Language Mannequin (LLM) isn’t essentially the ultimate step in productionizing your Generative AI utility. An typically forgotten, but essential a part of the MLOPs lifecycle is correctly load testing your LLM and guaranteeing it is able to stand up to your anticipated manufacturing site visitors. Load testing at a excessive stage is the apply of testing your utility or on this case your mannequin with the site visitors it might expect in a manufacturing surroundings to make sure that it’s performant.
Prior to now we’ve mentioned load testing conventional ML fashions utilizing open supply Python instruments reminiscent of Locust. Locust helps seize basic efficiency metrics reminiscent of requests per second (RPS) and latency percentiles on a per request foundation. Whereas that is efficient with extra conventional APIs and ML fashions it doesn’t seize the complete story for LLMs.
LLMs historically have a a lot decrease RPS and better latency than conventional ML fashions resulting from their measurement and bigger compute necessities. Typically the RPS metric does not likely present essentially the most correct image both as requests can significantly range relying on the enter to the LLM. As an example you might need a question asking to summarize a big chunk of textual content and one other question which may require a one-word response.
Because of this tokens are seen as a way more correct illustration of an LLM’s efficiency. At a excessive stage a token is a bit of textual content, every time an LLM is processing your enter it “tokenizes” the enter. A token differs relying particularly on the LLM you might be utilizing, however you may think about it for example as a phrase, sequence of phrases, or characters in essence.
What we’ll do on this article is discover how we will generate token primarily based metrics so we will perceive how your LLM is acting from a serving/deployment perspective. After this text you’ll have an concept of how one can arrange a load-testing software particularly to benchmark totally different LLMs within the case that you’re evaluating many fashions or totally different deployment configurations or a mixture of each.
Let’s get arms on! In case you are extra of a video primarily based learner be at liberty to comply with my corresponding YouTube video down under:
NOTE: This text assumes a fundamental understanding of Python, LLMs, and Amazon Bedrock/SageMaker. In case you are new to Amazon Bedrock please check with my starter information right here. If you wish to study extra about SageMaker JumpStart LLM deployments check with the video right here.
DISCLAIMER: I’m a Machine Studying Architect at AWS and my opinions are my very own.
Desk of Contents
- LLM Particular Metrics
- LLMPerf Intro
- Making use of LLMPerf to Amazon Bedrock
- Further Sources & Conclusion
LLM-Particular Metrics
As we briefly mentioned within the introduction with reference to LLM internet hosting, token primarily based metrics typically present a significantly better illustration of how your LLM is responding to totally different payload sizes or forms of queries (summarization vs QnA).
Historically we’ve at all times tracked RPS and latency which we’ll nonetheless see right here nonetheless, however extra so at a token stage. Listed here are among the metrics to concentrate on earlier than we get began with load testing:
- Time to First Token: That is the length it takes for the primary token to generate. That is particularly helpful when streaming. As an example when utilizing ChatGPT we begin processing data when the primary piece of textual content (token) seems.
- Complete Output Tokens Per Second: That is the entire variety of tokens generated per second, you may consider this as a extra granular various to the requests per second we historically observe.
These are the key metrics that we’ll give attention to, and there’s a couple of others reminiscent of inter-token latency that may even be displayed as a part of the load assessments. Remember the parameters that additionally affect these metrics embody the anticipated enter and output token measurement. We particularly play with these parameters to get an correct understanding of how our LLM performs in response to totally different technology duties.
Now let’s check out a software that allows us to toggle these parameters and show the related metrics we’d like.
LLMPerf Intro
LLMPerf is constructed on prime of Ray, a preferred distributed computing Python framework. LLMPerf particularly leverages Ray to create distributed load assessments the place we will simulate real-time manufacturing stage site visitors.
Be aware that any load-testing software can be solely going to have the ability to generate your anticipated quantity of site visitors if the consumer machine it’s on has sufficient compute energy to match your anticipated load. As an example as you scale the concurrency or throughput anticipated in your mannequin, you’d additionally need to scale the consumer machine(s) the place you might be working your load check.
Now particularly inside LLMPerf there’s a couple of parameters which might be uncovered which might be tailor-made for LLM load testing as we’ve mentioned:
- Mannequin: That is the mannequin supplier and your hosted mannequin that you simply’re working with. For our use-case it’ll be Amazon Bedrock and Claude 3 Sonnet particularly.
- LLM API: That is the API format wherein the payload must be structured. We use LiteLLM which supplies a standardized payload construction throughout totally different mannequin suppliers, thus simplifying the setup course of for us particularly if we need to check totally different fashions hosted on totally different platforms.
- Enter Tokens: The imply enter token size, you can even specify a normal deviation for this quantity.
- Output Tokens: The imply output token size, you can even specify a normal deviation for this quantity.
- Concurrent Requests: The variety of concurrent requests for the load check to simulate.
- Check Period: You’ll be able to management the length of the check, this parameter is enabled in seconds.
LLMPerf particularly exposes all these parameters via their token_benchmark_ray.py script which we configure with our particular values. Let’s have a look now at how we will configure this particularly for Amazon Bedrock.
Making use of LLMPerf to Amazon Bedrock
Setup
For this instance we’ll be working in a SageMaker Traditional Pocket book Occasion with a conda_python3 kernel and ml.g5.12xlarge occasion. Be aware that you simply need to choose an occasion that has sufficient compute to generate the site visitors load that you simply need to simulate. Make sure that you even have your AWS credentials for LLMPerf to entry the hosted mannequin be it on Bedrock or SageMaker.
LiteLLM Configuration
We first configure our LLM API construction of alternative which is LiteLLM on this case. With LiteLLM there’s assist throughout varied mannequin suppliers, on this case we configure the completion API to work with Amazon Bedrock:
import os
from litellm import completion
os.environ["AWS_ACCESS_KEY_ID"] = "Enter your entry key ID"
os.environ["AWS_SECRET_ACCESS_KEY"] = "Enter your secret entry key"
os.environ["AWS_REGION_NAME"] = "us-east-1"
response = completion(
mannequin="anthropic.claude-3-sonnet-20240229-v1:0",
messages=[{ "content": "Who is Roger Federer?","role": "user"}]
)
output = response.selections[0].message.content material
print(output)To work with Bedrock we configure the Mannequin ID to level in direction of Claude 3 Sonnet and go in our immediate. The neat half with LiteLLM is that messages key has a constant format throughout mannequin suppliers.
Publish-execution right here we will give attention to configuring LLMPerf for Bedrock particularly.
LLMPerf Bedrock Integration
To execute a load check with LLMPerf we will merely use the offered token_benchmark_ray.py script and go within the following parameters that we talked of earlier:
- Enter Tokens Imply & Commonplace Deviation
- Output Tokens Imply & Commonplace Deviation
- Max variety of requests for check
- Period of check
- Concurrent requests
On this case we additionally specify our API format to be LiteLLM and we will execute the load check with a easy shell script like the next:
%%sh
python llmperf/token_benchmark_ray.py
--model bedrock/anthropic.claude-3-sonnet-20240229-v1:0
--mean-input-tokens 1024
--stddev-input-tokens 200
--mean-output-tokens 1024
--stddev-output-tokens 200
--max-num-completed-requests 30
--num-concurrent-requests 1
--timeout 300
--llm-api litellm
--results-dir bedrock-outputsOn this case we hold the concurrency low, however be at liberty to toggle this quantity relying on what you’re anticipating in manufacturing. Our check will run for 300 seconds and publish length it’s best to see an output listing with two recordsdata representing statistics for every inference and likewise the imply metrics throughout all requests within the length of the check.
We are able to make this look somewhat neater by parsing the abstract file with pandas:
import json
from pathlib import Path
import pandas as pd
# Load JSON recordsdata
individual_path = Path("bedrock-outputs/bedrock-anthropic-claude-3-sonnet-20240229-v1-0_1024_1024_individual_responses.json")
summary_path = Path("bedrock-outputs/bedrock-anthropic-claude-3-sonnet-20240229-v1-0_1024_1024_summary.json")
with open(individual_path, "r") as f:
individual_data = json.load(f)
with open(summary_path, "r") as f:
summary_data = json.load(f)
# Print abstract metrics
df = pd.DataFrame(individual_data)
summary_metrics = {
"Mannequin": summary_data.get("mannequin"),
"Imply Enter Tokens": summary_data.get("mean_input_tokens"),
"Stddev Enter Tokens": summary_data.get("stddev_input_tokens"),
"Imply Output Tokens": summary_data.get("mean_output_tokens"),
"Stddev Output Tokens": summary_data.get("stddev_output_tokens"),
"Imply TTFT (s)": summary_data.get("results_ttft_s_mean"),
"Imply Inter-token Latency (s)": summary_data.get("results_inter_token_latency_s_mean"),
"Imply Output Throughput (tokens/s)": summary_data.get("results_mean_output_throughput_token_per_s"),
"Accomplished Requests": summary_data.get("results_num_completed_requests"),
"Error Charge": summary_data.get("results_error_rate")
}
print("Claude 3 Sonnet - Efficiency Abstract:n")
for okay, v in summary_metrics.objects():
print(f"{okay}: {v}")The ultimate load check outcomes will look one thing like the next:
As we will see we see the enter parameters that we configured, after which the corresponding outcomes with time to first token(s) and throughput with reference to imply output tokens per second.
In a real-world use case you may use LLMPerf throughout many alternative mannequin suppliers and run assessments throughout these platforms. With this software you should utilize it holistically to establish the proper mannequin and deployment stack in your use-case when used at scale.
Further Sources & Conclusion
The whole code for the pattern will be discovered at this related Github repository. In the event you additionally need to work with SageMaker endpoints you’ll find a Llama JumpStart deployment load testing pattern right here.
All in all load testing and analysis are each essential to making sure that your LLM is performant in opposition to your anticipated site visitors earlier than pushing to manufacturing. In future articles we’ll cowl not simply the analysis portion, however how we will create a holistic check with each parts.
As at all times thanks for studying and be at liberty to depart any suggestions and join with me on Linkedln and X.
