NeMo Framework has been updated with state-of-the-art features, such as FSDP, Mixture-of-Experts, and RLHF with TensorRT-LLM to provide speedups up to 4.2x for Llama-2 pre-training on H200. All of these features will be available in an upcoming release.
NVIDIA NeMo is a conversational AI toolkit built for researchers working on automatic speech recognition (ASR), text-to-speech synthesis (TTS), large language models (LLMs), and natural language processing (NLP). The primary objective of NeMo is to help researchers from industry and academia to reuse prior work (code and pretrained models) and make it easier to create new conversational AI models.
All NeMo models are trained with Lightning and training is automatically scalable to 1000s of GPUs. Additionally, NeMo Megatron LLM models can be trained up to 1 trillion parameters using tensor and pipeline model parallelism. NeMo models can be optimized for inference and deployed for production use-cases with NVIDIA Riva.
Getting started with NeMo is simple. State of the Art pretrained NeMo models are freely available on HuggingFace Hub and NVIDIA NGC. These models can be used to transcribe audio, synthesize speech, or translate text in just a few lines of code.
We have extensive tutorials that can be run on Google Colab.
For advanced users that want to train NeMo models from scratch or finetune existing NeMo models we have a full suite of example scripts that support multi-GPU/multi-node training.
For scaling NeMo LLM training on Slurm clusters or public clouds, please see the NVIDIA NeMo Megatron Launcher. The NM launcher has extensive recipes, scripts, utilities, and documentation for training NeMo LLMs and also has an Autoconfigurator which can be used to find the optimal model parallel configuration for training on a specific cluster.
- Speech processing
- HuggingFace Space for Audio Transcription (File, Microphone and YouTube)
- Pretrained models available in 14+ languages
- Automatic Speech Recognition (ASR)
- Supported ASR models:
- Jasper, QuartzNet, CitriNet, ContextNet
- Conformer-CTC, Conformer-Transducer, FastConformer-CTC, FastConformer-Transducer
- Squeezeformer-CTC and Squeezeformer-Transducer
- LSTM-Transducer (RNNT) and LSTM-CTC
- Supports the following decoders/losses:
- CTC
- Transducer/RNNT
- Hybrid Transducer/CTC
- NeMo Original Multi-blank Transducers and Token-and-Duration Transducers (TDT)
- Streaming/Buffered ASR (CTC/Transducer) - Chunked Inference Examples
- Cache-aware Streaming Conformer with multiple lookaheads.
- Beam Search decoding
- Language Modelling for ASR (CTC and RNNT): N-gram LM in fusion with Beam Search decoding, Neural Rescoring with Transformer
- Support of long audios for Conformer with memory efficient local attention
- Speech Classification, Speech Command Recognition and Language Identification: MatchboxNet (Command Recognition), AmberNet (LangID)
- Voice activity Detection (VAD): MarbleNet
- ASR with VAD Inference - Example
- Speaker Recognition: TitaNet, ECAPA_TDNN, SpeakerNet
- Speaker Diarization
- Clustering Diarizer: TitaNet, ECAPA_TDNN, SpeakerNet
- Neural Diarizer: MSDD (Multi-scale Diarization Decoder)
- Speech Intent Detection and Slot Filling: Conformer-Transformer
- Natural Language Processing
- NeMo Megatron pre-training of Large Language Models
- Neural Machine Translation (NMT)
- Punctuation and Capitalization
- Token classification (named entity recognition)
- Text classification
- Joint Intent and Slot Classification
- Question answering
- GLUE benchmark
- Information retrieval
- Entity Linking
- Dialogue State Tracking
- Prompt Learning
- NGC collection of pre-trained NLP models.
- Synthetic Tabular Data Generation
- Text-to-Speech Synthesis (TTS):
- Documentation
- Mel-Spectrogram generators: FastPitch, SSL FastPitch, Mixer-TTS/Mixer-TTS-X, RAD-TTS, Tacotron2
- Vocoders: HiFiGAN, UnivNet, WaveGlow
- End-to-End Models: VITS
- Pre-trained Model Checkpoints in NVIDIA GPU Cloud (NGC)
- Tools
- Text Processing (text normalization and inverse text normalization)
- NeMo Forced Aligner
- CTC-Segmentation tool
- Speech Data Explorer: a dash-based tool for interactive exploration of ASR/TTS datasets
- Speech Data Processor
Built for speed, NeMo can utilize NVIDIA's Tensor Cores and scale out training to multiple GPUs and multiple nodes.
- Python 3.10 or above
- Pytorch 1.13.1 or above
- NVIDIA GPU, if you intend to do model training
| Version | Status | Description |
|---|---|---|
| Latest | Documentation of the latest (i.e. main) branch. | |
| Stable | Documentation of the stable (i.e. most recent release) branch. |
A great way to start with NeMo is by checking one of our tutorials.
You can also get a high-level overview of NeMo by watching the talk NVIDIA NeMo: Toolkit for Conversational AI, presented at PyData Yerevan 2022:
FAQ can be found on NeMo's Discussions board. You are welcome to ask questions or start discussions there.
We recommend installing NeMo in a fresh Conda environment.
Install PyTorch using their configurator.
The command used to install PyTorch may depend on your system. Please use the configurator linked above to find the right command for your system.
Use this installation mode if you want the latest released version.
Depending on the shell used, you may need to use "nemo_toolkit[all]" instead in the above command.
Use this installation mode if you want the version from a particular GitHub branch (e.g main).
Use this installation mode if you are contributing to NeMo.
If you only want the toolkit without additional conda-based dependencies, you may replace reinstall.sh with pip install -e . when your PWD is the root of the NeMo repository.
To install NeMo on Mac with Apple M-Series GPU:
- create a new Conda environment
- install PyTorch 2.0 or higher
- run the following code:
# [optional] install mecab using Homebrew, to use sacrebleu for NLP collection
# you can install Homebrew here: https://brew.sh
brew install mecab
# [optional] install pynini using Conda, to use text normalization
conda install -c conda-forge pynini
# install Cython manually
pip install cython
# clone the repo and install in development mode
git clone https://github.com/NVIDIA/NeMo
cd NeMo
./reinstall.shOne of the options is using Windows Subsystem for Linux (WSL).
To install WSL:
- In PowerShell, run the following code:
wsl --install
# [note] If you run wsl --install and see the WSL help text, it means WSL is already installed.Learn more about installing WSL at Microsoft's official documentation.
- After Installing your Linux distribution with WSL:
- Option 1: Open the distribution (Ubuntu by default) from the Start menu and follow the instructions.
- Option 2: Launch the Terminal application. Download it from Microsoft's Windows Terminal page if not installed.
Next, follow the instructions for Linux systems, as provided above. For example:
Note that RNNT requires numba to be installed from conda.
NeMo Megatron training requires NVIDIA Apex to be installed. Install it manually if not using the NVIDIA PyTorch container.
To install Apex, run
git clone https://github.com/NVIDIA/apex.git
cd apex
git checkout 52e18c894223800cb611682dce27d88050edf1de
pip install install -v --no-build-isolation --disable-pip-version-check --no-cache-dir --config-settings "--build-option=--cpp_ext --cuda_ext --fast_layer_norm --distributed_adam --deprecated_fused_adam" ./It is highly recommended to use the NVIDIA PyTorch or NeMo container if having issues installing Apex or any other dependencies.
While installing Apex, it may raise an error if the CUDA version on your system does not match the CUDA version torch was compiled with. This raise can be avoided by commenting it here: https://github.com/NVIDIA/apex/blob/master/setup.py#L32
cuda-nvprof is needed to install Apex. The version should match the CUDA version that you are using:
packaging is also needed:
With the latest versions of Apex, the pyproject.toml file in Apex may need to be deleted in order to install locally.
NeMo Megatron GPT has been integrated with NVIDIA Transformer Engine Transformer Engine enables FP8 training on NVIDIA Hopper GPUs. Install it manually if not using the NVIDIA PyTorch container.
It is highly recommended to use the NVIDIA PyTorch or NeMo container if having issues installing Transformer Engine or any other dependencies.
Transformer Engine requires PyTorch to be built with CUDA 11.8.
Transformer Engine already supports Flash Attention for GPT models. If you want to use Flash Attention for non-causal models, please install flash-attn. If you want to use Flash Attention with attention bias (introduced from position encoding, e.g. Alibi), please also install triton pinned version following the implementation.
To launch the inference web UI server, please install the gradio gradio.
NeMo Text Processing, specifically (Inverse) Text Normalization, is now a separate repository https://github.com/NVIDIA/NeMo-text-processing.
We release NeMo containers alongside NeMo releases. For example, NeMo r1.21.0 comes with container nemo:23.08, you may find more details about released containers in releases page.
To use built container, please run
To build a nemo container with Dockerfile from a branch, please run
If you choose to work with the main branch, we recommend using NVIDIA's PyTorch container version 23.08-py3 and then installing from GitHub.
Many examples can be found under the "Examples" folder.
We welcome community contributions! Please refer to CONTRIBUTING.md for the process.
We provide an ever-growing list of publications that utilize the NeMo framework.
If you would like to add your own article to the list, you are welcome to do so via a pull request to this repository's gh-pages-src branch. Please refer to the instructions in the README of that branch.
NeMo is released under an Apache 2.0 license.