Your go-to toolkit for lightning-fast, flexible LLM evaluation, from Hugging Face's Leaderboard and Evals Team.

---

---

Lighteval is your all-in-one toolkit for evaluating LLMs across multiple backends—whether your model is being served somewhere or already loaded in memory. Dive deep into your model's performance by saving and exploring detailed, sample-by-sample results to debug and see how your models stack-up.

Customization at your fingertips: letting you either browse all our existing tasks and metrics or effortlessly create your own custom task and custom metric, tailored to your needs.

Available Tasks

Lighteval supports 7,000+ evaluation tasks across multiple domains and languages. Here's an overview of some popular benchmarks:

📚 Knowledge

• General Knowledge: MMLU, MMLU-Pro, MMMU, BIG-Bench
• Question Answering: TriviaQA, Natural Questions, SimpleQA, Humanity's Last Exam (HLE)
• Specialized: GPQA, AGIEval

🧮 Math and Code

• Math Problems: GSM8K, GSM-Plus, MATH, MATH500
• Competition Math: AIME24, AIME25
• Multilingual Math: MGSM (Grade School Math in 10+ languages)
• Coding Benchmarks: LCB (LiveCodeBench)

🎯 Chat Model Evaluation

• Instruction Following: IFEval, IFEval-fr
• Reasoning: MUSR, DROP (discrete reasoning)
• Long Context: RULER
• Dialogue: MT-Bench
• Holistic Evaluation: HELM, BIG-Bench

🌍 Multilingual Evaluation

• Cross-lingual: XTREME, Flores200 (200 languages), XCOPA, XQuAD
• Language-specific:
  • Arabic: ArabicMMLU
  • Filipino: FilBench
  • French: IFEval-fr, GPQA-fr, BAC-fr
  • German: German RAG Eval
  • Serbian: Serbian LLM Benchmark, OZ Eval
  • Turkic: TUMLU (9 Turkic languages)
  • Chinese: CMMLU, CEval, AGIEval
  • Russian: RUMMLU, Russian SQuAD
  • And many more...

🧠 Core Language Understanding

• NLU: GLUE, SuperGLUE, TriviaQA, Natural Questions
• Commonsense: HellaSwag, WinoGrande, ProtoQA
• Natural Language Inference: XNLI
• Reading Comprehension: SQuAD, XQuAD, MLQA, Belebele

⚡️ Installation

Note: lighteval is currently completely untested on Windows, and we don't support it yet. (Should be fully functional on Mac/Linux)

pip install lighteval

Lighteval allows for many extras when installing, see here for a complete list.

If you want to push results to the Hugging Face Hub, add your access token as an environment variable:

huggingface-cli login

🚀 Quickstart

Lighteval offers the following entry points for model evaluation:

• lighteval accelerate: Evaluate models on CPU or one or more GPUs using 🤗 Accelerate
• lighteval nanotron: Evaluate models in distributed settings using ⚡️ Nanotron
• lighteval vllm: Evaluate models on one or more GPUs using 🚀 VLLM
• lighteval sglang: Evaluate models using SGLang as backend
• lighteval endpoint: Evaluate models using various endpoints as backend
  • lighteval endpoint inference-endpoint: Evaluate models using Hugging Face's Inference Endpoints API
  • lighteval endpoint tgi: Evaluate models using 🔗 Text Generation Inference running locally
  • lighteval endpoint litellm: Evaluate models on any compatible API using LiteLLM
  • lighteval endpoint inference-providers: Evaluate models using HuggingFace's inference providers as backend

Did not find what you need ? You can always make your custom model API by following this guide

• lighteval custom: Evaluate custom models (can be anything)

Here's a quick command to evaluate using the Accelerate backend:

lighteval accelerate \
    "model_name=gpt2" \
    "leaderboard|truthfulqa:mc|0"

Or use the Python API to run a model already loaded in memory!

from transformers import AutoModelForCausalLM

from lighteval.logging.evaluation_tracker import EvaluationTracker
from lighteval.models.transformers.transformers_model import TransformersModel, TransformersModelConfig
from lighteval.pipeline import ParallelismManager, Pipeline, PipelineParameters


MODEL_NAME = "meta-llama/Meta-Llama-3-8B-Instruct"
BENCHMARKS = "lighteval|gsm8k|0"

evaluation_tracker = EvaluationTracker(output_dir="./results")
pipeline_params = PipelineParameters(
    launcher_type=ParallelismManager.NONE,
    max_samples=2
)

model = AutoModelForCausalLM.from_pretrained(
  MODEL_NAME, device_map="auto"
)
config = TransformersModelConfig(model_name=MODEL_NAME, batch_size=1)
model = TransformersModel.from_model(model, config)

pipeline = Pipeline(
    model=model,
    pipeline_parameters=pipeline_params,
    evaluation_tracker=evaluation_tracker,
    tasks=BENCHMARKS,
)

results = pipeline.evaluate()
pipeline.show_results()
results = pipeline.get_results()

🙏 Acknowledgements

Lighteval started as an extension of the fantastic Eleuther AI Harness (which powers the Open LLM Leaderboard) and draws inspiration from the amazing HELM framework.

While evolving Lighteval into its own standalone tool, we are grateful to the Harness and HELM teams for their pioneering work on LLM evaluations.

🌟 Contributions Welcome 💙💚💛💜🧡

Got ideas? Found a bug? Want to add a task or metric? Contributions are warmly welcomed!

If you're adding a new feature, please open an issue first.

If you open a PR, don't forget to run the styling!

pip install -e .[dev]
pre-commit install
pre-commit run --all-files

📜 Citation

@misc{lighteval,
  author = {Habib, Nathan and Fourrier, Clémentine and Kydlíček, Hynek and Wolf, Thomas and Tunstall, Lewis},
  title = {LightEval: A lightweight framework for LLM evaluation},
  year = {2023},
  version = {0.10.0},
  url = {https://github.com/huggingface/lighteval}
}