Nanotron is a library for pretraining transformer models. It provides a simple and flexible API to pretrain models on custom datasets. Nanotron is designed to be easy to use, fast, and scalable. It is built with the following principles in mind:
- Simplicity: Nanotron is designed to be easy to use. It provides a simple and flexible API to pretrain models on custom datasets.
- Performance: Optimized for speed and scalability, Nanotron uses the latest techniques to train models faster and more efficiently.
# Requirements: Python>=3.10
git clone https://github.com/huggingface/nanotron
cd nanotron
pip install --upgrade pip
pip install --pre torch --index-url https://download.pytorch.org/whl/nightly/cu121
pip install -e .
# Install dependencies if you want to use the example scripts
pip install datasets transformers
pip install triton "flash-attn>=2.5.0" --no-build-isolation
Note
If you get undefined symbol: ncclCommRegister
error you should install torch 2.1.2 instead: pip install torch==2.1.2 --index-url https://download.pytorch.org/whl/cu121
Tip
We log to wandb automatically if it's installed. For that you can use pip install wandb
. If you don't want to use wandb, you can run wandb disabled
.
The following command will train a tiny Llama model on a single node with 8 GPUs. The model will be saved in the checkpoints
directory as specified in the config file.
CUDA_DEVICE_MAX_CONNECTIONS=1 torchrun --nproc_per_node=8 run_train.py --config-file examples/config_tiny_llama.yaml
torchrun --nproc_per_node=1 run_generate.py --ckpt-path checkpoints/10/ --tp 1 --pp 1
# We could set a larger TP for faster generation, and a larger PP in case of very large models.
You can find more examples in the /examples
directory:
Example | Description |
---|---|
custom-dataloader |
Plug a custom dataloader to nanotron |
datatrove |
Use the datatrove library to load data |
doremi |
Use DoReMi to speed up training |
mamba |
Train an example Mamba model |
moe |
Train an example Mixture-of-Experts (MoE) model |
mup |
Use spectral µTransfer to scale up your model |
We're working on adding more examples soon! Feel free to add a PR to add your own example. 🚀
We currently support the following features:
- 3D parallelism (DP+TP+PP)
- Expert parallelism for MoEs
- AFAB and 1F1B schedules for PP
- Explicit APIs for TP and PP which enables easy debugging
- ZeRO-1 optimizer
- FP32 gradient accumulation
- Parameter tying/sharding
- Custom module checkpointing for large models
- Spectral µTransfer parametrization for scaling up neural networks
- Mamba example
And we have on our roadmap:
- FP8 training
- ZeRO-3 optimizer (a.k.a FSDP)
-
torch.compile
support - Ring attention
- Interleaved 1f1b schedule
We would like to thank everyone working on LLMs, especially those sharing their work openly from which we took great inspiration: Nvidia for Megatron-LM/apex
, Microsoft for DeepSpeed
, HazyResearch for flash-attn
..