Merge pull request #63 from NSAPH-Projects/automate-script-creation

Automate script creation

Author: ekarais

environment.yaml

@@ -9,6 +9,7 @@ channels:
 dependencies:
   - coverage=7.4.3
   - gudhi=3.8.0
+  - jinja2=3.1.4
   - matplotlib=3.7.2
   - networkx=3.1
   - numpy=1.22.4

scripts/experiments/9.sh

@@ -0,0 +1,13 @@
+experiment_name: "experiment_1"
+description: "isolated ranks: atoms + bonds"
+
+# Dataset args
+lifters: ["atom:0", "bond:1", "supercell:2"]
+neighbor_types: "max"
+connectivity: "self"
+visible_dims: [0, 1]
+initial_features: "hetero"
+dim: 2
+
+# Model args
+num_hidden: 128

scripts/experiments/colab_parallel.ipynb

@@ -1,26 +1,19 @@
 #!/bin/bash
 
-# isolated ranks: atoms + bonds
+# Dataset generation script for experiment_1
 
 # DEFINE EXP ARGUMENTS
-LIFTERS=("atom:0" "bond:1" "supercell:2")
+LIFTERS=(atom:0 bond:1 supercell:2)
 NEIGHBOR_TYPES="max"
 CONNECTIVITY="self"
 VISIBLE_DIMS=(0 1)
 INITIAL_FEATURES="hetero"
 DIM=2
 
-
-# Train EGNN in parallel for each TARGET_NAME
+# Command to generate dataset
 python src/create_dataset.py --lifters "${LIFTERS[@]}" \
                              --neighbor_types "$NEIGHBOR_TYPES" \
                              --connectivity "$CONNECTIVITY" \
                              --visible_dims "${VISIBLE_DIMS[@]}" \
                              --initial_features "$INITIAL_FEATURES" \
-                             --dim "$DIM" \
-                       
-                        
-                        
-                        
-
-
+                             --dim "$DIM"

scripts/experiments/experiment_1/config.yaml

@@ -1,18 +1,19 @@
 #!/bin/bash
 
-# isolated ranks: atoms + bonds
+# Training script for experiment_1
 
 # DEFINE EXP ARGUMENTS
-LIFTERS=("atom:0" "bond:1" "supercell:2")
+EXP_NAME=experiment_1
+LIFTERS=(atom:0 bond:1 supercell:2)
 DIM=2
 VISIBLE_DIMS=(0 1)
 INITIAL_FEATURES="hetero"
 NEIGHBOR_TYPES="max"
 CONNECTIVITY="self"
-EPOCHS=350
 NUM_HIDDEN=128
 
 # Constants
+EPOCHS=350
 BATCH_SIZE=96
 WEIGHT_DECAY=1e-16
 MIN_LR=0
@@ -21,7 +22,6 @@ MODEL_NAME="ten"
 SPLITS="egnn"
 CHECKPOINT_DIR="checkpoints/"
 
-
 # Define the predefined target names lists
 TARGET_NAMES_0=("alpha")
 TARGET_NAMES_1=("mu" "alpha" "homo" "lumo" "gap" "r2")
@@ -50,7 +50,7 @@ esac
 for TARGET_NAME in "${TARGET_NAMES[@]}"
 do
     # Check if target is one of the first three
-    if [[ "$target" == "gap" || "$target" == "homo" || "$target" == "lumo" ]]; then
+    if [[ "$TARGET_NAME" == "gap" || "$TARGET_NAME" == "homo" || "$TARGET_NAME" == "lumo" ]]; then
         LR="1e-3"
     else
         LR="5e-4"
@@ -77,6 +77,7 @@ do
                            --splits "$SPLITS" \
                            --normalize_invariants \
                            --clip_gradient \
+                           --run_name "${EXP_NAME} ${TARGET_NAME}" \
                            --checkpoint_dir "$CHECKPOINT_DIR" &
 
     # Wait for 30 minutes before moving to the next iteration
@@ -86,5 +87,4 @@ done
 # Wait for all background jobs to finish
 echo "All jobs have been launched."
 wait
-echo "All jobs have been completed."
-
+echo "All jobs have been completed."

scripts/experiments/1_create_dataset.sh renamed to scripts/experiments/experiment_1/generate_dataset.sh

@@ -0,0 +1,13 @@
+experiment_name: "experiment_10"
+description: "EGNN-like"
+
+# Dataset args
+lifters: ["atom:0", "supercell:1"]
+neighbor_types: "max"
+connectivity: "self"
+visible_dims: [0]
+initial_features: "hetero"
+dim: 1
+
+# Model args
+num_hidden: 128

scripts/experiments/1.sh renamed to scripts/experiments/experiment_1/train_etnn.sh

@@ -0,0 +1,19 @@
+#!/bin/bash
+
+# Dataset generation script for experiment_10
+
+# DEFINE EXP ARGUMENTS
+LIFTERS=(atom:0 supercell:1)
+NEIGHBOR_TYPES="max"
+CONNECTIVITY="self"
+VISIBLE_DIMS=(0)
+INITIAL_FEATURES="hetero"
+DIM=1
+
+# Command to generate dataset
+python src/create_dataset.py --lifters "${LIFTERS[@]}" \
+                             --neighbor_types "$NEIGHBOR_TYPES" \
+                             --connectivity "$CONNECTIVITY" \
+                             --visible_dims "${VISIBLE_DIMS[@]}" \
+                             --initial_features "$INITIAL_FEATURES" \
+                             --dim "$DIM"

scripts/experiments/experiment_10/config.yaml

@@ -1,18 +1,19 @@
 #!/bin/bash
 
-# EGNN-like, no VC
+# Training script for experiment_10
 
 # DEFINE EXP ARGUMENTS
-LIFTERS=("atom:0" "bond:1")
+EXP_NAME=experiment_10
+LIFTERS=(atom:0 supercell:1)
 DIM=1
 VISIBLE_DIMS=(0)
 INITIAL_FEATURES="hetero"
 NEIGHBOR_TYPES="max"
 CONNECTIVITY="self"
-EPOCHS=350
 NUM_HIDDEN=128
 
 # Constants
+EPOCHS=350
 BATCH_SIZE=96
 WEIGHT_DECAY=1e-16
 MIN_LR=0
@@ -21,7 +22,6 @@ MODEL_NAME="ten"
 SPLITS="egnn"
 CHECKPOINT_DIR="checkpoints/"
 
-
 # Define the predefined target names lists
 TARGET_NAMES_0=("alpha")
 TARGET_NAMES_1=("mu" "alpha" "homo" "lumo" "gap" "r2")
@@ -50,7 +50,7 @@ esac
 for TARGET_NAME in "${TARGET_NAMES[@]}"
 do
     # Check if target is one of the first three
-    if [[ "$target" == "gap" || "$target" == "homo" || "$target" == "lumo" ]]; then
+    if [[ "$TARGET_NAME" == "gap" || "$TARGET_NAME" == "homo" || "$TARGET_NAME" == "lumo" ]]; then
         LR="1e-3"
     else
         LR="5e-4"
@@ -77,14 +77,14 @@ do
                            --splits "$SPLITS" \
                            --normalize_invariants \
                            --clip_gradient \
+                           --run_name "${EXP_NAME} ${TARGET_NAME}" \
                            --checkpoint_dir "$CHECKPOINT_DIR" &
-
+    
     # Wait for 30 minutes before moving to the next iteration
     sleep 1800
 done
 
 # Wait for all background jobs to finish
 echo "All jobs have been launched."
 wait
-echo "All jobs have been completed."
-
+echo "All jobs have been completed."

scripts/experiments/experiment_10/generate_dataset.sh

@@ -0,0 +1,13 @@
+experiment_name: "experiment_11"
+description: "EGNN-Wide"
+
+# Dataset args
+lifters: ["atom:0", "supercell:1"]
+neighbor_types: "max"
+connectivity: "self"
+visible_dims: [0]
+initial_features: "hetero"
+dim: 1
+
+# Model args
+num_hidden: 182