updating FSDP slurm documentation #745
Open
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
mhuguesaws
reviewed
Jun 15, 2025
|
|
||
| ```bash | ||
| . ./create_venv.sh | ||
| bash create_venv.sh |
There was a problem hiding this comment.
Please the venv creation command as it is.
mhuguesaws
requested changes
Jun 15, 2025
| * By creating this environment on the shared FSx for Lustre volume, all compute nodes in our cluster will have access to it. | ||
|
|
||
| ### Option 2: Build a container image and the squash file | ||
|
|
||
| We first build the container image with the command below: |
There was a problem hiding this comment.
Suggested change
| We first build the container image with the command below: | |
| You will first build the container image with the command below: |
| export AWS_REGION=$(aws ec2 describe-availability-zones --output text --query 'AvailabilityZones[0].[RegionName]') | ||
| export ACCOUNT=$(aws sts get-caller-identity --query Account --output text) | ||
| export REGISTRY=${ACCOUNT}.dkr.ecr.${AWS_REGION}.amazonaws.com/ | ||
| pushd ../ |
There was a problem hiding this comment.
Suggested change
| pushd ../ |
| export REGISTRY=${ACCOUNT}.dkr.ecr.${AWS_REGION}.amazonaws.com/ | ||
| pushd ../ | ||
| docker build -f Dockerfile -t ${REGISTRY}fsdp:pytorch2.7.1 . | ||
| popd |
| export ACCOUNT=$(aws sts get-caller-identity --query Account --output text) | ||
| export REGISTRY=${ACCOUNT}.dkr.ecr.${AWS_REGION}.amazonaws.com/ | ||
| pushd ../ | ||
| docker build -f Dockerfile -t ${REGISTRY}fsdp:pytorch2.7.1 . |
There was a problem hiding this comment.
Suggested change
| docker build -f Dockerfile -t ${REGISTRY}fsdp:pytorch2.7.1 . | |
| docker build -f ../Dockerfile -t ${REGISTRY}fsdp:pytorch2.7.1 . |
| ... | ||
| 1: p5-dy-gpu-2:62571:62571 [1] NCCL INFO NCCL version 2.26.2+cuda12.2 |
There was a problem hiding this comment.
Please keep those information to have trace of what version were using to display those numbers.
|
|
||
| ```text | ||
| + TORCHRUN_ARGS=('--nproc_per_node=8' '--nnodes=4' '--rdzv_id=2513' '--rdzv_backend=c10d' '--rdzv_endpoint=p5-dy-gpu-1') | ||
| + TORCHRUN_ARGS=('--nproc_per_node=8' '--nnodes=2' '--rdzv_id=98' '--rdzv_backend=c10d' '--rdzv_endpoint=p5-dy-gpu-1') |
There was a problem hiding this comment.
Suggested change
| + TORCHRUN_ARGS=('--nproc_per_node=8' '--nnodes=2' '--rdzv_id=98' '--rdzv_backend=c10d' '--rdzv_endpoint=p5-dy-gpu-1') | |
| + TORCHRUN_ARGS=('--nproc_per_node=8' '--nnodes=4' '--rdzv_id=98' '--rdzv_backend=c10d' '--rdzv_endpoint=p5-dy-gpu-1') |
| + TRAINING_ARGS=('--max_context_width=4096' '--num_key_value_heads=32' '--intermediate_size=11008' '--hidden_width=4096' '--num_layers=32' '--num_heads=32' '--model_type=llama_v2' '--tokenizer=hf-internal-testing/llama-tokenizer' '--checkpoint_freq=5000' '--validation_freq=500' '--max_steps=5000' '--checkpoint_dir=./checkpoints' '--dataset=c4' '--dataset_config_name=en' '--resume_from_checkpoint=./checkpoints' '--train_batch_size=1' '--val_batch_size=1' '--sharding_strategy=full' '--offload_activations=1') | ||
| + export TRAIN_SCRIPT=../src/train.py | ||
| + TRAIN_SCRIPT=../src/train.py | ||
| + TRAINING_ARGS=('--max_context_width=128256' '--num_key_value_heads=8' '--intermediate_size=14336' '--hidden_width=4096' '--num_layers=32' '--num_heads=32' '--model_type=llama_v3' '--tokenizer=hf-internal-testing/llama-tokenizer' '--checkpoint_freq=5000' '--validation_freq=500' '--max_steps=5000' '--checkpoint_dir=./checkpoints' '--dataset=allenai/c4' '--dataset_config_name=en' '--resume_from_checkpoint=./checkpoints' '--train_batch_size=1' '--val_batch_size=1' '--sharding_strategy=full' '--offload_activations=1') |
There was a problem hiding this comment.
Suggested change
| + TRAINING_ARGS=('--max_context_width=128256' '--num_key_value_heads=8' '--intermediate_size=14336' '--hidden_width=4096' '--num_layers=32' '--num_heads=32' '--model_type=llama_v3' '--tokenizer=hf-internal-testing/llama-tokenizer' '--checkpoint_freq=5000' '--validation_freq=500' '--max_steps=5000' '--checkpoint_dir=./checkpoints' '--dataset=allenai/c4' '--dataset_config_name=en' '--resume_from_checkpoint=./checkpoints' '--train_batch_size=1' '--val_batch_size=1' '--sharding_strategy=full' '--offload_activations=1') | |
| + TRAINING_ARGS=('--max_context_width=128256' '--num_key_value_heads=8' '--intermediate_size=14336' '--hidden_width=4096' '--num_layers=32' '--num_heads=32' '--model_type=llama_v3' '--tokenizer=hf-internal-testing/llama-tokenizer' '--checkpoint_freq=50' '--validation_freq=25' '--max_steps=100' '--checkpoint_dir=./checkpoints' '--dataset=allenai/c4' '--dataset_config_name=en' '--resume_from_checkpoint=./checkpoints' '--train_batch_size=1' '--val_batch_size=1' '--sharding_strategy=full' '--offload_activations=1') |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Issue #, if available:
Description of changes:
updating FSDP slurm documentation to add instructions on using container image and adding Llama 3 models.
By submitting this pull request, I confirm that you can use, modify, copy, and redistribute this contribution, under the terms of your choice.