roll back to autogluon 1.3.0 due to stabiulity issue

modules/ml-pipeline/src/pipeline/configs/build_model.yaml

@@ -19,5 +19,6 @@ default:
       excluded_model_types: ['RF', 'CAT', 'NN_TORCH', 'KNN', 'XT']
       infer_limit: 0.0005
       infer_limit_batch_size: 10000
-      "fit_strategy": "parallel"
+      fit_strategy: "parallel"
       ag_args_ensemble: {'num_folds_parallel': 2}
+      num_gpus: auto

modules/ml-pipeline/src/pipeline/core/MLModels.py

@@ -153,6 +153,7 @@ class AutogluonAutoML:
         "infer_limit_batch_size",
         "ag_args_ensemble",
         "fit_strategy",
+        "num_gpus",
     ]
 
     def load_model(self, path: Union[Path, str]) -> None:
@@ -184,6 +185,8 @@ class AutogluonAutoML:
         """
         Method to train a model
         """
+        # Force Parallel Model fitting
+        os.environ["AG_FORCE_PARALLEL"] = "True"
 
         validate_dict_keys(
             keys_1=list(model_hyperparameters.keys()),
@@ -211,6 +214,7 @@ class AutogluonAutoML:
             infer_limit_batch_size=model_hyperparameters["infer_limit_batch_size"],
             ag_args_ensemble=model_hyperparameters["ag_args_ensemble"],
             fit_strategy=model_hyperparameters["fit_strategy"],
+            num_gpus=model_hyperparameters["num_gpus"],
         )
 
     def predict(

modules/ml-pipeline/src/pipeline/dvc.lock

@@ -61,8 +61,8 @@ stages:
     outs:
     - path: data/prepared_data/
       hash: md5
-      md5: 2feba8772c240b507eb900934efcb8ca.dir
-      size: 46064555
+      md5: 7b780ea01da913d9d8cadcff73fbde0f.dir
+      size: 46092230
       nfiles: 3
   build_model:
     cmd: python 2_build_model.py
@@ -73,8 +73,8 @@ stages:
       size: 4820
     - path: data/prepared_data
       hash: md5
-      md5: 2feba8772c240b507eb900934efcb8ca.dir
-      size: 46064555
+      md5: 7b780ea01da913d9d8cadcff73fbde0f.dir
+      size: 46092230
       nfiles: 3
     params:
       configs/build_model.yaml:
@@ -104,20 +104,21 @@ stages:
               fit_strategy: parallel
               ag_args_ensemble:
                 num_folds_parallel: 2
+              num_gpus: auto
     outs:
     - path: data/fit_predictions/
       hash: md5
-      md5: 29036f4f42b1fdcab7f9e40a87f38a8c.dir
-      size: 3474783
+      md5: 01328a1cc5a1ff35e701a3c44902afc6.dir
+      size: 3474659
       nfiles: 1
     - path: data/model/
       hash: md5
-      md5: 77cab231e3d51bbebbae5a7af310c18a.dir
-      size: 791390619
-      nfiles: 34
+      md5: 70f076a248524dfce60412f83969ae63.dir
+      size: 760254863
+      nfiles: 33
     - path: metrics/fit_metrics.json
       hash: md5
-      md5: 4f39064fb6b31c7c879299621bcea28d
+      md5: 4726c52b2f27650ab1bbf97b5bf61e54
       size: 224
   generate_predictions:
     cmd: python 3_generate_predictions.py
@@ -128,13 +129,13 @@ stages:
       size: 2464
     - path: data/model
       hash: md5
-      md5: 77cab231e3d51bbebbae5a7af310c18a.dir
-      size: 791390619
-      nfiles: 34
+      md5: 70f076a248524dfce60412f83969ae63.dir
+      size: 760254863
+      nfiles: 33
     - path: data/prepared_data
       hash: md5
-      md5: 2feba8772c240b507eb900934efcb8ca.dir
-      size: 46064555
+      md5: 7b780ea01da913d9d8cadcff73fbde0f.dir
+      size: 46092230
       nfiles: 3
     params:
       configs/settings.yaml:
@@ -148,8 +149,8 @@ stages:
     outs:
     - path: data/predictions/
       hash: md5
-      md5: 8dfa69b48586da6b0ef33a6fbedb7c4a.dir
-      size: 484314
+      md5: 312f9106eb18d34df75124f0536f0603.dir
+      size: 484470
       nfiles: 1
   generate_metrics:
     cmd: python 4_generate_metrics.py
@@ -160,13 +161,13 @@ stages:
       size: 3484
     - path: data/predictions
       hash: md5
-      md5: 8dfa69b48586da6b0ef33a6fbedb7c4a.dir
-      size: 484314
+      md5: 312f9106eb18d34df75124f0536f0603.dir
+      size: 484470
       nfiles: 1
     - path: data/prepared_data
       hash: md5
-      md5: 2feba8772c240b507eb900934efcb8ca.dir
-      size: 46064555
+      md5: 7b780ea01da913d9d8cadcff73fbde0f.dir
+      size: 46092230
       nfiles: 3
     params:
       configs/settings.yaml:
@@ -176,8 +177,8 @@ stages:
     outs:
     - path: metrics/metrics.json
       hash: md5
-      md5: bf980dad2dc5b97651546b0b755419ae
-      size: 223
+      md5: 661388682aa1ca888b256e4667211379
+      size: 222
   generate_scenerio_metrics:
     cmd: python 5_generate_scenarios.py
     deps:
@@ -197,9 +198,9 @@ stages:
     outs:
     - path: metrics/scenario_metrics.md
       hash: md5
-      md5: 05e2cce8e61d5005398659e9f3465cd6
+      md5: 88ebca8dccf907692675301ffe06b10d
       size: 356
     - path: metrics/scenario_table.md
       hash: md5
-      md5: 92446d2f3836c6f790d06e3b268b05f3
+      md5: 3ec419e883b812b254b331f055999cc9
       size: 872

modules/ml-pipeline/src/pipeline/requirements/predictions/requirements-dev.txt

@@ -1,7 +1,7 @@
 joblib==1.5.2
 boto3==1.40.61
 pandas==2.2.3
-autogluon.tabular[all]==1.4
+autogluon.tabular[all]==1.3
 dynaconf==3.2.12
 pyarrow==20.0.0
 pre-commit==4.3.0

modules/ml-pipeline/src/pipeline/requirements/predictions/requirements.txt

@@ -1,7 +1,7 @@
 joblib==1.5.2
 boto3==1.40.61
 pandas==2.2.3
-autogluon.tabular[all]==1.4
+autogluon.tabular[all]==1.3
 dynaconf==3.2.12
 pyarrow==20.0.0
 PyYAML==6.0.3

modules/ml-pipeline/src/pipeline/requirements/training/requirements-dev.txt

@@ -1,10 +1,10 @@
 joblib==1.5.2
 boto3==1.40.61
 pandas==2.2.3
-autogluon.tabular[all]==1.4
+autogluon.tabular[all]==1.3
 ray==2.44.1
 dynaconf==3.2.12
 alibi==0.5.5
 shap==0.49.1
-pyarrow
+pyarrow==20.0.0
 pre-commit==4.3.0

modules/ml-pipeline/src/pipeline/requirements/training/requirements.txt

@@ -1,4 +1,4 @@
 boto3==1.40.61
 pandas==2.2.3
-autogluon.tabular[all]==1.4
+autogluon.tabular[all]==1.3
 dynaconf==3.2.12