seperate dmc setup...

2025-07-29 14:58:43 +02:00
parent 30df18239a
commit a02e258f1c
3 changed files with 209 additions and 0 deletions
@@ -0,0 +1,55 @@
+#!/bin/bash
+#SBATCH --job-name=reppo_dmc_dev
+#SBATCH --account=hk-project-p0022232
+#SBATCH --partition=dev_accelerated
+#SBATCH --gres=gpu:1
+#SBATCH --nodes=1
+#SBATCH --ntasks-per-node=1
+#SBATCH --cpus-per-task=4
+#SBATCH --time=00:30:00
+#SBATCH --mem=16G
+#SBATCH --output=logs/reppo_dmc_dev_%j.out
+#SBATCH --error=logs/reppo_dmc_dev_%j.err
+
+# Load required modules
+module load devel/cuda/12.4
+
+# Set environment variables
+export WANDB_MODE=online
+export WANDB_PROJECT=reppo_dmc_dev
+export WANDB_API_KEY=01fbfaf5e2f64bedd68febedfcaa7e3bbd54952c
+export WANDB_ENTITY=dominik_roth
+
+# Change to project directory
+cd /hkfs/home/project/hk-project-robolear/ys1087/Projects/reppo
+
+# Activate virtual environment
+source .venv/bin/activate
+
+# Run DMC (mujoco_playground) test
+echo "Starting REPPO dev test with DMC..."
+echo "Job ID: $SLURM_JOB_ID"
+echo "Node: $SLURM_NODELIST"
+echo "GPU: $CUDA_VISIBLE_DEVICES"
+
+# Default to CartpoleBalance for quick test
+ENV_NAME=${ENV_NAME:-CartpoleBalance}
+EXPERIMENT_TYPE=${EXPERIMENT_TYPE:-mjx_dmc_small_data}
+
+echo "Environment: $ENV_NAME"
+echo "Experiment type: $EXPERIMENT_TYPE"
+
+# Run the experiment with mjx_dmc environment
+python reppo_alg/jaxrl/reppo.py \
+    env=mjx_dmc \
+    env.name=$ENV_NAME \
+    hyperparameters.num_envs=256 \
+    hyperparameters.num_steps=32 \
+    hyperparameters.num_mini_batches=8 \
+    hyperparameters.num_epochs=4 \
+    hyperparameters.total_time_steps=1000000 \
+    wandb.mode=online \
+    wandb.entity=$WANDB_ENTITY \
+    wandb.project=$WANDB_PROJECT
+
+echo "DMC dev test completed!"
@@ -0,0 +1,56 @@
+#!/bin/bash
+#SBATCH --job-name=reppo_dmc_prod
+#SBATCH --account=hk-project-p0022232
+#SBATCH --partition=accelerated
+#SBATCH --gres=gpu:1
+#SBATCH --nodes=1
+#SBATCH --ntasks-per-node=1
+#SBATCH --cpus-per-task=8
+#SBATCH --time=24:00:00
+#SBATCH --mem=32G
+#SBATCH --output=logs/reppo_dmc_prod_%j.out
+#SBATCH --error=logs/reppo_dmc_prod_%j.err
+
+# Load required modules
+module load devel/cuda/12.4
+
+# Set environment variables
+export WANDB_MODE=online
+export WANDB_PROJECT=reppo_dmc_production
+export WANDB_API_KEY=01fbfaf5e2f64bedd68febedfcaa7e3bbd54952c
+export WANDB_ENTITY=dominik_roth
+
+# Change to project directory
+cd /hkfs/home/project/hk-project-robolear/ys1087/Projects/reppo
+
+# Activate virtual environment
+source .venv/bin/activate
+
+# Run DMC experiment
+echo "Starting REPPO production run with DMC..."
+echo "Job ID: $SLURM_JOB_ID"
+echo "Node: $SLURM_NODELIST"
+echo "GPU: $CUDA_VISIBLE_DEVICES"
+
+# Environment name passed as variable
+ENV_NAME=${ENV_NAME:-CartpoleBalance}
+SEED=${SEED:-0}
+
+echo "Environment: $ENV_NAME"
+echo "Seed: $SEED"
+
+# Run the experiment with full 50M steps
+python reppo_alg/jaxrl/reppo.py \
+    env=mjx_dmc \
+    env.name=$ENV_NAME \
+    hyperparameters.num_envs=1024 \
+    hyperparameters.num_steps=128 \
+    hyperparameters.num_mini_batches=128 \
+    hyperparameters.num_epochs=4 \
+    hyperparameters.total_time_steps=50000000 \
+    seed=$SEED \
+    wandb.mode=online \
+    wandb.entity=$WANDB_ENTITY \
+    wandb.project=$WANDB_PROJECT
+
+echo "Training completed!"