NVIDIA provides configurations for Nemotron 8B/15B/22B/340B model. The configuration includes carefully selected hyperparameters, which you may use as guidelines for any custom model configurations.
To run Nemotron training update conf/config.yaml:
defaults:
- training: nemotron/nemotron_340b
stages:
- training
Execute launcher pipeline: python3 main.py
Configuration
Default configurations for model size specific training can be found in the folder conf/training/nemotron.
The configuration is divided into four sections run, trainer, exp_manager, and model.
run:
name: nemotron_340b
results_dir: ${base_results_dir}/${.name}
time_limit: "0-04:00:00"
dependency: "singleton"
Set the number of nodes and devices for training:
trainer:
num_nodes: 768
devices: 8
max_steps: 300000 # consumed_samples = global_step * global_batch_size
max_time: "05:23:30:00" # days:hours:minutes:seconds
Set configurations for creating a checkpoint:
exp_manger:
create_checkpoint_callback: True
checkpoint_callback_params:
monitor: val_loss
save_top_k: 10
mode: min
always_save_nemo: False # saves nemo file during validation, not implemented for model parallel
save_nemo_train_end: False # not recommended when training large models on clusters with short time limits
filename: 'megatron_nemotron--{val_loss:.2f}-{step}-{consumed_amples}'
model_parallel_size: ${multiply:${training.model.tensor_model_parallel_size}, ${training.model.pipeline_model_parallel_size}}
Set wandb configurations:
exp_manager:
create_wandb_logger: True
wandb_logger_kwargs:
project: nemo_nemotron
name: ${training.run.name}
Set tensor parallel and pipeline parallel size:
model:
tensor_model_parallel_size: 8
pipeline_model_parallel_size: 12
Set data distribution configuration:
model:
data:
data_prefix:
- .0333
- ${data_dir}/my-nemotron_00_text_document
- .0333
- ${data_dir}/my-nemotron_00_text_document