Feasible Learning

The trl directory contains the code to run Orca experiments. This code is primarily built on Hugging Face's TRL library, which is why it is maintained separately from the main codebase. For additional details, please refer to the README file located within the trl directory.

Usage

CIFAR-10

python main.py \
    --data_config=configs/data.py:"dataset_type=cifar10" \
    --model_config=configs/model.py:"model_type=CIFAR_ResNet18_SGD model.init_seed=135" \
    --task_config=configs/task.py:"task_type=feasible_classification task.pointwise_probability=0.6 optim.dual_optimizer.kwargs.lr=1e-4" \
    --config.train.total_epochs=200 --config.train.seed=0 \
    --metrics_config=configs/metrics.py:"classification" \
    --resources_config=configs/resources.py:"job_type=debug" \
    --config.logging.wandb_mode="online"  --config.logging.wandb_tags="('debug','aistats')" \
    -config.logging.save_losses_and_multipliers=True --config.logging.save_eval_logs_every_n_epochs=1

UTKFace

python main.py \
    --data_config=configs/data.py:"dataset_type=utkface" \
    --model_config=configs/model.py:"model_type=UTKFace_ResNet18" \
    --task_config=configs/task.py:"task_type=feasible_regression task.pointwise_loss=0.02 optim.dual_optimizer.kwargs.lr=1e-3" \
    --config.train.total_epochs=150 --config.train.seed=0 \
    --metrics_config=configs/metrics.py:"regression" \
    --resources_config=configs/resources.py:"job_type=debug" \
    --config.logging.wandb_mode="online"  --config.logging.wandb_tags="('debug','aistats')" \
    -config.logging.save_losses_and_multipliers=True --config.logging.save_eval_logs_every_n_epochs=1

Two-Moons

python main.py \
    --data_config=configs/data.py:"dataset_type=two_moons" \
    --model_config=configs/model.py:"model_type=TwoDim_MLP optim.primal_optimizer.kwargs.lr=5e-4" \
    --task_config=configs/task.py:"task_type=feasible_classification task.pointwise_probability=0.6 optim.dual_optimizer.kwargs.lr=1e-2" \
    --config.train.total_epochs=250 --config.train.seed=0 \
    --metrics_config=configs/metrics.py:"classification" \
    --resources_config=configs/resources.py:"job_type=debug" \
    --config.logging.wandb_mode="online"  --config.logging.wandb_tags="('debug','aistats')" \
    -config.logging.save_losses_and_multipliers=True --config.logging.save_eval_logs_every_n_epochs=1

Note that the parsed text arguments for file-based configs need to point to the "full path" of the configuration item. For example, if you want to change restart_on_feasible in config.task.multiplier_kwargs, you need to use task.multiplier_kwargs.restart_on_feasible=BOOL (and NOT multiplier_kwargs.restart_on_feasible=BOOL).

Cluster	Is local?	Is SLURM?	Is interactive?	Is background?	Notes
debug	✅	🚫	✅	🚫	For local execution on your machine or on a compute node
local	✅	🚫	🚫	✅	This could be locally on your machine or on a compute node
main	🚫	✅	🚫	✅	To run jobs on SLURM main partition
long	🚫	✅	🚫	✅	To run jobs on SLURM long partition

• Keep in mind that you can run the code from a compute node on SLURM via debug or local.
• Currently, using multiple GPUs is not supported.

Required enviroment variables

We use dotenv to manage environment variables. Please create a .env file in the root directory of the project and add the following variables:

# Location of the directory containing your datasets
DATA_DIR=

# The directory where the results will be saved
CHECKPOINT_DIR=

# If you want to use Weights & Biases, add the entity name here
WANDB_ENTITY=

WANDB_PROJECT=
# Directory for Weights & Biases local storage
WANDB_DIR=

# Directory for logs created by submitit
SUBMITIT_DIR=

KAGGLE_USERNAME=
KAGGLE_KEY=

Code style

• This project uses black, isort, and flake8 for enforcing code style. See requirements.txt for version numbers.
• We use pre-commit hooks to ensure that all code committed respects the code style.
• After (1) cloning the repo, (2) creating your environment and (3) installing the required packages, you are strongly encouraged to run pre-commit install to set-up pre-commit hooks.

Logging format

Whenever you are using logging inside a module, please remember to use the rich formatting.

Do NOT do this:

import logging
logger = logging.getLogger(__name__)

DO this instead:

import shared
logger = shared.fetch_main_logger()