Supervised Fine-tuning (SFT)

Please prepare the datasets according to Data Preparation for SFT and PEFT section before proceeding.

Step 1: Start NeMo Container

If the container is not already running use the following command

Copy
Copied!
            

docker run %%gpus device=1 --shm-size=2g --net=host --ulimit memlock=-1 --rm -it -v ${PWD}:/workspace -w /workspace -v ${PWD}/results:/results nvcr.io/nvidia/nemo:24.01.gemma bash

Step 2: Run SFT

Set the environment variables, pass the paths to your train, test, and validation data files

Copy
Copied!
            

MODEL="YOUR PATH TO gemma-7b.nemo" TRAIN="[YOUR PATH TO databricks-dolly-15k/train.jsonl]" VALID="[YOUR PATH TO databricks-dolly-15k/validation.jsonl]" TEST="[YOUR PATH TO databricks-dolly-15k/test.jsonl]" VALID_NAMES="[databricks-dolly-15k]"

Set the concat sampling probability. This depends on the number of files being passed in the train set and how much percentage of the fine tuning data would you like to use from each file. Note sum of concat sampling probabilities should be 1.0. For example, the following is an example for setting concat sampling probability for a train set with 2 jsonl files.

Copy
Copied!
            

TRAIN="[/path/to/dataset_1.jsonl,/path/to/dataset_2.jsonl]" CONCAT_SAMPLING_PROBS="[0.3,0.7]"

In our example we are using 1 train file so CONCAT_SAMPLING_PROBS="[1.0]" Set the tensor parallelism and pipeline parallelism values based on the model you are using.

Copy
Copied!
            

CONCAT_SAMPLING_PROBS="[1]" TP_SIZE=2 PP_SIZE=1

Run the SFT command by appropriately setting the values for the parameters such as the number of steps, model checkpoint path, batch sizes etc. For a full reference of parameter settings refer to the config file

Copy
Copied!
            

torchrun --nproc_per_node=8 \ /opt/NeMo/examples/nlp/language_modeling/tuning/megatron_gpt_finetuning.py \ trainer.precision=bf16 \ trainer.devices=8 \ trainer.num_nodes=1 \ trainer.val_check_interval=0.1 \ trainer.max_steps=50 \ model.restore_from_path=${MODEL} \ model.peft.peft_scheme=none \ model.micro_batch_size=1 \ model.global_batch_size=128 \ model.tensor_model_parallel_size=${TP_SIZE} \ model.pipeline_model_parallel_size=${PP_SIZE} \ model.megatron_amp_O2=True \ model.sequence_parallel=True \ model.activations_checkpoint_granularity=selective \ model.activations_checkpoint_method=uniform \ model.optim.name=distributed_fused_adam \ model.optim.lr=5e-6 \ model.answer_only_loss=True \ model.data.train_ds.file_names=${TRAIN_DS} \ model.data.validation_ds.file_names=${VALID_DS} \ model.data.test_ds.file_names=${TEST_DS} \ model.data.train_ds.concat_sampling_probabilities=${CONCAT_SAMPLING_PROBS} \ model.data.train_ds.max_seq_length=4096 \ model.data.validation_ds.max_seq_length=4096 \ model.data.train_ds.micro_batch_size=1 \ model.data.train_ds.global_batch_size=128 \ model.data.validation_ds.micro_batch_size=1 \ model.data.validation_ds.global_batch_size=128 \ model.data.test_ds.micro_batch_size=1 \ model.data.test_ds.global_batch_size=256 \ model.data.train_ds.num_workers=0 \ model.data.validation_ds.num_workers=0 \ model.data.test_ds.num_workers=0 \ model.data.validation_ds.metric.name=loss \ model.data.test_ds.metric.name=loss \ exp_manager.create_wandb_logger=False \ exp_manager.explicit_log_dir=/results \ exp_manager.resume_if_exists=True \ exp_manager.resume_ignore_no_checkpoint=True \ exp_manager.create_checkpoint_callback=True \ exp_manager.checkpoint_callback_params.monitor=validation_loss \ exp_manager.checkpoint_callback_params.save_best_model=False \ exp_manager.checkpoint_callback_params.save_nemo_on_train_end=True \ ++cluster_type=BCP \ model.sequence_parallel=True \ ++model.bias_activation_fusion=True \ ++model.apply_rope_fusion=True \ ++model.optim.overlap_grad_sync=True \ ++model.optim.overlap_param_sync=True \ ++model.optim.contiguous_grad_buffer=True \ ++model.optim.grad_sync_dtype=bf16 \ ++model.fp8=False \ ++model.fp8_e4m3=False \ ++model.fp8_hybrid=True \ ++model.fp8_margin=0 \ ++model.fp8_interval=1 \ ++model.fp8_amax_history_len=128 \ ++model.fp8_amax_compute_algo=max

Note: For running SFT on multiple nodes (for example on a Slurm cluster, replace the torchrun --nproc_per_node=8 with python.

Tuning with packed dataset: Enable training with packed sequences by adjusting configs. We also need to reduce both micro batch size and global batch size due to packing. Here we set global_batch_size=1 with sequence length 4096:

Copy
Copied!
            

model.data.train_ds.file_names=/path/to/dolly/packed_4096_seed0.npy \ +model.data.train_ds.packed_sequence=True \ model.micro_batch_size=1 \ model.global_batch_size=8

Tuning with FP8: Enable training with FP8 by adjusting configs:

Copy
Copied!
            

++model.fp8=True

Step 3: Run evaluation

Run evaluation using megatron_gpt_peft_eval.py

Set the appropriate model checkpoint path, test file path, batch sizes, number of tokens etc. and run evaluation on the test file

Copy
Copied!
            

PATH_TO_TRAINED_MODEL=/results/megatron_gpt_peft_none_tuning/checkpoints/megatron_gpt_peft_none_tuning.nemo TEST_DS="[YOUR PATH TO test.jsonl]" python /opt/NeMo/examples/nlp/language_modeling/tuning/megatron_gpt_peft_eval.py \ model.restore_from_path=${PATH_TO_TRAINED_MODEL} \ trainer.devices=8 \ model.data.test_ds.file_names=${TEST_DS} \ model.data.test_ds.names=['dolly-15k_test'] \ model.data.test_ds.global_batch_size=16 \ model.data.test_ds.micro_batch_size=2 \ model.data.test_ds.tokens_to_generate=20 \ model.tensor_model_parallel_size=1 \ model.pipeline_model_parallel_size=1 \ inference.greedy=True \ model.data.test_ds.output_file_path_prefix=/results/sft_results \ model.data.test_ds.write_predictions_to_file=True

Sample Output

Copy
Copied!
            

$ tail -n 4 sft_results.jsonl {"sentence": "What is Azure HDInsight? Azure HDInsight is a cloud service that provides a high-performance, scalable, and cost-effective way to run Apache Hadoop on the"} {"sentence": "What is carnitine? Carnitine is a fatty acid that is found in the body. It is used to produce energy in the mitochondria of the cells. Carnit"} {"sentence": "List some TV shows that Canadian actor William B. Davis has been in."} {"sentence": "Identify which instrument is string or percussion: Handbell, Dobro, Drum"}

Note, This is only a sample output (based of a toy SFT example) and your output may vary. The performance can be further improved by fine tuning the model for more steps.

Please refer to Launcher Guide section to understand the NeMo Launcher basics. To run SFT update conf/config.yaml:

Copy
Copied!
            

defaults: - peft: gemma/sft stages: - peft

Execute launcher pipeline: python3 main.py

Configuration

Default configurations for PEFT with squad can be found in conf/peft/gemma/squad.yaml. Fine-tuning configuration is divided into four sections run, trainer, exp_manger and model.

Copy
Copied!
            

run: name: sft_gemma_7b time_limit: "04:00:00" dependency: "singleton" model_train_name: gemma_7b convert_dir: ${base_results_dir}/${peft.run.model_train_name}/${peft.run.convert_name} task_name: "squad" results_dir: ${base_results_dir}/${.model_train_name}/peft_${.task_name}

Set the number of nodes and devices for fine-tuning:

Copy
Copied!
            

trainer: num_nodes: 1 devices: 8

Copy
Copied!
            

model: restore_from_path: /path/to/gemma-7b.nemo

restore_from_path sets the path to the .nemo checkpoint to run fine-tuning.

Previous Checkpoint Conversion
Next Parameter Efficient Fine-Tuning (PEFT)
© Copyright 2023-2024, NVIDIA. Last updated on Feb 22, 2024.