561Z",
"iopub.status.busy": "2024-11-09T19:49:04.543172Z",
"iopub.status.idle": "2024-11-09T19:49:04.552542Z",
"shell.execute_reply": "2024-11-09T19:49:04.551542Z",
"shell.execute_reply.started": "2024-11-09T19:49:04.543522Z"
},
"trusted": true
},
"outputs": [],
"source": "Personality = {random.choice(TitanPersonality)}\nHealth = {random.choice(TitanHealth)}\n\n\nasync def req():\n # Start a streaming message request\n stream = client.messages.create(\n max_tokens=1024,\n model=\"claude-3-7-sonnet-20250219\",\n messages=[\n {\n \"role\": \"user\",\n \"content\": \"You are a Titan; a mech from Titanfall 2. Based on your titan's personality and status, generate a message for your pilot. If Near Destruction, make an all caps death message such as AVENGE ME or UNTIL NEXT TIME.\",\n },\n {\n \"role\": \"assistant\",\n \"content\": \"Personality: Legion is a relentless and heavy-hitting Titan that embodies brute strength and defensive firepower. He speaks bluntly. Status: Considerable Damage\",\n },\n {\n \"role\": \"assistant\",\n \"content\": \"Heavy damage detected. Reinforcements would be appreciated, but I can still fight.\",\n },\n {\n \"role\": \"user\",\n \"content\": \"You are a Titan; a mech from Titanfall 2. Based on your titan's personality and status, generate a message for your pilot. If Near Destruction, make an all caps death message such as AVENGE ME or UNTIL NEXT TIME.\",\n },\n {\n \"role\": \"assistant\",\n \"content\": f\"Personality: {Personality}. Status: {Health}\",\n },\n ],\n stream=True,\n )\n\n response = \"\"\n for event in stream:\n if event.type == \"content_block_delta\":\n response += event.delta.text\n elif event.type == \"message_stop\":\n break # Exit the loop when the message completes\n\n return response\n\n\nasync def generate_uuids():\n uuids = [str(uuid.uuid4()) for _ in range(4)]\n return uuids"
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"Now we wrap it all in a nice main function! Run this for the magic to happen! Go to your AgentOps dashboard and you should see this trace reflected!\n"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {
"execution": {
"iopub.execute_input": "2024-11-09T19:49:06.598601Z",
"iopub.status.busy": "2024-11-09T19:49:06.597657Z",
"iopub.status.idle": "2024-11-09T19:49:07.565561Z",
"shell.execute_reply": "2024-11-09T19:49:07.564647Z",
"shell.execute_reply.started": "2024-11-09T19:49:06.598561Z"
},
"trusted": true
},
"outputs": [],
"source": [
"async def main():\n",
" # Start both tasks concurrently\n",
" uuids, message = await asyncio.gather(generate_uuids(), req())\n",
"\n",
" print(\"Personality:\", Personality)\n",
" print(\"Health Status:\", Health)\n",
" print(\"Combat log incoming from encrypted area\")\n",
"\n",
" print(\"Verification matrix activated.:\")\n",
" for u in uuids:\n",
" print(u)\n",
"\n",
" print(\". Titan Message: \", message)\n",
"\n",
"\n",
"# Run the main function\n",
"await main()"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"We can observe the trace in the AgentOps dashboard by going to the trace URL provided above."
]
}
],
"metadata": {
"kaggle": {
"accelerator": "gpu",
"dataSources": [],
"dockerImageVersionId": 30786,
"isGpuEnabled": true,
"isInternetEnabled": true,
"language": "python",
"sourceType": "notebook"
},
"kernelspec": {
"display_name": "agentops (3.11.11)",
"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": "3.11.11"
}
},
"nbformat": 4,
"nbformat_minor": 4
}