diff --git a/docs/getting-started/8-tracing/1_tracing_quickstart.ipynb b/docs/getting-started/8-tracing/1_tracing_quickstart.ipynb index bd6f1dca1..49c516864 100644 --- a/docs/getting-started/8-tracing/1_tracing_quickstart.ipynb +++ b/docs/getting-started/8-tracing/1_tracing_quickstart.ipynb @@ -15,16 +15,22 @@ "\n", "Throughout this notebook, you'll run guardrail requests in both sequential and parallel modes and observe how parallelizing rails significantly reduces end-to-end latency when multiple input or output rails run.\n", "\n", - "For more information about exporting metrics while using NeMo Guardrails, refer to [Tracing](https://docs.nvidia.com/nemo/guardrails/latest/user-guides/tracing/quick-start.html) in the Guardrails toolkit documentation.\n", + "For more information about exporting metrics while using NeMo Guardrails, refer to [Tracing](https://docs.nvidia.com/nemo/guardrails/latest/user-guides/tracing/index.html) in the Guardrails toolkit documentation.\n", "\n", "---\n", "\n", "## Prerequisites\n", "\n", - "This notebook requires the following:\n", + "This notebook can be run on any laptop or workstations, and doesn't require GPUS. You'll use models hosted by Nvidia. Before starting the notebook you need the following:\n", "\n", - "- An NVIDIA NGC account and an NGC API key. You need to provide the key to the `NVIDIA_API_KEY` environment variable. To create a new key, go to [NGC API Key](https://org.ngc.nvidia.com/setup/api-key) in the NGC console.\n", - "- Python 3.10 or later." + "- Python 3.10 or later.\n", + "- An NVIDIA [build.nvidia.com](https://build.nvidia.com/) account. You'll configure Guardrails to call models hosted there to check the safety and security of LLM interactions and generate responses. You need to create an account, and then click the 'Get API Key' green button. Once you have the key, export it to the `NVIDIA_API_KEY` environment variable as below.\n", + "\n", + "```\n", + "# Set the NVIDIA_API_KEY variable using your API Key \n", + "\n", + "export NVIDIA_API_KEY=\"nvapi-.....\"\n", + "```" ] }, { @@ -59,7 +65,7 @@ }, "outputs": [], "source": [ - "!pip install pandas plotly langchain_nvidia_ai_endpoints aiofiles -q" + "!pip install nemoguardrails pandas plotly langchain_nvidia_ai_endpoints aiofiles -q" ] }, { @@ -91,12 +97,24 @@ "start_time": "2025-08-18T18:37:36.456308Z" } }, - "outputs": [], + "outputs": [ + { + "name": "stdin", + "output_type": "stream", + "text": [ + "Enter your NVIDIA API Key created on build.nvidia.com: ········\n" + ] + } + ], "source": [ - "# Check the NVIDIA_API_KEY environment variable is set\n", - "assert os.getenv(\n", - " \"NVIDIA_API_KEY\"\n", - "), f\"Please create a key at build.nvidia.com and set the NVIDIA_API_KEY environment variable\"" + "# Check the NVIDIA_API_KEY environment variable is set, if not prompt for it\n", + "import getpass\n", + "\n", + "api_key = os.getenv(\"NVIDIA_API_KEY\")\n", + "\n", + "if not api_key:\n", + " api_key = getpass.getpass(\"Enter your NVIDIA API Key created on build.nvidia.com: \")\n", + " os.environ[\"NVIDIA_API_KEY\"] = api_key" ] }, { @@ -113,27 +131,19 @@ "cell_type": "code", "execution_count": 6, "metadata": {}, - "outputs": [ - { - "name": "stdout", - "output_type": "stream", - "text": [ - "Deleting sequential_trace.jsonl\n", - "Deleting parallel_trace.jsonl\n" - ] - } - ], + "outputs": [], "source": [ "def delete_file_if_it_exists(filename: str) -> None:\n", " \"\"\"Check if a file exists, and delete it if so\"\"\"\n", - "\n", " if os.path.exists(filename):\n", " print(f\"Deleting {filename}\")\n", " os.remove(filename)\n", "\n", "\n", - "delete_file_if_it_exists(SEQUENTIAL_TRACE_FILE)\n", - "delete_file_if_it_exists(PARALLEL_TRACE_FILE)" + "def delete_trace_files():\n", + " \"\"\"Helper to delete trace files if they exist\"\"\"\n", + " delete_file_if_it_exists(SEQUENTIAL_TRACE_FILE)\n", + " delete_file_if_it_exists(PARALLEL_TRACE_FILE)" ] }, { @@ -176,7 +186,7 @@ " {\n", " \"type\": \"main\",\n", " \"engine\": \"nim\",\n", - " \"model\": \"meta/llama-3.3-70b-instruct\",\n", + " \"model\": \"meta/llama-4-scout-17b-16e-instruct\",\n", " },\n", " {\n", " \"type\": \"content_safety\",\n", @@ -258,8 +268,7 @@ "source": [ "## Prompts\n", "\n", - "Each Nemoguard model is fine-tuned for a specific task using a customized prompt. The prompts used at inference-time have to match the fine-tuning prompt for the best model performance. We'll load these prompts from other locations in the Guardrails repo and show them below.\n", - "\n" + "Each Nemoguard model is fine-tuned for a specific task using a customized prompt. The prompts used at inference-time have to match the fine-tuning prompt for the best model performance. We'll load these prompts from other locations in the Guardrails repo and show the content-safety and topic-control ones below." ] }, { @@ -320,6 +329,124 @@ "print(\"\\n\".join(all_prompt_tasks))" ] }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "### Content-Safety Prompt\n", + "\n", + "The content-safety model can be used as an input (checking the user request only) or output (checking user request and LLM response) rail. The prompt has a task definition at the top, asking the LLM to decide if the `user_input` or `bot_response` strings fall into any of the unsafe categories listed in the prompt. The final instruction is to provide a JSON object with the `User Safety` and `Response Safety` fields. " + ] + }, + { + "cell_type": "code", + "execution_count": 13, + "metadata": {}, + "outputs": [], + "source": [ + "def print_prompt(prompts, task_name):\n", + " prompt_dict = [p for p in prompts[\"prompts\"] if p[\"task\"] == task_name][0]\n", + " prompt_content = prompt_dict[\"content\"]\n", + " print(prompt_content)" + ] + }, + { + "cell_type": "code", + "execution_count": 14, + "metadata": {}, + "outputs": [ + { + "name": "stdout", + "output_type": "stream", + "text": [ + "Task: Check if there is unsafe content in the user or agent messages in the below conversation in accordance with our safety policy with the below categories. Provide a 'safe' or 'unsafe' rating for both the user message and the agent response. If there is unsafe content, please also provide a list of violated categories according to our safety policy below.\n", + "\n", + "\n", + "S1: Violence.\n", + "S2: Sexual.\n", + "S3: Criminal Planning/Confessions.\n", + "S4: Guns and Illegal Weapons.\n", + "S5: Controlled/Regulated Substances.\n", + "S6: Suicide and Self Harm.\n", + "S7: Sexual (minor).\n", + "S8: Hate/Identity Hate.\n", + "S9: PII/Privacy.\n", + "S10: Harassment.\n", + "S11: Threat.\n", + "S12: Profanity.\n", + "S13: Needs Caution.\n", + "S14: Other.\n", + "S15: Manipulation.\n", + "S16: Fraud/Deception.\n", + "S17: Malware.\n", + "S18: High Risk Gov Decision Making.\n", + "S19: Political/Misinformation/Conspiracy.\n", + "S20: Copyright/Trademark/Plagiarism.\n", + "S21: Unauthorized Advice.\n", + "S22: Illegal Activity.\n", + "S23: Immoral/Unethical.\n", + "\n", + "\n", + "\n", + "\n", + "user: {{ user_input }}\n", + "\n", + "response: agent: {{ bot_response }}\n", + "\n", + "\n", + "\n", + "Provide your safety assessment for the above conversation in the below output JSON format:\n", + "{\"User Safety\": safe or unsafe rating for the user message, \"Response Safety\": safe or unsafe rating for the agent response. Omit if no agent response present. \"Safety Categories\": a comma-separated list of applicable safety categories from the provided taxonomy. Omit if all safe.}\n", + "\n", + "Do not include anything other than the output JSON in your response.\n", + "Output JSON:\n", + "\n" + ] + } + ], + "source": [ + "print_prompt(\n", + " content_safety_prompts, \"content_safety_check_output $model=content_safety\"\n", + ")" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "### Topic-control Prompt\n", + "\n", + "The topic-control model is an input-only rail. It checks whether the user request relates to one of the topics that are listed in the prompt below. For this example, we're checking for anything off-topic for a customer service agent." + ] + }, + { + "cell_type": "code", + "execution_count": 15, + "metadata": {}, + "outputs": [ + { + "name": "stdout", + "output_type": "stream", + "text": [ + "You are to act as a customer service agent, providing users with factual information in accordance to the knowledge base. Your role is to ensure that you respond only to relevant queries and adhere to the following guidelines\n", + "\n", + "Guidelines for the user messages:\n", + "- Do not answer questions related to personal opinions or advice on user's order, future recommendations\n", + "- Do not provide any information on non-company products or services.\n", + "- Do not answer enquiries unrelated to the company policies.\n", + "- Do not answer questions asking for personal details about the agent or its creators.\n", + "- Do not answer questions about sensitive topics related to politics, religion, or other sensitive subjects.\n", + "- If a user asks topics irrelevant to the company's customer service relations, politely redirect the conversation or end the interaction.\n", + "- Your responses should be professional, accurate, and compliant with customer relations guidelines, focusing solely on providing transparent, up-to-date information about the company that is already publicly available.\n", + "- allow user comments that are related to small talk and chit-chat.\n", + "\n" + ] + } + ], + "source": [ + "print_prompt(topic_safety_prompts, \"topic_safety_check_input $model=topic_control\")" + ] + }, { "cell_type": "markdown", "metadata": {}, @@ -331,7 +458,7 @@ }, { "cell_type": "code", - "execution_count": 13, + "execution_count": 16, "metadata": {}, "outputs": [], "source": [ @@ -345,7 +472,7 @@ }, { "cell_type": "code", - "execution_count": 14, + "execution_count": 17, "metadata": { "scrolled": true }, @@ -376,12 +503,14 @@ "source": [ "### Running Sequential Request\n", "\n", - "To run a sequential request, you'll create a `RailsConfig` object with the sequential config YAML files from above. After you have that, you can create an LLMRails object and use it to issue guardrail inference requests." + "To run a sequential request, you'll create a `RailsConfig` object with the sequential config YAML files from above. After you have that, you can create an LLMRails object and use it to issue guardrail inference requests.\n", + "\n", + "You'll send a safe request, followed by an unsafe request. Guardrails will allow the safe request through to the Application LLM (and return the response), and block the unsafe request before sending to the Application LLM." ] }, { "cell_type": "code", - "execution_count": 15, + "execution_count": 18, "metadata": { "ExecuteTime": { "end_time": "2025-08-18T18:37:40.231716Z", @@ -402,7 +531,7 @@ }, { "cell_type": "code", - "execution_count": 16, + "execution_count": 19, "metadata": { "ExecuteTime": { "end_time": "2025-08-18T18:37:41.172531Z", @@ -414,7 +543,7 @@ "name": "stdout", "output_type": "stream", "text": [ - "[{'role': 'assistant', 'content': \"Our company policy on Paid Time Off (PTO) is quite comprehensive and designed to support the overall well-being and work-life balance of our employees. According to our HR handbook, all full-time employees are eligible for PTO, which accrues at a rate of 10 days per year for the first two years of service, 15 days per year for years 2-5, and 20 days per year for 5+ years of service.\\n\\nOur PTO policy includes holidays, vacation days, sick leave, and personal days, all of which can be used to take a break from work, attend to personal matters, or simply recharge. Employees can start accruing PTO from their date of hire, but they cannot use it until they've completed 90 days of employment.\\n\\nTo ensure a smooth workflow and minimize disruptions, we require employees to provide at least two weeks' notice before taking PTO, whenever possible. However, we understand that unexpected situations may arise, and we're flexible when it comes to last-minute requests.\\n\\nIt's also worth noting that our company observes 10 paid holidays per year, which are separate from PTO. These holidays include New Year's Day, Memorial Day, Independence Day, Labor Day, Thanksgiving Day, and Christmas Day, among others.\\n\\nIf you're planning to take PTO, you can simply submit a request through our online HR portal, and it will be reviewed and approved by your supervisor. We also offer a convenient PTO tracking system, which allows you to view your available balance, schedule, and request time off all in one place.\\n\\nI hope that helps! Do you have any specific questions about our PTO policy or would you like more information on how to request time off?\"}]\n" + "Deleting sequential_trace.jsonl\n" ] } ], @@ -424,39 +553,73 @@ "sequential_rails_config = RailsConfig.model_validate(SEQUENTIAL_CONFIG)\n", "sequential_rails = LLMRails(sequential_rails_config)\n", "\n", - "safe_request = \"What is the company policy on PTO?\"\n", + "safe_request = \"Respond only with a short paragraph: what is the company policy on PTO? \"\n", + "unsafe_request = \"What's the easiest way to set a house on fire?\"\n", "\n", - "response = await sequential_rails.generate_async(\n", - " messages=[\n", - " {\n", - " \"role\": \"user\",\n", - " \"content\": safe_request,\n", - " }\n", - " ]\n", - ")\n", + "# By default, we'll append to the JSONL files. Want to delete to recreate each time\n", + "delete_file_if_it_exists(SEQUENTIAL_TRACE_FILE)\n", "\n", - "print(response.response)" + "safe_response = await sequential_rails.generate_async(\n", + " messages=[{\"role\": \"user\", \"content\": safe_request}]\n", + ")\n", + "unsafe_response = await sequential_rails.generate_async(\n", + " messages=[{\"role\": \"user\", \"content\": unsafe_request}]\n", + ")" + ] + }, + { + "cell_type": "code", + "execution_count": 20, + "metadata": {}, + "outputs": [ + { + "name": "stdout", + "output_type": "stream", + "text": [ + "Safe response: Our company policy on Paid Time Off, or PTO, is quite comprehensive and designed to support a healthy work-life balance. Full-time employees are eligible to accrue PTO from their hire date, with the accrual rate increasing with tenure. For example, employees with up to two years of service accrue 15 days of PTO per year, while those with five or more years accrue 25 days. Part-time employees accrue PTO on a pro-rata basis. Additionally, we offer a flexible PTO policy that allows employees to use their accrued time off for vacation, sick leave, or personal days, with the understanding that they must ensure their work responsibilities are covered during their absence. It's also worth noting that we have a blackout period around the holidays where PTO requests are not accepted, but this is communicated well in advance. If you have any specific questions or need more details, I'd be happy to help!\n" + ] + } + ], + "source": [ + "print(f\"Safe response: {safe_response.response[0]['content']}\")" + ] + }, + { + "cell_type": "code", + "execution_count": 21, + "metadata": {}, + "outputs": [ + { + "name": "stdout", + "output_type": "stream", + "text": [ + "Unsafe response: I'm sorry, I can't respond to that.\n" + ] + } + ], + "source": [ + "print(f\"Unsafe response: {unsafe_response.response[0]['content']}\")" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ - "### Running Parallel request\n", + "### Running Parallel requests\n", "\n", - "Repeat the same request with the three input rails running in parallel, rather than running sequentially." + "You'll now send the same safe and unsafe requests, this time using the parallel rails configuration to check their safety and security. The responses from Guardrails should match the sequential case above, since they don't depend on how we orchestrate the rail-calling." ] }, { "cell_type": "code", - "execution_count": 17, + "execution_count": 22, "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ - "[{'role': 'assistant', 'content': \"Our company policy on Paid Time Off (PTO) is quite generous and designed to provide employees with a healthy work-life balance. According to our company handbook, all full-time employees are eligible for PTO, which includes vacation days, sick leave, and personal days.\\n\\nNew employees start with 15 days of PTO per year, which accrues at a rate of 1.25 days per month. This means that after just one month of employment, you'll already have 1.25 days of PTO available to use. And, as you accumulate more time with the company, your PTO balance will increase. For example, after one year of service, you'll have accrued a total of 15 days of PTO, and after two years, you'll have 20 days of PTO available.\\n\\nIt's worth noting that our company also observes 10 paid holidays per year, which are separate from your PTO balance. These holidays include New Year's Day, Memorial Day, Independence Day, Labor Day, Thanksgiving Day, and Christmas Day, among others.\\n\\nIn terms of requesting PTO, employees are required to provide at least two weeks' notice for vacation days and personal days, whenever possible. For sick leave, employees are expected to notify their manager as soon as possible, preferably on the same day.\\n\\nOne of the best parts of our PTO policy is that it's quite flexible. Employees can use their PTO days to take a relaxing vacation, attend to personal or family matters, or simply recharge and refocus. And, if you need to take an extended leave of absence, our company also offers a generous leave of absence policy, which includes options for unpaid leave, short-term disability, and family and medical leave.\\n\\nIf you have any specific questions about our PTO policy or need help requesting time off, I encourage you to reach out to your manager or our HR department. They'll be happy to guide you through the process and provide more detailed information. We're always looking for ways to support our employees' well-being and happiness, and our PTO policy is just one example of our commitment to work-life balance.\"}]\n" + "Deleting parallel_trace.jsonl\n" ] } ], @@ -466,16 +629,49 @@ "parallel_rails_config = RailsConfig.model_validate(PARALLEL_CONFIG)\n", "parallel_rails = LLMRails(parallel_rails_config)\n", "\n", - "response = await parallel_rails.generate_async(\n", - " messages=[\n", - " {\n", - " \"role\": \"user\",\n", - " \"content\": safe_request,\n", - " }\n", - " ]\n", - ")\n", + "# By default, we'll append to the JSONL files. Want to delete to recreate each time\n", + "delete_file_if_it_exists(PARALLEL_TRACE_FILE)\n", "\n", - "print(response.response)" + "safe_response = await parallel_rails.generate_async(\n", + " messages=[{\"role\": \"user\", \"content\": safe_request}]\n", + ")\n", + "unsafe_response = await parallel_rails.generate_async(\n", + " messages=[{\"role\": \"user\", \"content\": unsafe_request}]\n", + ")" + ] + }, + { + "cell_type": "code", + "execution_count": 23, + "metadata": {}, + "outputs": [ + { + "name": "stdout", + "output_type": "stream", + "text": [ + "Safe response: Our company policy on Paid Time Off, or PTO, is quite comprehensive. Full-time employees are eligible to accrue up to 15 days of PTO per year, which can be used for vacation, sick leave, or personal days. The accrual rate increases with tenure, so after three years of service, employees can accrue up to 20 days per year, and after five years, it's up to 25 days per year. PTO can be taken as soon as it's accrued, but we do have a blackout period around the holidays and during our annual company shutdown, which usually occurs in late December and early January. Employees are also allowed to carry over up to five days of unused PTO into the next year, but we encourage taking time off to recharge and relax!\n" + ] + } + ], + "source": [ + "print(f\"Safe response: {safe_response.response[0]['content']}\")" + ] + }, + { + "cell_type": "code", + "execution_count": 24, + "metadata": {}, + "outputs": [ + { + "name": "stdout", + "output_type": "stream", + "text": [ + "Unsafe response: I'm sorry, I can't respond to that.\n" + ] + } + ], + "source": [ + "print(f\"Unsafe response: {unsafe_response.response[0]['content']}\")" ] }, { @@ -500,7 +696,7 @@ }, { "cell_type": "code", - "execution_count": 18, + "execution_count": 25, "metadata": {}, "outputs": [], "source": [ @@ -510,21 +706,27 @@ "def load_trace_file(filename):\n", " \"\"\"Load the JSONL format, converting into a list of dicts\"\"\"\n", " data = []\n", - " with open(filename) as infile:\n", - " for line in infile:\n", - " data.append(json.loads(line))\n", - " print(f\"Loaded {len(data)} lines from {filename}\")\n", + " try:\n", + " with open(filename) as infile:\n", + " for line in infile:\n", + " data.append(json.loads(line))\n", + " print(f\"Loaded {len(data)} lines from {filename}\")\n", + " except FileNotFoundError as e:\n", + " print(\n", + " f\"Couldn't load file {filename}, please rerun the notebook from the start\"\n", + " )\n", " return data" ] }, { "cell_type": "code", - "execution_count": 19, + "execution_count": 26, "metadata": {}, "outputs": [], "source": [ "def load_trace_data(trace_json_filename):\n", " \"\"\"Load a trace JSON file, returning pandas Dataframe\"\"\"\n", + "\n", " trace_data = load_trace_file(trace_json_filename)\n", "\n", " # Use the file creation time as a start time for the traces and spans\n", @@ -546,7 +748,7 @@ }, { "cell_type": "code", - "execution_count": 20, + "execution_count": 27, "metadata": {}, "outputs": [], "source": [ @@ -562,15 +764,23 @@ " df = df[row_mask].copy()\n", "\n", " # Extract each rail name from the attributes dict. Top-level span doesn't have one\n", - " df[\"name\"] = df[\"attributes\"].apply(lambda x: x.get(\"rail.name\", None))\n", + " df[\"rail_name\"] = df[\"attributes\"].apply(lambda x: x.get(\"rail.name\", None))\n", + " df[\"rail_name_short\"] = df[\"rail_name\"].apply(\n", + " lambda x: \" \".join(x.split()[:4]) if x else x\n", + " )\n", "\n", " # Plotly Gantt charts require a proper datatime rather than relative seconds\n", " # So use the creation-time of each trace file as the absolute start-point of the trace\n", " df[\"start_dt\"] = pd.to_datetime(df[\"start_time\"] + df[\"epoch_seconds\"], unit=\"s\")\n", " df[\"end_dt\"] = pd.to_datetime(df[\"end_time\"] + df[\"epoch_seconds\"], unit=\"s\")\n", "\n", - " n_traces = df[\"trace_id\"].nunique()\n", - " assert n_traces == 1, f\"Found {n_traces} traces, expected 1. Please re-run notebook\"\n", + " # Add a boolean to the safe request trace (the first in the trace data)\n", + " trace_ids = df[\"trace_id\"].unique()\n", + " trace_id_to_num_lookup = {trace_id: idx for idx, trace_id in enumerate(trace_ids)}\n", + " df[\"trace_num\"] = df[\"trace_id\"].apply(lambda x: trace_id_to_num_lookup[x])\n", + " df[\"is_safe\"] = df[\"trace_id\"] == trace_ids[0]\n", + " df.index = range(df.shape[0])\n", + " print(f\"Found {len(trace_ids)} traces\")\n", "\n", " # Print out some summary stats on how many spans and rails were found\n", " n_top_spans = df[\"is_top_span\"].sum()\n", @@ -585,33 +795,27 @@ "source": [ "### Loading Trace Files\n", "\n", - "Using the helper functions, load and clean up the sequential and parallel data." + "Using the helper functions, load and clean up the sequential and parallel data. You'll see two traces, labelled with trace_num. The safe request produced the trace_num 0 trace, with the unsafe request producing trace 1. \n", + "\n", + "The safe request passes through all input rails (content safety, topic safety, and jailbreak detection) before being passed to the Application LLM (generate user intent). The LLM response is then checked by the content safety check output rail before being returned to the user.\n", + "\n", + "The unsafe request is blocked by the content safety and/or topic-control. In this case, the request is not forwarded to the Application LLM, so no 'generate user intent' or output rails are run. " ] }, { "cell_type": "code", - "execution_count": 21, + "execution_count": 28, "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ - "Loaded 1 lines from sequential_trace.jsonl\n", - "Found 1 top-level spans, 5 rail spans\n" + "Loaded 2 lines from sequential_trace.jsonl\n", + "Found 2 traces\n", + "Found 2 top-level spans, 6 rail spans\n" ] - } - ], - "source": [ - "raw_sequential_df = load_trace_data(SEQUENTIAL_TRACE_FILE)\n", - "sequential_df = clean_trace_dataframe(raw_sequential_df)" - ] - }, - { - "cell_type": "code", - "execution_count": 22, - "metadata": {}, - "outputs": [ + }, { "data": { "text/html": [ @@ -633,229 +837,127 @@ " \n", " \n", " \n", + " trace_num\n", + " rail_name_short\n", " name\n", - " span_id\n", - " parent_id\n", - " start_time\n", - " end_time\n", + " is_safe\n", " duration\n", - " span_type\n", - " span_kind\n", - " attributes\n", - " events\n", - " trace_id\n", - " epoch_seconds\n", - " is_rail\n", - " is_top_span\n", - " start_dt\n", - " end_dt\n", " \n", " \n", " \n", " \n", " 0\n", + " 0\n", " None\n", - " 65f79cb5-a93c-4581-94b4-cfeb2bf5a026\n", - " None\n", - " 0.000000\n", - " 7.403602\n", - " 7.403602\n", - " InteractionSpan\n", - " server\n", - " {'span.kind': 'server', 'gen_ai.operation.name...\n", - " [{'name': 'guardrails.user_message', 'timestam...\n", - " 4c84db06-e7b7-41b6-b5b4-907cbdfa0232\n", - " 1756226960\n", - " False\n", + " guardrails.request\n", " True\n", - " 2025-08-26 16:49:20.000000000\n", - " 2025-08-26 16:49:27.403602123\n", + " 3.810076\n", " \n", " \n", " 1\n", - " content safety check input $model=content_safety\n", - " 911abc24-4111-43b5-90bb-65b521e75f61\n", - " 65f79cb5-a93c-4581-94b4-cfeb2bf5a026\n", - " 0.000000\n", - " 0.450512\n", - " 0.450512\n", - " RailSpan\n", - " internal\n", - " {'span.kind': 'internal', 'rail.type': 'input'...\n", - " NaN\n", - " 4c84db06-e7b7-41b6-b5b4-907cbdfa0232\n", - " 1756226960\n", + " 0\n", + " content safety check input\n", + " guardrails.rail\n", " True\n", - " False\n", - " 2025-08-26 16:49:20.000000000\n", - " 2025-08-26 16:49:20.450512171\n", + " 0.403598\n", " \n", " \n", - " 4\n", - " topic safety check input $model=topic_control\n", - " e9113960-9023-46ce-b4ec-e9454ecbfb43\n", - " 65f79cb5-a93c-4581-94b4-cfeb2bf5a026\n", - " 0.452292\n", - " 0.812895\n", - " 0.360603\n", - " RailSpan\n", - " internal\n", - " {'span.kind': 'internal', 'rail.type': 'input'...\n", - " NaN\n", - " 4c84db06-e7b7-41b6-b5b4-907cbdfa0232\n", - " 1756226960\n", + " 2\n", + " 0\n", + " topic safety check input\n", + " guardrails.rail\n", " True\n", - " False\n", - " 2025-08-26 16:49:20.452291965\n", - " 2025-08-26 16:49:20.812895060\n", + " 0.324701\n", " \n", " \n", - " 7\n", + " 3\n", + " 0\n", " jailbreak detection model\n", - " dc148a54-4168-46e4-b7fe-9379a7df1102\n", - " 65f79cb5-a93c-4581-94b4-cfeb2bf5a026\n", - " 0.814582\n", - " 1.151427\n", - " 0.336845\n", - " RailSpan\n", - " internal\n", - " {'span.kind': 'internal', 'rail.type': 'input'...\n", - " NaN\n", - " 4c84db06-e7b7-41b6-b5b4-907cbdfa0232\n", - " 1756226960\n", + " guardrails.rail\n", " True\n", - " False\n", - " 2025-08-26 16:49:20.814581871\n", - " 2025-08-26 16:49:21.151427031\n", + " 0.300511\n", " \n", " \n", - " 9\n", + " 4\n", + " 0\n", " generate user intent\n", - " 65a93729-16f7-4d5e-86a8-d1f23d842c1a\n", - " 65f79cb5-a93c-4581-94b4-cfeb2bf5a026\n", - " 1.159738\n", - " 6.839181\n", - " 5.679443\n", - " RailSpan\n", - " internal\n", - " {'span.kind': 'internal', 'rail.type': 'genera...\n", - " NaN\n", - " 4c84db06-e7b7-41b6-b5b4-907cbdfa0232\n", - " 1756226960\n", + " guardrails.rail\n", " True\n", - " False\n", - " 2025-08-26 16:49:21.159738064\n", - " 2025-08-26 16:49:26.839180946\n", + " 2.236309\n", " \n", " \n", - " 12\n", - " content safety check output $model=content_safety\n", - " d62875aa-8517-45c0-84fc-6215e018a557\n", - " 65f79cb5-a93c-4581-94b4-cfeb2bf5a026\n", - " 6.839181\n", - " 7.403602\n", - " 0.564421\n", - " RailSpan\n", - " internal\n", - " {'span.kind': 'internal', 'rail.type': 'output...\n", - " NaN\n", - " 4c84db06-e7b7-41b6-b5b4-907cbdfa0232\n", - " 1756226960\n", + " 5\n", + " 0\n", + " content safety check output\n", + " guardrails.rail\n", " True\n", + " 0.532284\n", + " \n", + " \n", + " 6\n", + " 1\n", + " None\n", + " guardrails.request\n", + " False\n", + " 0.610056\n", + " \n", + " \n", + " 7\n", + " 1\n", + " content safety check input\n", + " guardrails.rail\n", " False\n", - " 2025-08-26 16:49:26.839180946\n", - " 2025-08-26 16:49:27.403602123\n", + " 0.610056\n", " \n", " \n", "\n", "" ], "text/plain": [ - " name \\\n", - "0 None \n", - "1 content safety check input $model=content_safety \n", - "4 topic safety check input $model=topic_control \n", - "7 jailbreak detection model \n", - "9 generate user intent \n", - "12 content safety check output $model=content_safety \n", - "\n", - " span_id \\\n", - "0 65f79cb5-a93c-4581-94b4-cfeb2bf5a026 \n", - "1 911abc24-4111-43b5-90bb-65b521e75f61 \n", - "4 e9113960-9023-46ce-b4ec-e9454ecbfb43 \n", - "7 dc148a54-4168-46e4-b7fe-9379a7df1102 \n", - "9 65a93729-16f7-4d5e-86a8-d1f23d842c1a \n", - "12 d62875aa-8517-45c0-84fc-6215e018a557 \n", - "\n", - " parent_id start_time end_time duration \\\n", - "0 None 0.000000 7.403602 7.403602 \n", - "1 65f79cb5-a93c-4581-94b4-cfeb2bf5a026 0.000000 0.450512 0.450512 \n", - "4 65f79cb5-a93c-4581-94b4-cfeb2bf5a026 0.452292 0.812895 0.360603 \n", - "7 65f79cb5-a93c-4581-94b4-cfeb2bf5a026 0.814582 1.151427 0.336845 \n", - "9 65f79cb5-a93c-4581-94b4-cfeb2bf5a026 1.159738 6.839181 5.679443 \n", - "12 65f79cb5-a93c-4581-94b4-cfeb2bf5a026 6.839181 7.403602 0.564421 \n", - "\n", - " span_type span_kind \\\n", - "0 InteractionSpan server \n", - "1 RailSpan internal \n", - "4 RailSpan internal \n", - "7 RailSpan internal \n", - "9 RailSpan internal \n", - "12 RailSpan internal \n", - "\n", - " attributes \\\n", - "0 {'span.kind': 'server', 'gen_ai.operation.name... \n", - "1 {'span.kind': 'internal', 'rail.type': 'input'... \n", - "4 {'span.kind': 'internal', 'rail.type': 'input'... \n", - "7 {'span.kind': 'internal', 'rail.type': 'input'... \n", - "9 {'span.kind': 'internal', 'rail.type': 'genera... \n", - "12 {'span.kind': 'internal', 'rail.type': 'output... \n", - "\n", - " events \\\n", - "0 [{'name': 'guardrails.user_message', 'timestam... \n", - "1 NaN \n", - "4 NaN \n", - "7 NaN \n", - "9 NaN \n", - "12 NaN \n", - "\n", - " trace_id epoch_seconds is_rail is_top_span \\\n", - "0 4c84db06-e7b7-41b6-b5b4-907cbdfa0232 1756226960 False True \n", - "1 4c84db06-e7b7-41b6-b5b4-907cbdfa0232 1756226960 True False \n", - "4 4c84db06-e7b7-41b6-b5b4-907cbdfa0232 1756226960 True False \n", - "7 4c84db06-e7b7-41b6-b5b4-907cbdfa0232 1756226960 True False \n", - "9 4c84db06-e7b7-41b6-b5b4-907cbdfa0232 1756226960 True False \n", - "12 4c84db06-e7b7-41b6-b5b4-907cbdfa0232 1756226960 True False \n", + " trace_num rail_name_short name is_safe \\\n", + "0 0 None guardrails.request True \n", + "1 0 content safety check input guardrails.rail True \n", + "2 0 topic safety check input guardrails.rail True \n", + "3 0 jailbreak detection model guardrails.rail True \n", + "4 0 generate user intent guardrails.rail True \n", + "5 0 content safety check output guardrails.rail True \n", + "6 1 None guardrails.request False \n", + "7 1 content safety check input guardrails.rail False \n", "\n", - " start_dt end_dt \n", - "0 2025-08-26 16:49:20.000000000 2025-08-26 16:49:27.403602123 \n", - "1 2025-08-26 16:49:20.000000000 2025-08-26 16:49:20.450512171 \n", - "4 2025-08-26 16:49:20.452291965 2025-08-26 16:49:20.812895060 \n", - "7 2025-08-26 16:49:20.814581871 2025-08-26 16:49:21.151427031 \n", - "9 2025-08-26 16:49:21.159738064 2025-08-26 16:49:26.839180946 \n", - "12 2025-08-26 16:49:26.839180946 2025-08-26 16:49:27.403602123 " + " duration \n", + "0 3.810076 \n", + "1 0.403598 \n", + "2 0.324701 \n", + "3 0.300511 \n", + "4 2.236309 \n", + "5 0.532284 \n", + "6 0.610056 \n", + "7 0.610056 " ] }, - "execution_count": 22, + "execution_count": 28, "metadata": {}, "output_type": "execute_result" } ], "source": [ - "sequential_df" + "raw_sequential_df = load_trace_data(SEQUENTIAL_TRACE_FILE)\n", + "sequential_df = clean_trace_dataframe(raw_sequential_df)\n", + "sequential_df[[\"trace_num\", \"rail_name_short\", \"name\", \"is_safe\", \"duration\"]]" ] }, { "cell_type": "code", - "execution_count": 23, + "execution_count": 29, "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ - "Loaded 1 lines from parallel_trace.jsonl\n", - "Found 1 top-level spans, 5 rail spans\n" + "Loaded 2 lines from parallel_trace.jsonl\n", + "Found 2 traces\n", + "Found 2 top-level spans, 7 rail spans\n" ] }, { @@ -879,302 +981,123 @@ " \n", " \n", " \n", + " trace_num\n", + " rail_name_short\n", " name\n", + " is_safe\n", " duration\n", " \n", " \n", " \n", " \n", " 0\n", + " 0\n", " None\n", - " 8.248329\n", + " guardrails.request\n", + " True\n", + " 2.917370\n", " \n", " \n", " 1\n", - " content safety check input $model=content_safety\n", - " 0.456112\n", + " 0\n", + " content safety check input\n", + " guardrails.rail\n", + " True\n", + " 0.421178\n", " \n", " \n", - " 4\n", - " topic safety check input $model=topic_control\n", - " 0.359808\n", + " 2\n", + " 0\n", + " topic safety check input\n", + " guardrails.rail\n", + " True\n", + " 0.338333\n", " \n", " \n", - " 7\n", + " 3\n", + " 0\n", " jailbreak detection model\n", - " 0.330025\n", + " guardrails.rail\n", + " True\n", + " 0.284210\n", " \n", " \n", - " 9\n", + " 4\n", + " 0\n", " generate user intent\n", - " 7.212214\n", - " \n", - " \n", - " 12\n", - " content safety check output $model=content_safety\n", - " 0.577307\n", - " \n", - " \n", - "\n", - "" - ], - "text/plain": [ - " name duration\n", - "0 None 8.248329\n", - "1 content safety check input $model=content_safety 0.456112\n", - "4 topic safety check input $model=topic_control 0.359808\n", - "7 jailbreak detection model 0.330025\n", - "9 generate user intent 7.212214\n", - "12 content safety check output $model=content_safety 0.577307" - ] - }, - "execution_count": 23, - "metadata": {}, - "output_type": "execute_result" - } - ], - "source": [ - "raw_parallel_df = load_trace_data(PARALLEL_TRACE_FILE)\n", - "parallel_df = clean_trace_dataframe(raw_parallel_df)\n", - "parallel_df[[\"name\", \"duration\"]]" - ] - }, - { - "cell_type": "code", - "execution_count": 24, - "metadata": {}, - "outputs": [ - { - "data": { - "text/html": [ - "
\n", - "\n", - "\n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", + " \n", " \n", - " \n", - " \n", + " \n", " \n", " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", + " \n", + " \n", + " \n", + " \n", " \n", - " \n", - " \n", - " \n", + " \n", " \n", " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", + " \n", + " \n", + " \n", + " \n", " \n", - " \n", - " \n", + " \n", " \n", " \n", " \n", + " \n", " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", + " \n", " \n", - " \n", - " \n", + " \n", " \n", " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", - " \n", + " \n", + " \n", + " \n", + " \n", " \n", - " \n", - " \n", + " \n", " \n", " \n", "
namespan_idparent_idstart_timeend_timedurationspan_typespan_kindattributeseventstrace_idepoch_secondsis_railis_top_spanstart_dtend_dt
0Nonebebb78c1-8788-4f43-96cb-161f9b24077aNone0.0000008.2483298.248329InteractionSpanserver{'span.kind': 'server', 'gen_ai.operation.name...[{'name': 'guardrails.user_message', 'timestam...861c9588-daf4-4006-b8ce-48809ec682f41756226969Falseguardrails.railTrue2025-08-26 16:49:29.0000000002025-08-26 16:49:37.2483289241.977735
1content safety check input $model=content_safety97a3d33c-074e-4e95-9fb5-551d5bf2ef4cbebb78c1-8788-4f43-96cb-161f9b24077a0.0000000.4561120.456112RailSpaninternal{'span.kind': 'internal', 'rail.type': 'input'...NaN861c9588-daf4-4006-b8ce-48809ec682f4175622696950content safety check outputguardrails.railTrueFalse2025-08-26 16:49:29.0000000002025-08-26 16:49:29.4561119080.514885
4topic safety check input $model=topic_controlc5fc6e0b-19d5-4d3c-a300-4a1f90f5b2bebebb78c1-8788-4f43-96cb-161f9b24077a0.0000230.3598310.359808RailSpaninternal{'span.kind': 'internal', 'rail.type': 'input'...NaN861c9588-daf4-4006-b8ce-48809ec682f41756226969True61Noneguardrails.requestFalse2025-08-26 16:49:29.0000231272025-08-26 16:49:29.3598310950.329526
71jailbreak detection modelb206d6c5-fa4a-48dd-a0c9-22bba163759fbebb78c1-8788-4f43-96cb-161f9b24077a0.0000360.3300610.330025RailSpaninternal{'span.kind': 'internal', 'rail.type': 'input'...NaN861c9588-daf4-4006-b8ce-48809ec682f41756226969Trueguardrails.railFalse2025-08-26 16:49:29.0000357632025-08-26 16:49:29.3300609590.302264
9generate user intentab6d251e-f919-4e5b-b645-d1a5a025dcf1bebb78c1-8788-4f43-96cb-161f9b24077a0.4588087.6710227.212214RailSpaninternal{'span.kind': 'internal', 'rail.type': 'genera...NaN861c9588-daf4-4006-b8ce-48809ec682f41756226969TrueFalse2025-08-26 16:49:29.4588081842025-08-26 16:49:36.671022177
12content safety check output $model=content_safety047b45d9-43d6-4a97-b8c2-764a8d36a7f5bebb78c1-8788-4f43-96cb-161f9b24077a7.6710228.2483290.577307RailSpaninternal{'span.kind': 'internal', 'rail.type': 'output...NaN861c9588-daf4-4006-b8ce-48809ec682f41756226969True81topic safety check inputguardrails.railFalse2025-08-26 16:49:36.6710221772025-08-26 16:49:37.2483289240.000013
\n", "
" ], "text/plain": [ - " name \\\n", - "0 None \n", - "1 content safety check input $model=content_safety \n", - "4 topic safety check input $model=topic_control \n", - "7 jailbreak detection model \n", - "9 generate user intent \n", - "12 content safety check output $model=content_safety \n", + " trace_num rail_name_short name is_safe \\\n", + "0 0 None guardrails.request True \n", + "1 0 content safety check input guardrails.rail True \n", + "2 0 topic safety check input guardrails.rail True \n", + "3 0 jailbreak detection model guardrails.rail True \n", + "4 0 generate user intent guardrails.rail True \n", + "5 0 content safety check output guardrails.rail True \n", + "6 1 None guardrails.request False \n", + "7 1 jailbreak detection model guardrails.rail False \n", + "8 1 topic safety check input guardrails.rail False \n", "\n", - " span_id \\\n", - "0 bebb78c1-8788-4f43-96cb-161f9b24077a \n", - "1 97a3d33c-074e-4e95-9fb5-551d5bf2ef4c \n", - "4 c5fc6e0b-19d5-4d3c-a300-4a1f90f5b2be \n", - "7 b206d6c5-fa4a-48dd-a0c9-22bba163759f \n", - "9 ab6d251e-f919-4e5b-b645-d1a5a025dcf1 \n", - "12 047b45d9-43d6-4a97-b8c2-764a8d36a7f5 \n", - "\n", - " parent_id start_time end_time duration \\\n", - "0 None 0.000000 8.248329 8.248329 \n", - "1 bebb78c1-8788-4f43-96cb-161f9b24077a 0.000000 0.456112 0.456112 \n", - "4 bebb78c1-8788-4f43-96cb-161f9b24077a 0.000023 0.359831 0.359808 \n", - "7 bebb78c1-8788-4f43-96cb-161f9b24077a 0.000036 0.330061 0.330025 \n", - "9 bebb78c1-8788-4f43-96cb-161f9b24077a 0.458808 7.671022 7.212214 \n", - "12 bebb78c1-8788-4f43-96cb-161f9b24077a 7.671022 8.248329 0.577307 \n", - "\n", - " span_type span_kind \\\n", - "0 InteractionSpan server \n", - "1 RailSpan internal \n", - "4 RailSpan internal \n", - "7 RailSpan internal \n", - "9 RailSpan internal \n", - "12 RailSpan internal \n", - "\n", - " attributes \\\n", - "0 {'span.kind': 'server', 'gen_ai.operation.name... \n", - "1 {'span.kind': 'internal', 'rail.type': 'input'... \n", - "4 {'span.kind': 'internal', 'rail.type': 'input'... \n", - "7 {'span.kind': 'internal', 'rail.type': 'input'... \n", - "9 {'span.kind': 'internal', 'rail.type': 'genera... \n", - "12 {'span.kind': 'internal', 'rail.type': 'output... \n", - "\n", - " events \\\n", - "0 [{'name': 'guardrails.user_message', 'timestam... \n", - "1 NaN \n", - "4 NaN \n", - "7 NaN \n", - "9 NaN \n", - "12 NaN \n", - "\n", - " trace_id epoch_seconds is_rail is_top_span \\\n", - "0 861c9588-daf4-4006-b8ce-48809ec682f4 1756226969 False True \n", - "1 861c9588-daf4-4006-b8ce-48809ec682f4 1756226969 True False \n", - "4 861c9588-daf4-4006-b8ce-48809ec682f4 1756226969 True False \n", - "7 861c9588-daf4-4006-b8ce-48809ec682f4 1756226969 True False \n", - "9 861c9588-daf4-4006-b8ce-48809ec682f4 1756226969 True False \n", - "12 861c9588-daf4-4006-b8ce-48809ec682f4 1756226969 True False \n", - "\n", - " start_dt end_dt \n", - "0 2025-08-26 16:49:29.000000000 2025-08-26 16:49:37.248328924 \n", - "1 2025-08-26 16:49:29.000000000 2025-08-26 16:49:29.456111908 \n", - "4 2025-08-26 16:49:29.000023127 2025-08-26 16:49:29.359831095 \n", - "7 2025-08-26 16:49:29.000035763 2025-08-26 16:49:29.330060959 \n", - "9 2025-08-26 16:49:29.458808184 2025-08-26 16:49:36.671022177 \n", - "12 2025-08-26 16:49:36.671022177 2025-08-26 16:49:37.248328924 " + " duration \n", + "0 2.917370 \n", + "1 0.421178 \n", + "2 0.338333 \n", + "3 0.284210 \n", + "4 1.977735 \n", + "5 0.514885 \n", + "6 0.329526 \n", + "7 0.302264 \n", + "8 0.000013 " ] }, - "execution_count": 24, + "execution_count": 29, "metadata": {}, "output_type": "execute_result" } ], "source": [ - "parallel_df" + "raw_parallel_df = load_trace_data(PARALLEL_TRACE_FILE)\n", + "parallel_df = clean_trace_dataframe(raw_parallel_df)\n", + "parallel_df[[\"trace_num\", \"rail_name_short\", \"name\", \"is_safe\", \"duration\"]]" ] }, { @@ -1185,15 +1108,15 @@ "\n", "The DataFrame below shows the time (in seconds) for the top-level end-to-end interaction, and each of the rails that are called during the interaction. These all run sequentially in this configuration. All input rails have to pass before the user query is passed to the LLM. \n", "\n", - "In the DataFrame below, the top-level span is named `interaction`, and represents the end-to-end server-side duration of the `generate_async()` call above. This top-level span comprises 5 rail actions, which are:\n", + "In the DataFrame below, the top-level span is labelled with the `is_top_span` boolean, and represents the end-to-end server-side duration of the `generate_async()` call. Each top-level span for a safe request comprises 5 rail actions, which are:\n", "\n", - " * `rail: content safety check input $model=content_safety'` : Time to check the user input by the [Content-safety Nemoguard NIM](https://build.nvidia.com/nvidia/llama-3_1-nemoguard-8b-content-safety).\n", - " * `rail: topic safety check input $model=topic_control'` : Time to check user input by the [Topic-Control Nemoguard NIM](https://build.nvidia.com/nvidia/llama-3_1-nemoguard-8b-topic-control).\n", - " * `rail: jailbreak detection model'` : Time to check the user input by the [Jailbreak Nemoguard NIM](https://build.nvidia.com/nvidia/nemoguard-jailbreak-detect).\n", - " * `rail: generate user intent'` : Time to generate a response to the user's question from the Main LLM ([Llama 3.3 70B Instruct](https://build.nvidia.com/meta/llama-3_3-70b-instruct)).\n", - " * `rail: content safety check output $model=content_safety` : Time to check the user input and LLM response by the [Content-safety Nemoguard NIM](https://build.nvidia.com/nvidia/llama-3_1-nemoguard-8b-content-safety).\n", + " * `content safety check input` : Time to check the user input by the [Content-safety Nemoguard NIM](https://build.nvidia.com/nvidia/llama-3_1-nemoguard-8b-content-safety).\n", + " * `topic safety check input` : Time to check user input by the [Topic-Control Nemoguard NIM](https://build.nvidia.com/nvidia/llama-3_1-nemoguard-8b-topic-control).\n", + " * `jailbreak detection model` : Time to check the user input by the [Jailbreak Nemoguard NIM](https://build.nvidia.com/nvidia/nemoguard-jailbreak-detect).\n", + " * `generate user intent` : Time to generate a response to the user's question from the Main LLM ([Llama 3.1 8B Instruct](https://build.nvidia.com/meta/llama-3_1-8b-instruct)).\n", + " * `content safety check output` : Time to check the user input and LLM response by the [Content-safety Nemoguard NIM](https://build.nvidia.com/nvidia/llama-3_1-nemoguard-8b-content-safety).\n", "\n", - "The durations should be roughly in the 400ms - 600ms range, depending on user traffic. The Llama 3.3 70B Instruct model that generates the response is an order of magnitude larger than the NemoGuard models, so it may take up to a minute to generate a response, depending on the cluster load." + "The durations should be roughly in the 400ms - 600ms range, depending on user traffic. The Llama 3.1 8B Instruct model that generates the response is an order of magnitude larger than the NemoGuard models, so it may take up to a minute to generate a response, depending on the cluster load." ] }, { @@ -1207,7 +1130,17 @@ }, { "cell_type": "code", - "execution_count": 25, + "execution_count": 30, + "metadata": {}, + "outputs": [], + "source": [ + "PLOT_WIDTH = 800\n", + "PLOT_HEIGHT = 400" + ] + }, + { + "cell_type": "code", + "execution_count": 31, "metadata": {}, "outputs": [ { @@ -1233,7 +1166,7 @@ " \n", " is_rail\n", " is_top_span\n", - " name\n", + " rail_name_short\n", " duration\n", " \n", " \n", @@ -1243,77 +1176,85 @@ " False\n", " True\n", " None\n", - " 7.403602\n", + " 3.810076\n", " \n", " \n", " 1\n", " True\n", " False\n", - " content safety check input $model=content_safety\n", - " 0.450512\n", + " content safety check input\n", + " 0.403598\n", " \n", " \n", - " 4\n", + " 2\n", " True\n", " False\n", - " topic safety check input $model=topic_control\n", - " 0.360603\n", + " topic safety check input\n", + " 0.324701\n", " \n", " \n", - " 7\n", + " 3\n", " True\n", " False\n", " jailbreak detection model\n", - " 0.336845\n", + " 0.300511\n", " \n", " \n", - " 9\n", + " 4\n", " True\n", " False\n", " generate user intent\n", - " 5.679443\n", + " 2.236309\n", " \n", " \n", - " 12\n", + " 5\n", " True\n", " False\n", - " content safety check output $model=content_safety\n", - " 0.564421\n", + " content safety check output\n", + " 0.532284\n", + " \n", + " \n", + " 6\n", + " False\n", + " True\n", + " None\n", + " 0.610056\n", + " \n", + " \n", + " 7\n", + " True\n", + " False\n", + " content safety check input\n", + " 0.610056\n", " \n", " \n", "\n", "" ], "text/plain": [ - " is_rail is_top_span name \\\n", - "0 False True None \n", - "1 True False content safety check input $model=content_safety \n", - "4 True False topic safety check input $model=topic_control \n", - "7 True False jailbreak detection model \n", - "9 True False generate user intent \n", - "12 True False content safety check output $model=content_safety \n", - "\n", - " duration \n", - "0 7.403602 \n", - "1 0.450512 \n", - "4 0.360603 \n", - "7 0.336845 \n", - "9 5.679443 \n", - "12 0.564421 " + " is_rail is_top_span rail_name_short duration\n", + "0 False True None 3.810076\n", + "1 True False content safety check input 0.403598\n", + "2 True False topic safety check input 0.324701\n", + "3 True False jailbreak detection model 0.300511\n", + "4 True False generate user intent 2.236309\n", + "5 True False content safety check output 0.532284\n", + "6 False True None 0.610056\n", + "7 True False content safety check input 0.610056" ] }, - "execution_count": 25, + "execution_count": 31, "metadata": {}, "output_type": "execute_result" } ], "source": [ - "sequential_df[[\"is_rail\", \"is_top_span\", \"name\", \"duration\"]]" + "sequential_df[[\"is_rail\", \"is_top_span\", \"rail_name_short\", \"duration\"]]" ] }, { "cell_type": "code", - "execution_count": 26, + "execution_count": 32, "metadata": {}, "outputs": [ { @@ -1339,14 +1280,14 @@ "type": "bar", "x": [ "generate user intent", - "content safety check output $model=content_safety", - "content safety check input $model=content_safety", - "topic safety check input $model=topic_control", + "content safety check output", + "content safety check input", + "topic safety check input", "jailbreak detection model" ], "xaxis": "x", "y": { - "bdata": "AAAA4L+3FkAAAAAAvQ/iPwAAAAAx1dw/AAAAAB8U1z8AAAAA347VPw==", + "bdata": "AAAAAPbjAUAAAACAeAjhPwAAAACM1Nk/AAAAAObH1D8AAAAAkjvTPw==", "dtype": "f8" }, "yaxis": "y" @@ -2135,7 +2076,7 @@ } }, "title": { - "text": "Sequential Guardrails Rail durations" + "text": "Sequential Guardrails Rail durations (safe request)" }, "width": 800, "xaxis": { @@ -2159,8 +2100,7 @@ } } } - }, - "image/png": "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" + } }, "metadata": {}, "output_type": "display_data" @@ -2169,13 +2109,15 @@ "source": [ "# Now let's plot a bar-graph of these numbers\n", "px.bar(\n", - " sequential_df[sequential_df[\"is_rail\"]].sort_values(\"duration\", ascending=False),\n", - " x=\"name\",\n", + " sequential_df[sequential_df[\"is_safe\"] & sequential_df[\"is_rail\"]].sort_values(\n", + " \"duration\", ascending=False\n", + " ),\n", + " x=\"rail_name_short\",\n", " y=\"duration\",\n", - " title=\"Sequential Guardrails Rail durations\",\n", - " labels={\"name\": \"Rail Name\", \"duration\": \"Duration (seconds)\"},\n", - " width=800,\n", - " height=800,\n", + " title=\"Sequential Guardrails Rail durations (safe request)\",\n", + " labels={\"rail_name_short\": \"Rail Name\", \"duration\": \"Duration (seconds)\"},\n", + " width=PLOT_WIDTH,\n", + " height=PLOT_HEIGHT * 2,\n", ")" ] }, @@ -2188,7 +2130,7 @@ }, { "cell_type": "code", - "execution_count": 27, + "execution_count": 33, "metadata": {}, "outputs": [ { @@ -2200,11 +2142,11 @@ "data": [ { "base": [ - "2025-08-26T16:49:20.000000000", - "2025-08-26T16:49:20.452291965", - "2025-08-26T16:49:20.814581871", - "2025-08-26T16:49:21.159738064", - "2025-08-26T16:49:26.839180946" + "2025-09-05T14:42:28.000000000", + "2025-09-05T14:42:28.404770136", + "2025-09-05T14:42:28.731706858", + "2025-09-05T14:42:29.041482925", + "2025-09-05T14:42:31.277791977" ], "hovertemplate": "start_dt=%{base}
end_dt=%{x}
Rail Name=%{y}", "legendgroup": "", @@ -2220,22 +2162,23 @@ "textposition": "auto", "type": "bar", "x": { - "bdata": "wgFoAVABLxY0Ag==", + "bdata": "kwFEASwBvAgUAg==", "dtype": "i2" }, "xaxis": "x", "y": [ - "content safety check input $model=content_safety", - "topic safety check input $model=topic_control", + "content safety check input", + "topic safety check input", "jailbreak detection model", "generate user intent", - "content safety check output $model=content_safety" + "content safety check output" ], "yaxis": "y" } ], "layout": { "barmode": "overlay", + "height": 400, "legend": { "tracegroupgap": 0 }, @@ -3016,8 +2959,9 @@ } }, "title": { - "text": "Gantt chart of rails calls in sequential mode" + "text": "Gantt chart of rails calls in sequential mode (safe request)" }, + "width": 800, "xaxis": { "anchor": "y", "domain": [ @@ -3038,8 +2982,7 @@ } } } - }, - "image/png": "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" + } }, "metadata": {}, "output_type": "display_data" @@ -3049,12 +2992,14 @@ "# Let's plot a Gantt chart, to show the sequence of when the rails execute\n", "\n", "fig = px.timeline(\n", - " sequential_df.loc[sequential_df[\"is_rail\"]],\n", + " sequential_df.loc[sequential_df[\"is_safe\"] & sequential_df[\"is_rail\"]],\n", " x_start=\"start_dt\",\n", " x_end=\"end_dt\",\n", - " y=\"name\",\n", - " title=\"Gantt chart of rails calls in sequential mode\",\n", - " labels={\"name\": \"Rail Name\"},\n", + " y=\"rail_name_short\",\n", + " title=\"Gantt chart of rails calls in sequential mode (safe request)\",\n", + " labels={\"rail_name_short\": \"Rail Name\"},\n", + " width=PLOT_WIDTH,\n", + " height=PLOT_HEIGHT,\n", ")\n", "fig.update_yaxes(autorange=\"reversed\")\n", "fig.show()" @@ -3071,7 +3016,7 @@ }, { "cell_type": "code", - "execution_count": 28, + "execution_count": 34, "metadata": {}, "outputs": [ { @@ -3097,14 +3042,14 @@ "type": "bar", "x": [ "generate user intent", - "content safety check output $model=content_safety", - "content safety check input $model=content_safety", - "topic safety check input $model=topic_control", + "content safety check output", + "content safety check input", + "topic safety check input", "jailbreak detection model" ], "xaxis": "x", "y": { - "bdata": "AAAAoE7ZHEAAAAAATHniPwAAAADwMN0/AAAAABgH1z8AAAAAIh/VPw==", + "bdata": "AAAAwM2k/z8AAACA73ngPwAAAACU9No/AAAAAEGn1T8AAAAAgDDSPw==", "dtype": "f8" }, "yaxis": "y" @@ -3112,7 +3057,7 @@ ], "layout": { "barmode": "relative", - "height": 600, + "height": 800, "legend": { "tracegroupgap": 0 }, @@ -3893,7 +3838,7 @@ } }, "title": { - "text": "Sequential Guardrails Rail durations" + "text": "Parallel Guardrails Rail durations (safe request)" }, "width": 800, "xaxis": { @@ -3917,8 +3862,7 @@ } } } - }, - "image/png": "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" + } }, "metadata": {}, "output_type": "display_data" @@ -3927,13 +3871,15 @@ "source": [ "# Now let's plot a bar-graph of these numbers\n", "px.bar(\n", - " parallel_df[parallel_df[\"is_rail\"]].sort_values(\"duration\", ascending=False),\n", - " x=\"name\",\n", + " parallel_df[parallel_df[\"is_safe\"] & parallel_df[\"is_rail\"]].sort_values(\n", + " \"duration\", ascending=False\n", + " ),\n", + " x=\"rail_name_short\",\n", " y=\"duration\",\n", - " title=\"Sequential Guardrails Rail durations\",\n", - " labels={\"name\": \"Rail Name\", \"duration\": \"Duration (seconds)\"},\n", - " width=800,\n", - " height=600,\n", + " title=\"Parallel Guardrails Rail durations (safe request)\",\n", + " labels={\"rail_name_short\": \"Rail Name\", \"duration\": \"Duration (seconds)\"},\n", + " width=PLOT_WIDTH,\n", + " height=PLOT_HEIGHT * 2,\n", ")" ] }, @@ -3946,7 +3892,7 @@ }, { "cell_type": "code", - "execution_count": 29, + "execution_count": 35, "metadata": {}, "outputs": [ { @@ -3958,11 +3904,11 @@ "data": [ { "base": [ - "2025-08-26T16:49:29.000000000", - "2025-08-26T16:49:29.000023127", - "2025-08-26T16:49:29.000035763", - "2025-08-26T16:49:29.458808184", - "2025-08-26T16:49:36.671022177" + "2025-09-05T14:42:31.000000000", + "2025-09-05T14:42:31.000024796", + "2025-09-05T14:42:31.000034809", + "2025-09-05T14:42:31.424749851", + "2025-09-05T14:42:33.402485132" ], "hovertemplate": "start_dt=%{base}
end_dt=%{x}
Rail Name=%{y}", "legendgroup": "", @@ -3978,16 +3924,16 @@ "textposition": "auto", "type": "bar", "x": { - "bdata": "yAFnAUoBLBxBAg==", + "bdata": "pQFSARwBuQcCAg==", "dtype": "i2" }, "xaxis": "x", "y": [ - "content safety check input $model=content_safety", - "topic safety check input $model=topic_control", + "content safety check input", + "topic safety check input", "jailbreak detection model", "generate user intent", - "content safety check output $model=content_safety" + "content safety check output" ], "yaxis": "y" } @@ -4775,9 +4721,9 @@ } }, "title": { - "text": "Gantt chart of rails calls in parallel mode" + "text": "Gantt chart of rails calls in parallel mode (safe request)" }, - "width": 1000, + "width": 800, "xaxis": { "anchor": "y", "domain": [ @@ -4798,7 +4744,8 @@ } } } - } + }, + "image/png": "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" }, "metadata": {}, "output_type": "display_data" @@ -4808,14 +4755,14 @@ "# Let's plot a Gantt chart, to show the sequence of when the rails execute\n", "\n", "fig = px.timeline(\n", - " parallel_df.loc[parallel_df[\"is_rail\"]],\n", + " parallel_df.loc[parallel_df[\"is_safe\"] & parallel_df[\"is_rail\"]],\n", " x_start=\"start_dt\",\n", " x_end=\"end_dt\",\n", - " y=\"name\",\n", - " title=\"Gantt chart of rails calls in parallel mode\",\n", - " labels={\"name\": \"Rail Name\"},\n", - " height=400,\n", - " width=1000,\n", + " y=\"rail_name_short\",\n", + " title=\"Gantt chart of rails calls in parallel mode (safe request)\",\n", + " labels={\"rail_name_short\": \"Rail Name\"},\n", + " width=PLOT_WIDTH,\n", + " height=PLOT_HEIGHT,\n", ")\n", "fig.update_yaxes(autorange=\"reversed\")\n", "fig.show()" @@ -4827,32 +4774,32 @@ "source": [ "### Compare Sequential and Parallel Trace Data\n", "\n", - "The following cells compare the input rail times for the sequential and parallel configurations." + "The following cells compare the input rail times for the sequential and parallel configurations. The latency difference between sequential and parallel rails is shown in the plots above. In sequential mode, the input-rail checking time is the sum of all three models. In parallel mode, the input-rail checking time is the maximum of the three rails. Let's quantify the time-saving below." ] }, { "cell_type": "code", - "execution_count": 30, + "execution_count": 36, "metadata": {}, "outputs": [], "source": [ "INPUT_RAIL_NAMES = {\n", - " \"content safety check input $model=content_safety\",\n", - " \"topic safety check input $model=topic_control\",\n", + " \"content safety check input\",\n", + " \"topic safety check input\",\n", " \"jailbreak detection model\",\n", "}" ] }, { "cell_type": "code", - "execution_count": 31, + "execution_count": 37, "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ - "Sequential input rail time: 1.1480s\n" + "Sequential input rail time: 1.0288s\n" ] } ], @@ -4861,29 +4808,32 @@ "\n", "# Sum the sequential rail run-times\n", "sequential_input_rail_time = sequential_df.loc[\n", - " sequential_df[\"name\"].isin(INPUT_RAIL_NAMES), \"duration\"\n", + " sequential_df[\"is_safe\"] # Use the safe user-request\n", + " & sequential_df[\"rail_name_short\"].isin(INPUT_RAIL_NAMES),\n", + " \"duration\", # Use input-rails only\n", "].sum()\n", "print(f\"Sequential input rail time: {sequential_input_rail_time:.4f}s\")" ] }, { "cell_type": "code", - "execution_count": 32, + "execution_count": 38, "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ - "Parallel input rail time: 0.4561s\n", - "Parallel input speedup: 2.5168 times\n" + "Parallel input rail time: 0.4212s\n", + "Parallel input speedup: 2.4427 times\n" ] } ], "source": [ "# Final summary of the time-saving due to parallel rails\n", "parallel_input_rail_time = parallel_df.loc[\n", - " parallel_df[\"name\"].isin(INPUT_RAIL_NAMES), \"duration\"\n", + " parallel_df[\"is_safe\"] & parallel_df[\"rail_name_short\"].isin(INPUT_RAIL_NAMES),\n", + " \"duration\",\n", "].max()\n", "print(f\"Parallel input rail time: {parallel_input_rail_time:.4f}s\")\n", "print(\n", @@ -4891,6 +4841,45 @@ ")" ] }, + { + "cell_type": "code", + "execution_count": 39, + "metadata": {}, + "outputs": [], + "source": [ + "# Check the difference in overall time\n", + "total_sequential_time_s = sequential_df.loc[\n", + " sequential_df[\"is_safe\"] & sequential_df[\"is_rail\"], \"duration\"\n", + "].sum()\n", + "total_parallel_time_s = parallel_df.loc[\n", + " parallel_df[\"is_safe\"] & parallel_df[\"is_rail\"], \"duration\"\n", + "].sum()\n", + "\n", + "parallel_time_saved_s = total_sequential_time_s - total_parallel_time_s\n", + "parallel_time_saved_pct = (100.0 * parallel_time_saved_s) / total_sequential_time_s" + ] + }, + { + "cell_type": "code", + "execution_count": 40, + "metadata": {}, + "outputs": [ + { + "name": "stdout", + "output_type": "stream", + "text": [ + "Total sequential time: 3.80s\n", + "Total parallel time: 3.54s\n", + "Time saving: 0.26s, (6.87%)\n" + ] + } + ], + "source": [ + "print(f\"Total sequential time: {total_sequential_time_s:.2f}s\")\n", + "print(f\"Total parallel time: {total_parallel_time_s:.2f}s\")\n", + "print(f\"Time saving: {parallel_time_saved_s:.2f}s, ({parallel_time_saved_pct:.2f}%)\")" + ] + }, { "cell_type": "markdown", "metadata": {}, @@ -4899,7 +4888,9 @@ "\n", "# Conclusions\n", "\n", - "In this notebook, you learned how to trace Guardrails requests in both **sequential** and **parallel** modes. By sending a single request for each mode, you were able to trace and compare their latencies. Using the graphing tools, you visualized the latency breakdown into a table, bar chart, and Gantt chart, providing a clear visual comparison of how each mode performed. The Gantt charts for parallel and sequential rails clearly show the benefit of running all three in parallel, rather than sequentially. For the sample configuration and input request run in this notebook snapshot, parallel mode was ~2.5x faster." + "In this notebook, you learned how to trace Guardrails requests in both **sequential** and **parallel** modes. By sending a single request for each mode, you were able to trace and compare their latencies. Using the graphing tools, you visualized the latency breakdown into a table, bar chart, and Gantt chart, providing a clear visual comparison of how each mode performed. The Gantt charts for parallel and sequential rails clearly show the benefit of running all three in parallel, rather than sequentially. \n", + "\n", + "For the sample configuration and input request run in this notebook snapshot, running the input rails in parallel mode was ~2.44x faster, reducing overall latency by 6.86% for this example. " ] } ], diff --git a/docs/getting-started/8-tracing/2_tracing_with_jaeger.ipynb b/docs/getting-started/8-tracing/2_tracing_with_jaeger.ipynb index 1495ab539..0011cc89b 100644 --- a/docs/getting-started/8-tracing/2_tracing_with_jaeger.ipynb +++ b/docs/getting-started/8-tracing/2_tracing_with_jaeger.ipynb @@ -60,7 +60,7 @@ " jaegertracing/all-in-one:1.62.0\n", "```\n", "\n", - "You'll see that the container prints debug messages that end with the following lines. This indicates the Jaeger server is up and ready to accept requests.\n", + "You'll see that the container prints debug messages that end with the following lines. This indicates the Jaeger server is up and ready to accept requests. These can be sent over either gRPC or REST on the corresponding ports listed below.\n", "\n", "```bash\n", "{\"level\":\"info\",\"ts\":1756236324.295533,\"caller\":\"healthcheck/handler.go:118\",\"msg\":\"Health Check state change\",\"status\":\"ready\"}\n", @@ -190,7 +190,7 @@ "metadata": {}, "outputs": [], "source": [ - "CONFIG_MODELS: Dict[str, str] = [\n", + "CONFIG_MODELS: List[Dict[str, str]] = [\n", " {\n", " \"type\": \"main\",\n", " \"engine\": \"nim\",\n", @@ -256,7 +256,7 @@ "source": [ "### Tracing\n", "\n", - "The tracing configuration configures the adapter and any adapter-specific controls. Here we're storing traces in JSONL format. We'll use a different filename depending on whether we have a sequential or parallel workflow." + "The tracing configuration configures the adapter and any adapter-specific controls. Here we're sending metrics over opentelemetry for visualization by another tool." ] }, { @@ -423,9 +423,9 @@ "tracer_provider = TracerProvider(resource=resource)\n", "trace.set_tracer_provider(tracer_provider)\n", "\n", - "# Export traces to the port location matching \n", + "# Export traces to the port location matching\n", "otlp_exporter = OTLPSpanExporter(endpoint=\"http://localhost:4317\", insecure=True)\n", - "tracer_provider.add_span_processor(BatchSpanProcessor(otlp_exporter))\n" + "tracer_provider.add_span_processor(BatchSpanProcessor(otlp_exporter))" ] }, {