remove debug info

Author: lijinf2

python/tests_large/test_large_logistic_regression.py

@@ -119,84 +119,3 @@ def get_nnz_func(pdf_iter: Iterable[pd.DataFrame]) -> Iterable[pd.DataFrame]:
 
 def test_sparse_int64_mg() -> None:
     test_sparse_int64(multi_gpus=True)
-
-
-def test_log_reg_sparsevector_int64_sparkrapidsml() -> None:
-    elasticNetParam=1.0
-    gpu_number = 2
-    """
-    This test requires minimum 128G CPU memory, 32 GB GPU memory
-    Only pass in x2 GPU, each one has >=32GB and CPU memory >=256GB environment
-    """
-    # Configuration for the 2.2GB dataset
-    output_num_files = 100  # large value smaller CPU memory for each spark task
-    data_shape = (int(1e7), 2200)
-    #data_shape = (int(1e5), 2200)
-
-    fraction_sampled_for_test = (
-        1.0 if data_shape[0] <= 100000 else 100000 / data_shape[0]
-    )
-    n_classes = 8
-    tolerance = 0.001
-    est_params: Dict[str, Any] = {
-        "regParam": 0.02,
-        "maxIter": 10,
-        "standardization": False,  # reduce GPU memory since standardization copies the value array
-    }
-    density = 0.1
-
-    data_gen_args = [
-        "--n_informative",
-        f"{math.ceil(data_shape[1] / 3)}",
-        "--num_rows",
-        str(data_shape[0]),
-        "--num_cols",
-        str(data_shape[1]),
-        "--output_num_files",
-        str(output_num_files),
-        "--dtype",
-        "float32",
-        "--feature_type",
-        "vector",
-        "--output_dir",
-        "./temp",
-        "--n_classes",
-        str(n_classes),
-        "--random_state",
-        "0",
-        "--logistic_regression",
-        "True",
-        "--density",
-        str(density),
-        "--use_gpu",
-        "True",
-    ]
-
-    data_gen = SparseRegressionDataGen(data_gen_args)
-    df, _, _ = data_gen.gen_dataframe_and_meta(_spark)
-
-    df = df.cache()
-    df_gpu = df
-
-    if gpu_number > 1:
-        main_pid = 0
-        pid_col = "pid"
-        delta_ratio = 0.1
-
-        delta_df = df.sample(fraction=delta_ratio, seed=0)
-
-        df = df.withColumn(pid_col, SparkF.lit(main_pid))
-        delta_df = delta_df.withColumn(
-            pid_col, SparkF.monotonically_increasing_id() % (gpu_number * 4)
-        )
-
-        df = df.union(delta_df)
-        df_gpu = df.repartition(gpu_number, pid_col)
-
-    df_test = df.sample(fraction=fraction_sampled_for_test, seed=0)
-    numrows = df_test.count()
-    print(f"finished with numrows {numrows}")
-
-    # Train the Logistic Regression model
-    lr = LogisticRegression(num_workers=gpu_number, verbose=True, **est_params, elasticNetParam=elasticNetParam)
-    # lr_model = lr.fit(df)