Moving Models CLI to GitHub Actions for Easier deployment

.github/workflows/run_models.yml

@@ -20,16 +20,33 @@ jobs:
       - name: Install dependencies
         run: |
           python -m pip install --upgrade pip
-          pip install pandas matplotlib numpy pathlib sklearn matplotlib joblib
+          pip install pandas matplotlib numpy pathlib scikit-learn matplotlib joblib
 
       - name: Run Models CLI
+        id: run_models
         run: |
-          cd models/
-          python -m scripts.run
+          cd models
+          OUTPUT=$(python -m scripts.run)
+          echo "$OUTPUT"
 
-      - name: Commit & push results
-        uses: actions-js/push@master
+          EXP_DIR=$(echo "$OUTPUT" | grep "^EXPERIMENT_DIR=" | cut -d= -f2)
+          echo "experiment_dir=$EXP_DIR" >> $GITHUB_OUTPUT
+
+      - name: Show experiment location
+        run: |
+          echo "## Models CLI Experiment Output" >> $GITHUB_STEP_SUMMARY
+          echo "" >> $GITHUB_STEP_SUMMARY
+          echo "**Contents:**" >> $GITHUB_STEP_SUMMARY
+          echo '```' >> $GITHUB_STEP_SUMMARY
+          ls -R models/experiments/${{ steps.run_models.outputs.experiment_dir }} >> $GITHUB_STEP_SUMMARY
+          echo '```' >> $GITHUB_STEP_SUMMARY
+          echo "" >> $GITHUB_STEP_SUMMARY
+          echo "You can download the experiment results as an artifact from the 'Artifacts' section below." >> $GITHUB_STEP_SUMMARY
+
+      - name: Upload experiment results
+        uses: actions/upload-artifact@v4
         with:
-          github_token: ${{ secrets.GITHUB_TOKEN }}
-          branch: ${{ github.ref_name }}
-          message: "Adding Experiment results [skip ci]"
+          name: experiment-results
+          path: models/experiments/${{ steps.run_models.outputs.experiment_dir }}
+
+

models/experiments/exp_All_v2_2026-01-17_00-29/analysis/3DBall/analysis.csv

@@ -0,0 +1,6 @@
+,Predicted 0,Predicted 1,3DBall
+Actual 0,53.0,1.0,
+Actual 1,61.0,633.0,
+Accuracy,,,0.9171122994652406
+Precision,,,0.998422712933754
+ROC AUC,,,0.946792613939588

models/experiments/exp_All_v2_2026-01-17_00-29/analysis/Basic/analysis.csv

@@ -0,0 +1,6 @@
+,Predicted 0,Predicted 1,Basic
+Actual 0,1.0,0.0,
+Actual 1,0.0,6.0,
+Accuracy,,,1.0
+Precision,,,1.0
+ROC AUC,,,1.0

models/experiments/exp_All_v2_2026-01-17_00-29/analysis/BigWallJump/analysis.csv

@@ -0,0 +1,6 @@
+,Predicted 0,Predicted 1,BigWallJump
+Actual 0,2.0,0.0,
+Actual 1,0.0,2.0,
+Accuracy,,,1.0
+Precision,,,1.0
+ROC AUC,,,1.0

models/experiments/exp_All_v2_2026-01-17_00-29/analysis/PushBlock/analysis.csv

@@ -0,0 +1,6 @@
+,Predicted 0,Predicted 1,PushBlock
+Actual 0,0.0,0.0,
+Actual 1,1.0,3.0,
+Accuracy,,,0.75
+Precision,,,1.0
+ROC AUC,,,

models/experiments/exp_All_v2_2026-01-17_00-29/analysis/metadata.json

@@ -0,0 +1,14 @@
+{
+  "name": "all on dataset",
+  "version": "v2",
+  "total_samples": 3847,
+  "samples per environment": {
+    "3DBall": 3736,
+    "Basic": 31,
+    "Hallway": 20,
+    "Crawler": 20,
+    "PushBlock": 20,
+    "BigWallJump": 20
+  },
+  "% reaching threshold": 92.77358981024176
+}

models/experiments/exp_All_v2_2026-01-17_00-29/cv/folds.csv

@@ -0,0 +1,6 @@
+fold,accuracy,f1,steps_mse,steps_mae,time_mse,time_mae
+0,0.8831168831168831,0.9326347305389222,8665995544.099108,20792.00480138349,16502.923007408288,29.220886992381736
+1,0.8831168831168831,0.9331352154531947,8010460316.885493,30778.434103350235,72074.08826083166,39.27999359152022
+2,0.8868660598179454,0.9356032568467801,1289652796.434547,18768.28678058233,2217.3410889146653,25.690729257995397
+3,0.8673602080624188,0.9235382308845578,634909461.0406405,16378.027367047129,1311.6920437203266,22.965142143900827
+4,0.8595578673602081,0.9190404797601199,1561773206.3131857,19857.363520993866,824.4540386839404,21.396068279679547

models/experiments/exp_All_v2_2026-01-17_00-29/cv/metadata.json

@@ -0,0 +1,67 @@
+[
+  {
+    "fold": 0,
+    "classifier": {
+      "accuracy": 0.8831168831168831,
+      "f1": 0.9326347305389222
+    },
+    "regression": {
+      "steps_mse": 8665995544.099108,
+      "steps_mae": 20792.00480138349,
+      "time_mse": 16502.923007408288,
+      "time_mae": 29.220886992381736
+    }
+  },
+  {
+    "fold": 1,
+    "classifier": {
+      "accuracy": 0.8831168831168831,
+      "f1": 0.9331352154531947
+    },
+    "regression": {
+      "steps_mse": 8010460316.885493,
+      "steps_mae": 30778.434103350235,
+      "time_mse": 72074.08826083166,
+      "time_mae": 39.27999359152022
+    }
+  },
+  {
+    "fold": 2,
+    "classifier": {
+      "accuracy": 0.8868660598179454,
+      "f1": 0.9356032568467801
+    },
+    "regression": {
+      "steps_mse": 1289652796.434547,
+      "steps_mae": 18768.28678058233,
+      "time_mse": 2217.3410889146653,
+      "time_mae": 25.690729257995397
+    }
+  },
+  {
+    "fold": 3,
+    "classifier": {
+      "accuracy": 0.8673602080624188,
+      "f1": 0.9235382308845578
+    },
+    "regression": {
+      "steps_mse": 634909461.0406405,
+      "steps_mae": 16378.027367047129,
+      "time_mse": 1311.6920437203266,
+      "time_mae": 22.965142143900827
+    }
+  },
+  {
+    "fold": 4,
+    "classifier": {
+      "accuracy": 0.8595578673602081,
+      "f1": 0.9190404797601199
+    },
+    "regression": {
+      "steps_mse": 1561773206.3131857,
+      "steps_mae": 19857.363520993866,
+      "time_mse": 824.4540386839404,
+      "time_mae": 21.396068279679547
+    }
+  }
+]

models/experiments/exp_All_v2_2026-01-17_00-29/metadata.json

@@ -0,0 +1,10 @@
+{
+  "name": "All on dataset",
+  "user": "runner",
+  "os": null,
+  "data_csv": "../data/normalized/normalized_v2.csv",
+  "env": "All",
+  "timestamp": "2026-01-17_00-29",
+  "version": "v2",
+  "experiment_name": "exp_All_v2_2026-01-17_00-29"
+}

models/experiments/exp_All_v2_2026-01-17_00-29/predictions/3DBall/metadata.json

@@ -0,0 +1,67 @@
+{
+  "name": "3DBall",
+  "version": 1,
+  "classifier": {
+    "n_train": 2988,
+    "n_test": 748,
+    "features": [
+      "steps",
+      "learning_rate",
+      "batch_size",
+      "buffer_size",
+      "epochs",
+      "num_agents",
+      "seed",
+      "average_cpu",
+      "average_ram",
+      "early_reward_mean",
+      "p_loss_mean",
+      "v_loss_mean",
+      "entropy_mean",
+      "algorithm"
+    ],
+    "threshold": 0.5,
+    "accuracy": 0.9171122994652406,
+    "roc_auc": 0.9802273455011208,
+    "confusion_matrix": [
+      [
+        53,
+        1
+      ],
+      [
+        61,
+        633
+      ]
+    ]
+  },
+  "regressors": {
+    "n_train_reached": 2772,
+    "n_test_reached": 694,
+    "features": [
+      "steps",
+      "learning_rate",
+      "batch_size",
+      "buffer_size",
+      "epochs",
+      "num_agents",
+      "seed",
+      "average_cpu",
+      "average_ram",
+      "early_reward_mean",
+      "p_loss_mean",
+      "v_loss_mean",
+      "entropy_mean",
+      "algorithm"
+    ],
+    "steps": {
+      "mae": 9471.143371728705,
+      "rmse": 15546.822958535264,
+      "r2": 0.7755699036683846
+    },
+    "time": {
+      "mae": 8.866254490919301,
+      "rmse": 13.189383949709963,
+      "r2": 0.796150797633937
+    }
+  }
+}