Important

NeMo 2.0 is an experimental feature and currently released in the dev container only: nvcr.io/nvidia/nemo:dev. Please refer to NeMo 2.0 overview for information on getting started.

Nemotron

Nemotron is a Large Language Model (LLM) that can be integrated into a synthetic data generation pipeline to produce training data, assisting researchers and developers in building their own LLMs.

NeMo 2.0 Pretraining Recipes

We provide recipes for pretraining nemotron models for the following sizes: 4B, 8B, 15B, 22B and 340B using NeMo 2.0 and NeMo-Run. These recipes configure a run.Partial for one of the nemo.collections.llm api functions introduced in NeMo 2.0. The recipes are hosted in nemotron3_4b, nemotron3_8b, nemotron4_15b, nemotron4_22b , and nemotron4_340b files.

Note

The pretraining recipes use the MockDataModule for the data argument. You are expected to replace the MockDataModule with your own custom dataset.

We provide an example below on how to invoke the default recipe and override the data argument:

from nemo.collections import llm

pretrain = llm.nemotron3_8b.pretrain_recipe(
    name="nemotron3_8b_pretraining",
    ckpt_dir=f"/path/to/checkpoints",
    num_nodes=2,
    num_gpus_per_node=8,
)

dataloader = a_function_that_configures_your_custom_dataset(
    global_batch_size=global_batch_size,
    micro_batch_size=micro_batch_size,
    seq_length=pretrain.model.config.seq_length,
)
pretrain.data = dataloader

Note

The configuration in the recipes is done using the NeMo-Run run.Config and run.Partial configuration objects. Please review the NeMo-Run documentation to learn more about its configuration and execution system.

Once you have your final configuration ready, you can execute it on any of the NeMo-Run supported executors. The simplest is the local executor, which just runs the pretraining locally in a separate process. You can use it as follows:

import nemo_run as run

run.run(pretrain, executor=run.LocalExecutor())

Additionally, you can also run it directly in the same Python process as follows:

run.run(pretrain, direct=True)

A comprehensive list of pretraining recipes that we currently support or plan to support soon is provided below for reference:

Recipe	Status
Nemotron3 4B	Yes
Nemotron3 4B FP8	N/A
Nemotron3 8B	Yes
Nemotron3 8B FP8	N/A
Nemotron4 15B	Yes
Nemotron4 15B 16k	Yes
Nemotron4 15B 64k	Yes
Nemotron4 15B FP8	N/A
Nemotron4 22B	Yes
Nemotron4 22B 16k	Yes
Nemotron4 22B 64k	Yes
Nemotron4 22B FP8	N/A
Nemotron4 340B	Yes
Nemotron4 340B FP8	N/A