evalscope

English | 简体中文

📖 Documents

⭐ If you like this project, please click the "Star" button at the top right to support us. Your support is our motivation to keep going!

📋 Table of Contents

• Introduction
• News
• Installation
• Quick Start
• Evaluation Backend
• Custom Dataset Evaluation
• Offline Evaluation
• Arena Mode
• Model Serving Performance Evaluation

📝 Introduction

EvalScope is the official model evaluation and performance benchmarking framework launched by the ModelScope community. It comes with built-in common benchmarks and evaluation metrics, such as MMLU, CMMLU, C-Eval, GSM8K, ARC, HellaSwag, TruthfulQA, MATH, and HumanEval. EvalScope supports various types of model evaluations, including LLMs, multimodal LLMs, embedding models, and reranker models. It is also applicable to multiple evaluation scenarios, such as end-to-end RAG evaluation, arena mode, and model inference performance stress testing. Moreover, with the seamless integration of the ms-swift training framework, evaluations can be initiated with a single click, providing full end-to-end support from model training to evaluation 🚀

EvalScope Framework.

The architecture includes the following modules:

1. Model Adapter: The model adapter is used to convert the outputs of specific models into the format required by the framework, supporting both API call models and locally run models.
2. Data Adapter: The data adapter is responsible for converting and processing input data to meet various evaluation needs and formats.
3. Evaluation Backend:
   • Native: EvalScope’s own default evaluation framework, supporting various evaluation modes, including single model evaluation, arena mode, baseline model comparison mode, etc.
   • OpenCompass: Supports OpenCompass as the evaluation backend, providing advanced encapsulation and task simplification, allowing you to submit tasks for evaluation more easily.
   • VLMEvalKit: Supports VLMEvalKit as the evaluation backend, enabling easy initiation of multi-modal evaluation tasks, supporting various multi-modal models and datasets.
   • RAGEval: Supports RAG evaluation, supporting independent evaluation of embedding models and rerankers using MTEB/CMTEB, as well as end-to-end evaluation using RAGAS.
   • ThirdParty: Other third-party evaluation tasks, such as ToolBench.
4. Performance Evaluator: Model performance evaluation, responsible for measuring model inference service performance, including performance testing, stress testing, performance report generation, and visualization.
5. Evaluation Report: The final generated evaluation report summarizes the model's performance, which can be used for decision-making and further model optimization.
6. Visualization: Visualization results help users intuitively understand evaluation results, facilitating analysis and comparison of different model performances.

🎉 News

• 🔥 [2024.11.26] The model inference service performance evaluator has been completely refactored: it now supports local inference service startup and Speed Benchmark; asynchronous call error handling has been optimized. For more details, refer to the 📖 User Guide.
• 🔥 [2024.10.31] The best practice for evaluating Multimodal-RAG has been updated, please check the 📖 Blog for more details.
• 🔥 [2024.10.23] Supports multimodal RAG evaluation, including the assessment of image-text retrieval using CLIP_Benchmark, and extends RAGAS to support end-to-end multimodal metrics evaluation.
• 🔥 [2024.10.8] Support for RAG evaluation, including independent evaluation of embedding models and rerankers using MTEB/CMTEB, as well as end-to-end evaluation using RAGAS.
• 🔥 [2024.09.18] Our documentation has been updated to include a blog module, featuring some technical research and discussions related to evaluations. We invite you to 📖 read it.
• 🔥 [2024.09.12] Support for LongWriter evaluation, which supports 10,000+ word generation. You can use the benchmark LongBench-Write to measure the long output quality as well as the output length.
• 🔥 [2024.08.30] Support for custom dataset evaluations, including text datasets and multimodal image-text datasets.
• 🔥 [2024.08.20] Updated the official documentation, including getting started guides, best practices, and FAQs. Feel free to 📖read it here!
• 🔥 [2024.08.09] Simplified the installation process, allowing for pypi installation of vlmeval dependencies; optimized the multimodal model evaluation experience, achieving up to 10x acceleration based on the OpenAI API evaluation chain.
• 🔥 [2024.07.31] Important change: The package name llmuses has been changed to evalscope. Please update your code accordingly.
• 🔥 [2024.07.26] Support for VLMEvalKit as a third-party evaluation framework to initiate multimodal model evaluation tasks.
• 🔥 [2024.06.29] Support for OpenCompass as a third-party evaluation framework, which we have encapsulated at a higher level, supporting pip installation and simplifying evaluation task configuration.
• 🔥 [2024.06.13] EvalScope seamlessly integrates with the fine-tuning framework SWIFT, providing full-chain support from LLM training to evaluation.
• 🔥 [2024.06.13] Integrated the Agent evaluation dataset ToolBench.

🛠️ Installation

Method 1: Install Using pip

We recommend using conda to manage your environment and installing dependencies with pip:

1. Create a conda environment (optional)

   # It is recommended to use Python 3.10
   conda create -n evalscope python=3.10
   # Activate the conda environment
   conda activate evalscope
   

2. Install dependencies using pip

   pip install evalscope                # Install Native backend (default)
   # Additional options
   pip install evalscope[opencompass]   # Install OpenCompass backend
   pip install evalscope[vlmeval]       # Install VLMEvalKit backend
   pip install evalscope[rag]           # Install RAGEval backend
   pip install evalscope[perf]          # Install Perf dependencies
   pip install evalscope[all]           # Install all backends (Native, OpenCompass, VLMEvalKit, RAGEval)
   

[!WARNING] As the project has been renamed to evalscope, for versions v0.4.3 or earlier, you can install using the following command:

pip install llmuses<=0.4.3

To import relevant dependencies using llmuses:

from llmuses import ...

Method 2: Install from Source

1. Download the source code

   git clone https://github.com/modelscope/evalscope.git
   

2. Install dependencies

   cd evalscope/
   pip install -e .                  # Install Native backend
   # Additional options
   pip install -e '.[opencompass]'   # Install OpenCompass backend
   pip install -e '.[vlmeval]'       # Install VLMEvalKit backend
   pip install -e '.[rag]'           # Install RAGEval backend
   pip install -e '.[perf]'          # Install Perf dependencies
   pip install -e '.[all]'           # Install all backends (Native, OpenCompass, VLMEvalKit, RAGEval)
   

🚀 Quick Start

1. Simple Evaluation

To evaluate a model using default settings on specified datasets, follow the process below:

Installation using pip

You can execute this in any directory:

python -m evalscope.run \
 --model Qwen/Qwen2.5-0.5B-Instruct \
 --template-type qwen \
 --datasets gsm8k ceval \
 --limit 10

Installation from source

You need to execute this in the evalscope directory:

python evalscope/run.py \
 --model Qwen/Qwen2.5-0.5B-Instruct \
 --template-type qwen \
 --datasets gsm8k ceval \
 --limit 10

If prompted with Do you wish to run the custom code? [y/N], please type y.

Results (tested with only 10 samples)

Report table:
+-----------------------+--------------------+-----------------+
| Model                 | ceval              | gsm8k           |
+=======================+====================+=================+
| Qwen2.5-0.5B-Instruct | (ceval/acc) 0.5577 | (gsm8k/acc) 0.5 |
+-----------------------+--------------------+-----------------+

Basic Parameter Descriptions

• --model: Specifies the model_id of the model on ModelScope, allowing automatic download. For example, see the Qwen2-0.5B-Instruct model link; you can also use a local path, such as /path/to/model.
• --template-type: Specifies the template type corresponding to the model. Refer to the Default Template field in the template table for filling in this field.
• --datasets: The dataset name, allowing multiple datasets to be specified, separated by spaces; these datasets will be automatically downloaded. Refer to the supported datasets list for available options.
• --limit: Maximum number of evaluation samples per dataset; if not specified, all will be evaluated, which is useful for quick validation.

2. Parameterized Evaluation

If you wish to conduct a more customized evaluation, such as modifying model parameters or dataset parameters, you can use the following commands:

Example 1:

python evalscope/run.py \
 --model qwen/Qwen2-0.5B-Instruct \
 --template-type qwen \
 --model-args revision=master,precision=torch.float16,device_map=auto \
 --datasets gsm8k ceval \
 --use-cache true \
 --limit 10

Example 2:

python evalscope/run.py \
 --model qwen/Qwen2-0.5B-Instruct \
 --template-type qwen \
 --generation-config do_sample=false,temperature=0.0 \
 --datasets ceval \
 --dataset-args '{"ceval": {"few_shot_num": 0, "few_shot_random": false}}' \
 --limit 10

Parameter Descriptions

In addition to the three basic parameters, the other parameters are as follows:

• --model-args: Model loading parameters, separated by commas, in key=value format.
• --generation-config: Generation parameters, separated by commas, in key=value format.
  • do_sample: Whether to use sampling, default is false.
  • max_new_tokens: Maximum generation length, default is 1024.
  • temperature: Sampling temperature.
  • top_p: Sampling threshold.
  • top_k: Sampling threshold.
• --use-cache: Whether to use local cache, default is false. If set to true, previously evaluated model and dataset combinations will not be evaluated again, and will be read directly from the local cache.
• --dataset-args: Evaluation dataset configuration parameters, provided in JSON format, where the key is the dataset name and the value is the parameter; note that these must correspond one-to-one with the values in --datasets.
  • --few_shot_num: Number of few-shot examples.
  • --few_shot_random: Whether to randomly sample few-shot data; if not specified, defaults to true.

3. Use the run_task Function to Submit an Evaluation Task

Using the run_task function to submit an evaluation task requires the same parameters as the command line. You need to pass a dictionary as the parameter, which includes the following fields:

1. Configuration Task Dictionary Parameters

import torch
from evalscope.constants import DEFAULT_ROOT_CACHE_DIR

# Example
your_task_cfg = {
        'model_args': {'revision': None, 'precision': torch.float16, 'device_map': 'auto'},
        'generation_config': {'do_sample': False, 'repetition_penalty': 1.0, 'max_new_tokens': 512},
        'dataset_args': {},
        'dry_run': False,
        'model': 'qwen/Qwen2-0.5B-Instruct',
        'template_type': 'qwen',
        'datasets': ['arc', 'hellaswag'],
        'work_dir': DEFAULT_ROOT_CACHE_DIR,
        'outputs': DEFAULT_ROOT_CACHE_DIR,
        'mem_cache': False,
        'dataset_hub': 'ModelScope',
        'dataset_dir': DEFAULT_ROOT_CACHE_DIR,
        'limit': 10,
        'debug': False
    }

Here, DEFAULT_ROOT_CACHE_DIR is set to '~/.cache/evalscope'.

2. Execute Task with run_task

from evalscope.run import run_task
run_task(task_cfg=your_task_cfg)

Evaluation Backend

EvalScope supports using third-party evaluation frameworks to initiate evaluation tasks, which we call Evaluation Backend. Currently supported Evaluation Backend includes:

• Native: EvalScope's own default evaluation framework, supporting various evaluation modes including single model evaluation, arena mode, and baseline model comparison mode.
• OpenCompass: Initiate OpenCompass evaluation tasks through EvalScope. Lightweight, easy to customize, supports seamless integration with the LLM fine-tuning framework ms-swift. 📖 User Guide
• VLMEvalKit: Initiate VLMEvalKit multimodal evaluation tasks through EvalScope. Supports various multimodal models and datasets, and offers seamless integration with the LLM fine-tuning framework ms-swift. 📖 User Guide
• RAGEval: Initiate RAG evaluation tasks through EvalScope, supporting independent evaluation of embedding models and rerankers using MTEB/CMTEB, as well as end-to-end evaluation using RAGAS: 📖 User Guide
• ThirdParty: Third-party evaluation tasks, such as ToolBench and LongBench-Write.

Model Serving Performance Evaluation

A stress testing tool focused on large language models, which can be customized to support various dataset formats and different API protocol formats.

Reference: Performance Testing 📖 User Guide

Supports wandb for recording results

Supports Speed Benchmark

It supports speed testing and provides speed benchmarks similar to those found in the official Qwen reports:

Speed Benchmark Results:
+---------------+-----------------+----------------+
| Prompt Tokens | Speed(tokens/s) | GPU Memory(GB) |
+---------------+-----------------+----------------+
|       1       |      50.69      |      0.97      |
|     6144      |      51.36      |      1.23      |
|     14336     |      49.93      |      1.59      |
|     30720     |      49.56      |      2.34      |
+---------------+-----------------+----------------+

Custom Dataset Evaluation

EvalScope supports custom dataset evaluation. For detailed information, please refer to the Custom Dataset Evaluation 📖User Guide

Offline Evaluation

You can use local dataset to evaluate the model without internet connection.

Refer to: Offline Evaluation 📖 User Guide

Arena Mode

The Arena mode allows multiple candidate models to be evaluated through pairwise battles, and can choose to use the AI Enhanced Auto-Reviewer (AAR) automatic evaluation process or manual evaluation to obtain the evaluation report.

Refer to: Arena Mode 📖 User Guide

TO-DO List

• RAG evaluation
• VLM evaluation
• Agents evaluation
• vLLM
• Distributed evaluating
• Multi-modal evaluation
• Benchmarks
  • GAIA
  • GPQA
  • MBPP
• Auto-reviewer
  • Qwen-max

Star History