├── LICENSE ├── README.md └── aibench ├── __init__.py ├── requirements.txt └── runner.py /LICENSE: -------------------------------------------------------------------------------- 1 | Apache License 2 | Version 2.0, January 2004 3 | http://www.apache.org/licenses/ 4 | 5 | TERMS AND CONDITIONS FOR USE, REPRODUCTION, AND DISTRIBUTION 6 | 7 | 1. Definitions. 8 | 9 | "License" shall mean the terms and conditions for use, reproduction, 10 | and distribution as defined by Sections 1 through 9 of this document. 11 | 12 | "Licensor" shall mean the copyright owner or entity authorized by 13 | the copyright owner that is granting the License. 14 | 15 | "Legal Entity" shall mean the union of the acting entity and all 16 | other entities that control, are controlled by, or are under common 17 | control with that entity. For the purposes of this definition, 18 | "control" means (i) the power, direct or indirect, to cause the 19 | direction or management of such entity, whether by contract or 20 | otherwise, or (ii) ownership of fifty percent (50%) or more of the 21 | outstanding shares, or (iii) beneficial ownership of such entity. 22 | 23 | "You" (or "Your") shall mean an individual or Legal Entity 24 | exercising permissions granted by this License. 25 | 26 | "Source" form shall mean the preferred form for making modifications, 27 | including but not limited to software source code, documentation 28 | source, and configuration files. 29 | 30 | "Object" form shall mean any form resulting from mechanical 31 | transformation or translation of a Source form, including but 32 | not limited to compiled object code, generated documentation, 33 | and conversions to other media types. 34 | 35 | "Work" shall mean the work of authorship, whether in Source or 36 | Object form, made available under the License, as indicated by a 37 | copyright notice that is included in or attached to the work 38 | (an example is provided in the Appendix below). 39 | 40 | "Derivative Works" shall mean any work, whether in Source or Object 41 | form, that is based on (or derived from) the Work and for which the 42 | editorial revisions, annotations, elaborations, or other modifications 43 | represent, as a whole, an original work of authorship. For the purposes 44 | of this License, Derivative Works shall not include works that remain 45 | separable from, or merely link (or bind by name) to the interfaces of, 46 | the Work and Derivative Works thereof. 47 | 48 | "Contribution" shall mean any work of authorship, including 49 | the original version of the Work and any modifications or additions 50 | to that Work or Derivative Works thereof, that is intentionally 51 | submitted to Licensor for inclusion in the Work by the copyright owner 52 | or by an individual or Legal Entity authorized to submit on behalf of 53 | the copyright owner. For the purposes of this definition, "submitted" 54 | means any form of electronic, verbal, or written communication sent 55 | to the Licensor or its representatives, including but not limited to 56 | communication on electronic mailing lists, source code control systems, 57 | and issue tracking systems that are managed by, or on behalf of, the 58 | Licensor for the purpose of discussing and improving the Work, but 59 | excluding communication that is conspicuously marked or otherwise 60 | designated in writing by the copyright owner as "Not a Contribution." 61 | 62 | "Contributor" shall mean Licensor and any individual or Legal Entity 63 | on behalf of whom a Contribution has been received by Licensor and 64 | subsequently incorporated within the Work. 65 | 66 | 2. Grant of Copyright License. Subject to the terms and conditions of 67 | this License, each Contributor hereby grants to You a perpetual, 68 | worldwide, non-exclusive, no-charge, royalty-free, irrevocable 69 | copyright license to reproduce, prepare Derivative Works of, 70 | publicly display, publicly perform, sublicense, and distribute the 71 | Work and such Derivative Works in Source or Object form. 72 | 73 | 3. Grant of Patent License. Subject to the terms and conditions of 74 | this License, each Contributor hereby grants to You a perpetual, 75 | worldwide, non-exclusive, no-charge, royalty-free, irrevocable 76 | (except as stated in this section) patent license to make, have made, 77 | use, offer to sell, sell, import, and otherwise transfer the Work, 78 | where such license applies only to those patent claims licensable 79 | by such Contributor that are necessarily infringed by their 80 | Contribution(s) alone or by combination of their Contribution(s) 81 | with the Work to which such Contribution(s) was submitted. If You 82 | institute patent litigation against any entity (including a 83 | cross-claim or counterclaim in a lawsuit) alleging that the Work 84 | or a Contribution incorporated within the Work constitutes direct 85 | or contributory patent infringement, then any patent licenses 86 | granted to You under this License for that Work shall terminate 87 | as of the date such litigation is filed. 88 | 89 | 4. Redistribution. You may reproduce and distribute copies of the 90 | Work or Derivative Works thereof in any medium, with or without 91 | modifications, and in Source or Object form, provided that You 92 | meet the following conditions: 93 | 94 | (a) You must give any other recipients of the Work or 95 | Derivative Works a copy of this License; and 96 | 97 | (b) You must cause any modified files to carry prominent notices 98 | stating that You changed the files; and 99 | 100 | (c) You must retain, in the Source form of any Derivative Works 101 | that You distribute, all copyright, patent, trademark, and 102 | attribution notices from the Source form of the Work, 103 | excluding those notices that do not pertain to any part of 104 | the Derivative Works; and 105 | 106 | (d) If the Work includes a "NOTICE" text file as part of its 107 | distribution, then any Derivative Works that You distribute must 108 | include a readable copy of the attribution notices contained 109 | within such NOTICE file, excluding those notices that do not 110 | pertain to any part of the Derivative Works, in at least one 111 | of the following places: within a NOTICE text file distributed 112 | as part of the Derivative Works; within the Source form or 113 | documentation, if provided along with the Derivative Works; or, 114 | within a display generated by the Derivative Works, if and 115 | wherever such third-party notices normally appear. The contents 116 | of the NOTICE file are for informational purposes only and 117 | do not modify the License. You may add Your own attribution 118 | notices within Derivative Works that You distribute, alongside 119 | or as an addendum to the NOTICE text from the Work, provided 120 | that such additional attribution notices cannot be construed 121 | as modifying the License. 122 | 123 | You may add Your own copyright statement to Your modifications and 124 | may provide additional or different license terms and conditions 125 | for use, reproduction, or distribution of Your modifications, or 126 | for any such Derivative Works as a whole, provided Your use, 127 | reproduction, and distribution of the Work otherwise complies with 128 | the conditions stated in this License. 129 | 130 | 5. Submission of Contributions. Unless You explicitly state otherwise, 131 | any Contribution intentionally submitted for inclusion in the Work 132 | by You to the Licensor shall be under the terms and conditions of 133 | this License, without any additional terms or conditions. 134 | Notwithstanding the above, nothing herein shall supersede or modify 135 | the terms of any separate license agreement you may have executed 136 | with Licensor regarding such Contributions. 137 | 138 | 6. Trademarks. This License does not grant permission to use the trade 139 | names, trademarks, service marks, or product names of the Licensor, 140 | except as required for reasonable and customary use in describing the 141 | origin of the Work and reproducing the content of the NOTICE file. 142 | 143 | 7. Disclaimer of Warranty. Unless required by applicable law or 144 | agreed to in writing, Licensor provides the Work (and each 145 | Contributor provides its Contributions) on an "AS IS" BASIS, 146 | WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or 147 | implied, including, without limitation, any warranties or conditions 148 | of TITLE, NON-INFRINGEMENT, MERCHANTABILITY, or FITNESS FOR A 149 | PARTICULAR PURPOSE. You are solely responsible for determining the 150 | appropriateness of using or redistributing the Work and assume any 151 | risks associated with Your exercise of permissions under this License. 152 | 153 | 8. Limitation of Liability. In no event and under no legal theory, 154 | whether in tort (including negligence), contract, or otherwise, 155 | unless required by applicable law (such as deliberate and grossly 156 | negligent acts) or agreed to in writing, shall any Contributor be 157 | liable to You for damages, including any direct, indirect, special, 158 | incidental, or consequential damages of any character arising as a 159 | result of this License or out of the use or inability to use the 160 | Work (including but not limited to damages for loss of goodwill, 161 | work stoppage, computer failure or malfunction, or any and all 162 | other commercial damages or losses), even if such Contributor 163 | has been advised of the possibility of such damages. 164 | 165 | 9. Accepting Warranty or Additional Liability. While redistributing 166 | the Work or Derivative Works thereof, You may choose to offer, 167 | and charge a fee for, acceptance of support, warranty, indemnity, 168 | or other liability obligations and/or rights consistent with this 169 | License. However, in accepting such obligations, You may act only 170 | on Your own behalf and on Your sole responsibility, not on behalf 171 | of any other Contributor, and only if You agree to indemnify, 172 | defend, and hold each Contributor harmless for any liability 173 | incurred by, or claims asserted against, such Contributor by reason 174 | of your accepting any such warranty or additional liability. 175 | 176 | END OF TERMS AND CONDITIONS 177 | 178 | APPENDIX: How to apply the Apache License to your work. 179 | 180 | To apply the Apache License to your work, attach the following 181 | boilerplate notice, with the fields enclosed by brackets "[]" 182 | replaced with your own identifying information. (Don't include 183 | the brackets!) The text should be enclosed in the appropriate 184 | comment syntax for the file format. We also recommend that a 185 | file or class name and description of purpose be included on the 186 | same "printed page" as the copyright notice for easier 187 | identification within third-party archives. 188 | 189 | Copyright [yyyy] [name of copyright owner] 190 | 191 | Licensed under the Apache License, Version 2.0 (the "License"); 192 | you may not use this file except in compliance with the License. 193 | You may obtain a copy of the License at 194 | 195 | http://www.apache.org/licenses/LICENSE-2.0 196 | 197 | Unless required by applicable law or agreed to in writing, software 198 | distributed under the License is distributed on an "AS IS" BASIS, 199 | WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. 200 | See the License for the specific language governing permissions and 201 | limitations under the License. 202 | -------------------------------------------------------------------------------- /README.md: -------------------------------------------------------------------------------- 1 | # AIBench LLM Endpoints 2 | 3 | ## Overview 4 | This code provides a benchmarking runner, `AIBench-LLM`, for evaluating the performance of a large language model (LLM) inference endpoint. The benchmark measures various metrics such as Time to First Token (TTFT), End to End Latency, Inter-Token Latency (ITL), Output Tokens per Second, and more. 5 | 6 | The AIBench Runner is in charge of collecting metrics from LLM inference endpoints for the [Unify Hub](https://unify.ai/hub). More information about the full methodology is available [here 📑](https://unify.ai/docs/hub/concepts/benchmarks.html) 7 | 8 | Contributions and discussions around the methodology and the runner are definitely welcome, you can join the [Unify Discord](https://discord.com/invite/sXyFF8tDtm) if this sounds interesting! 9 | 10 | ## Metrics 11 | The benchmark runner collects the following metrics: 12 | 13 | - `load`: Number of concurrent requests. 14 | - `input_policy`: Input policy used (short or long). 15 | - `ttft`: Time-to-first-token for each request. 16 | - `e2e_latency`: End-to-end latency for each request. 17 | - `itl`: Inter-token Latency. 18 | - `cold_start`: Cold start time (if applicable). 19 | - `prompt_tokens`: Number of tokens in the input prompt. 20 | - `output_tokens`: Number of tokens in the LLM output. 21 | - `total_tokens`: Total number of tokens (input + output). 22 | - `output_tks_per_sec`: Output tokens per second. 23 | - `failed_queries`: Number of failed queries. 24 | 25 | ## Usage and Examples 26 | To be added this week! 27 | -------------------------------------------------------------------------------- /aibench/__init__.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/unifyai/aibench-llm-endpoints/72359679904f40b7475a879ba32df319619a1dea/aibench/__init__.py -------------------------------------------------------------------------------- /aibench/requirements.txt: -------------------------------------------------------------------------------- 1 | certifi==2024.2.2 2 | charset-normalizer==3.3.2 3 | idna==3.6 4 | regex==2023.12.25 5 | requests==2.31.0 6 | tiktoken==0.5.2 7 | urllib3==2.2.0 8 | -------------------------------------------------------------------------------- /aibench/runner.py: -------------------------------------------------------------------------------- 1 | import asyncio 2 | import json 3 | import logging 4 | import random 5 | import time 6 | from typing import Callable, Dict, List, Tuple 7 | 8 | import tiktoken 9 | 10 | 11 | class AIBenchRunner: 12 | """ 13 | A benchmark runner for asynchronous LLM inference endpoints. 14 | 15 | This class facilitates benchmarking asynchronous functions that take a 16 | string as input and return a string asynchronously (streaming). It supports 17 | concurrent execution of requests and provides metrics such as 18 | time-to-first-token (ttft), end-to-end latency (e2e_latency), inter token 19 | latency (itl), cold start time, prompt tokens, output tokens, total tokens, 20 | and failed queries. 21 | """ 22 | 23 | def __init__(self, fn: Callable, load: int, input_policy: str, seed: int = 42): 24 | """ 25 | Initialize the AIBenchRunner. 26 | 27 | :param fn: Callable 28 | The function to benchmark. Takes as input a string and returns a 29 | string asynchronously (streaming). 30 | :param load: int 31 | The number of concurrent requests to run. 32 | :param input_policy: str 33 | The input policy to use (short or long). 34 | :param seed: int, optional 35 | The seed for the random number generator. Default is 42. 36 | """ 37 | self.fn = fn 38 | self.load = load 39 | assert input_policy in {"short", "long"} 40 | self.input_policy = input_policy 41 | 42 | self.ttft = [] 43 | self.end_to_end_latency = [] 44 | self.cold_start = 0 45 | self.prompt_tokens = [] 46 | self.output_tokens = [] 47 | self.total_tokens = [] 48 | self.failed_queries = 0 49 | 50 | self.prompt_queue = asyncio.Queue() 51 | self.results_queue = asyncio.Queue() 52 | 53 | self.tokenizer = tiktoken.get_encoding("cl100k_base") 54 | 55 | random.seed(seed) 56 | 57 | @property 58 | def itl(self) -> List[float]: 59 | """ 60 | Compute the inter-token latency. 61 | 62 | :return: List[float] 63 | The inter-token latency. 64 | """ 65 | return [ 66 | (e2e_lat - ttft) / (o_tks - 1) 67 | for e2e_lat, ttft, o_tks in zip( 68 | self.end_to_end_latency, 69 | self.ttft, 70 | self.output_tokens, 71 | ) 72 | ] 73 | 74 | @property 75 | def output_tks_per_sec(self) -> List[float]: 76 | """ 77 | Compute the output tokens per second based on ITL. 78 | 79 | :return: List[float] 80 | The output tokens per second. 81 | """ 82 | return [1000 / i for i in self.itl] 83 | 84 | def as_dict(self) -> Dict: 85 | """ 86 | Return the stored metrics as a dictionary. 87 | 88 | :return: Dict 89 | The metrics as a dictionary. 90 | """ 91 | return { 92 | "load": self.load, 93 | "input_policy": self.input_policy, 94 | "ttft": self.ttft, 95 | "e2e_latency": self.end_to_end_latency, 96 | "itl": self.itl, 97 | "cold_start": self.cold_start, 98 | "prompt_tokens": self.prompt_tokens, 99 | "output_tokens": self.output_tokens, 100 | "total_tokens": self.total_tokens, 101 | "output_tks_per_sec": self.output_tks_per_sec, 102 | "failed_queries": self.failed_queries, 103 | } 104 | 105 | async def unpack_metrics(self): 106 | """Unpack the metrics from the results queue.""" 107 | while not self.results_queue.empty(): 108 | req_result: dict = await self.results_queue.get() 109 | self.ttft.append(req_result["ttft"]) 110 | self.end_to_end_latency.append(req_result["e2e_latency"]) 111 | self.prompt_tokens.append(req_result["prompt_tokens"]) 112 | self.output_tokens.append(req_result["output_tokens"]) 113 | self.total_tokens.append(req_result["total_tokens"]) 114 | self.failed_queries += req_result["failed_queries"] 115 | self.results_queue.task_done() 116 | 117 | @staticmethod 118 | def _get_samples(filename: str) -> List[Tuple[str, int]]: 119 | """ 120 | Retrieve samples from a JSON file. 121 | 122 | :param filename: str 123 | The name of the file to read. 124 | 125 | :return: List[Tuple[str, int]] 126 | A list of tuples containing cleaned prompts and their lengths. 127 | """ 128 | with open(filename, "r") as file: 129 | data = json.load(file) 130 | samples = [ 131 | (item["prompt"].replace("\n", ""), item["length"]) for item in data 132 | ] 133 | return samples 134 | 135 | def prepare_prompts(self) -> List[str]: 136 | """ 137 | Prepare the prompts for the benchmark. 138 | 139 | :return: List[str] 140 | The prepared prompts. 141 | """ 142 | samples_fname = f"prompts_{self.input_policy}.json" 143 | prompts = random.sample(self._get_samples(samples_fname), self.load) 144 | all_prompts = [] 145 | for prompt in prompts: 146 | count = {random.randint(0, int((4096 - prompt[1]) / prompt[1]))} 147 | preamble = f"Repeat the following line {count} times without generating the EOS token earlier than that: \n" 148 | all_prompts.append(preamble + prompt[0]) 149 | return all_prompts 150 | 151 | def _max_token_sampler(self) -> int: 152 | """ 153 | Sample the maximum number of tokens to generate. 154 | 155 | :return: int 156 | The maximum number of tokens to generate. 157 | """ 158 | if self.input_policy == "short": 159 | return int(random.normalvariate(200, 20)) 160 | return int(random.normalvariate(1000, 100)) 161 | 162 | async def compute_metrics(self) -> None: 163 | """Compute the metrics for an individual request.""" 164 | prompt = await self.prompt_queue.get() 165 | max_tokens = self._max_token_sampler() 166 | completions = [] 167 | metrics_dict = {} 168 | 169 | result = self.fn(prompt=prompt, max_tokens=max_tokens, stream=True) 170 | 171 | # Store failed queries if any 172 | metrics_dict["failed_queries"] = 0 173 | if result is None: 174 | metrics_dict["failed_queries"] = 1 175 | return 176 | 177 | # Streaming response loop 178 | start_time = time.perf_counter() 179 | async for part in result.generator(): 180 | completions.append( 181 | { 182 | "content": part["choices"][0]["delta"]["content"], 183 | "reception_time": time.perf_counter(), 184 | }, 185 | ) 186 | await asyncio.sleep(0) 187 | end_time = time.perf_counter() 188 | 189 | # Compute and store metrics 190 | content = "".join( 191 | [ 192 | completion["content"] 193 | for completion in completions 194 | if completion["content"] is not None 195 | ], 196 | ) 197 | metrics_dict["ttft"] = (completions[0]["reception_time"] - start_time) * 1000 198 | metrics_dict["e2e_latency"] = (end_time - start_time) * 1000 199 | metrics_dict["prompt_tokens"] = len(self.tokenizer.encode(prompt)) 200 | metrics_dict["output_tokens"] = len(self.tokenizer.encode(content)) 201 | metrics_dict["total_tokens"] = ( 202 | metrics_dict["prompt_tokens"] + metrics_dict["output_tokens"] 203 | ) 204 | await self.results_queue.put(metrics_dict) 205 | self.prompt_queue.task_done() 206 | 207 | async def check_coldstart(self, threshold: float) -> float: 208 | """ 209 | Check if the endpoint suffers from a cold start. 210 | 211 | :param threshold: float 212 | The threshold for the cold start. 213 | 214 | :return: float 215 | The cold start time. 216 | """ 217 | prompt = "2+2 is " 218 | completions = [] 219 | 220 | result, _ = self.fn(prompt=prompt, max_tokens=10, stream=True) 221 | 222 | if result is None: 223 | logging.warning("Run during cold start failed") 224 | return 0 225 | 226 | start_time = time.perf_counter() 227 | async for part in result.generator(): 228 | completions.append( 229 | { 230 | "content": part["choices"][0]["delta"]["content"], 231 | "reception_time": time.perf_counter(), 232 | }, 233 | ) 234 | await asyncio.sleep(0) 235 | first_token_time = completions[0]["reception_time"] 236 | try: 237 | second_token_time = completions[1]["reception_time"] 238 | except IndexError: 239 | second_token_time = 0 240 | 241 | cold_start = first_token_time - start_time 242 | if ( 243 | cold_start > threshold 244 | and (second_token_time - first_token_time) * 10 <= cold_start 245 | ): 246 | return cold_start 247 | 248 | return 0 249 | 250 | async def __call__(self) -> Dict: 251 | """ 252 | Run the benchmark. 253 | 254 | This method performs the following steps: 255 | 1. Checks for cold start by calling the `check_coldstart` method. 256 | 2. Prepares prompts and adds them to the prompt queue. 257 | 3. Initiates tasks for the concurrent requests to compute metrics. 258 | 4. Waits for all concurrent requests to complete. 259 | 5. Unpacks computed metrics into the corresponding attributes. 260 | 6. Returns the computed metrics as a dictionary. 261 | 262 | Note: 263 | This method is designed to be asynchronous and should be awaited when called. 264 | 265 | :return: Dict 266 | Computed metrics as a dictionary. 267 | """ 268 | self.cold_start = await self.check_coldstart(threshold=15) 269 | concurrent_requests = [] 270 | for prompt in self.prepare_prompts(): 271 | await self.prompt_queue.put(prompt) 272 | for _ in range(self.load): 273 | concurrent_requests.append(asyncio.create_task(self.compute_metrics())) 274 | await asyncio.gather(*concurrent_requests) 275 | await self.unpack_metrics() 276 | return self.as_dict() 277 | --------------------------------------------------------------------------------