NeMo Eval

Documentation | Examples | Contributing

Overview

The NeMo Framework is NVIDIA’s GPU-accelerated, end-to-end training platform for large language models (LLMs), multimodal models, and speech models. It enables seamless scaling of both pretraining and post-training workloads, from a single GPU to clusters with thousands of nodes, supporting Hugging Face/PyTorch and Megatron models. NeMo includes a suite of libraries and curated training recipes to help users build models from start to finish.

The Eval library ("NeMo Eval") is a comprehensive evaluation module within the NeMo Framework for LLMs. It offers streamlined deployment and advanced evaluation capabilities for models trained using NeMo, leveraging state-of-the-art evaluation harnesses.

🚀 Features

• Multi-Backend Deployment: Supports PyTriton and multi-instance evaluations using the Ray Serve deployment backend
• Comprehensive Evaluation: Includes state-of-the-art evaluation harnesses for academic benchmarks, reasoning benchmarks, code generation, and safety testing
• Adapter System: Features a flexible architecture with chained interceptors for customizable request and response processing
• Production-Ready: Supports high-performance inference with CUDA graphs and flash decoding
• Multi-GPU and Multi-Node Support: Enables distributed inference across multiple GPUs and compute nodes
• OpenAI-Compatible API: Provides RESTful endpoints aligned with OpenAI API specifications

🔧 Install NeMo Eval

Prerequisites

• Python 3.10 or higher
• CUDA-compatible GPU(s) (tested on RTX A6000, A100, H100)
• NeMo Framework container (recommended)

Use pip

For quick exploration of NeMo Eval, we recommend installing our pip package:

pip install nemo-eval

Use Docker

For optimal performance and user experience, use the latest version of the NeMo Framework container. Please fetch the most recent $TAG and run the following command to start a container:

docker run --rm -it -w /workdir -v $(pwd):/workdir \
  --entrypoint bash \
  --gpus all \
  nvcr.io/nvidia/nemo:${TAG}

Use uv

To install NeMo Eval with uv, please refer to our Contribution guide.

🚀 Quick Start

1. Deploy a Model

from nemo_eval.api import deploy

# Deploy a NeMo checkpoint
deploy(
    nemo_checkpoint="/path/to/your/checkpoint",
    serving_backend="pytriton",  # or "ray"
    server_port=8080,
    num_gpus=1,
    max_input_len=4096,
    max_batch_size=8
)

2. Evaluate the Model

from nvidia_eval_commons.core.evaluate import evaluate
from nvidia_eval_commons.api.api_dataclasses import ApiEndpoint, EvaluationConfig, EvaluationTarget

# Configure evaluation
api_endpoint = ApiEndpoint(
    url="http://0.0.0.0:8080/v1/completions/",
    model_id="megatron_model"
)
target = EvaluationTarget(api_endpoint=api_endpoint)
config = EvaluationConfig(type="gsm8k", output_dir="results")

# Run evaluation
results = evaluate(target_cfg=target, eval_cfg=config)
print(results)

📊 Support Matrix

Checkpoint Type	Inference Backend	Deployment Server	Evaluation Harnesses Supported
NeMo FW checkpoint via Megatron Core backend	Megatron Core in-framework inference engine	PyTriton (single and multi node model parallelism), Ray (single node model parallelism with multi instance evals)	lm-evaluation-harness, simple-evals, BigCode, BFCL, safety-harness, garak

🏗️ Architecture

Core Components

1. Deployment Layer

• PyTriton Backend: Provides high-performance inference through the NVIDIA Triton Inference Server, with OpenAI API compatibility via a FastAPI interface. Supports model parallelism across single-node and multi-node configurations. Note: Multi-instance evaluation is not supported.
• Ray Backend: Enables multi-instance evaluation with model parallelism on a single node using Ray Serve, while maintaining OpenAI API compatibility. Multi-node support is coming soon.

2. Evaluation Layer

• NVIDIA Eval Factory: Provides standardized benchmark evaluations using packages from NVIDIA Eval Factory, bundled in the NeMo Framework container. The lm-evaluation-harness is pre-installed by default, and additional tools listed in the support matrix can be added as needed. For more information, see the documentation.

• Adapter System: Flexible request/response processing pipeline with Interceptors that provide modular processing:

  • Available Interceptors: Modular components for request/response processing
    • SystemMessageInterceptor: Customize system prompts
    • RequestLoggingInterceptor: Log incoming requests
    • ResponseLoggingInterceptor: Log outgoing responses
    • ResponseReasoningInterceptor: Process reasoning outputs
    • EndpointInterceptor: Route requests to the actual model

📖 Usage Examples

Basic Deployment with PyTriton as the Serving Backend

from nemo_eval.api import deploy

# Deploy model
deploy(
    nemo_checkpoint="/path/to/checkpoint",
    serving_backend="pytriton",
    server_port=8080,
    num_gpus=1,
    max_input_len=8192,
    max_batch_size=4
)

Basic Evaluation

from nvidia_eval_commons.core.evaluate import evaluate
from nvidia_eval_commons.api.api_dataclasses import ApiEndpoint, ConfigParams, EvaluationConfig, EvaluationTarget
# Configure Endpoint
api_endpoint = ApiEndpoint(
    url="http://0.0.0.0:8080/v1/completions/",
    model_id="megatron_model"
)
# Evaluation target configuration
target = EvaluationTarget(api_endpoint=api_endpoint)
# Configure EvaluationConfig with type, number of samples to evaluate on, etc.
config = EvaluationConfig(type="gsm8k",
            output_dir="results",
            params=ConfigParams(
                    limit_samples=10
                ))

# Run evaluation
results = evaluate(target_cfg=target, eval_cfg=config)

Use Adapters

The example below demonstrates how to configure an Adapter to provide a custom system prompt. Requests and responses are processed through interceptors, which are automatically selected based on the parameters defined in AdapterConfig.

from nemo_eval.utils.api import AdapterConfig

# Configure adapter for reasoning
adapter_config = AdapterConfig(
    api_url="http://0.0.0.0:8080/v1/completions/",
    use_reasoning=True,
    end_reasoning_token="</think>",
    custom_system_prompt="You are a helpful assistant that thinks step by step.",
    max_logged_requests=5,
    max_logged_responses=5
)

# Run evaluation with adapter
results = evaluate(
    target_cfg=target,
    eval_cfg=config,
    adapter_cfg=adapter_config
)

Deploy with Multiple GPUs

# Deploy with tensor parallelism or pipeline parallelism
deploy(
    nemo_checkpoint="/path/to/checkpoint",
    serving_backend="pytriton",
    num_gpus=4,
    tensor_parallelism_size=4,
    pipeline_parallelism_size=1,
    max_input_len=8192,
    max_batch_size=8
)

Deploy with Ray

# Deploy using Ray Serve
deploy(
    nemo_checkpoint="/path/to/checkpoint",
    serving_backend="ray",
    num_gpus=2,
    num_replicas=2,
    num_cpus_per_replica=8,
    server_port=8080,
    include_dashboard=True,
    cuda_visible_devices="0,1"
)

📁 Project Structure

Eval/
├── src/nemo_eval/           # Main package
│   ├── api.py               # Main API functions
│   ├── package_info.py      # Package metadata
│   ├── adapters/            # Adapter system
│   │   ├── server.py        # Adapter server
│   │   ├── utils.py         # Adapter utilities
│   │   └── interceptors/    # Request/response interceptors
│   └── utils/               # Utility modules
│       ├── api.py           # API configuration classes
│       ├── base.py          # Base utilities
│       └── ray_deploy.py    # Ray deployment utilities
├── tests/                   # Test suite
│   ├── unit_tests/          # Unit tests
│   └── functional_tests/    # Functional tests
├── tutorials/               # Tutorial notebooks
├── scripts/                 # Reference nemo-run scripts
├── docs/                    # Documentation
├── docker/                  # Docker configuration
└── external/                # External dependencies

🤝 Contributing

We welcome contributions! Please see our Contributing Guide for details on development setup, testing, and code style guidelines

📄 License

This project is licensed under the Apache License 2.0. See the LICENSE file for details.

📞 Support

• Issues: GitHub Issues
• Discussions: GitHub Discussions
• Documentation: NeMo Documentation

🔗 Related Projects

• NeMo Export Deploy - Model export and deployment

---

Note: This project is actively maintained by NVIDIA. For the latest updates and features, please check our releases page.