chronos repo name in deps, make style

Signed-off-by: Giridhar Ganapavarapu <[email protected]>

services/inference/pyproject.toml

@@ -49,7 +49,7 @@ urllib3 = { version = ">=1.26.19,<2" } # see https://github.com/urllib3/urllib3/
 aiohttp = { version = ">=3.10.11" }
 
 # ***********Chronos*********
-chronos = { git = "https://github.com/amazon-science/chronos-forecasting.git" }
+chronos-forecasting = { git = "https://github.com/amazon-science/chronos-forecasting.git" }
 
 [[tool.poetry.source]]
 name = "pytorch"

services/inference/tests/test_inference_lib.py

@@ -33,19 +33,14 @@
 MODEL_IDS = [
     os.path.basename(dirpath)
     for dirpath, _, _ in os.walk("./mytest-tsfm")
-    if ".git" not in dirpath
-    and "./mytest-tsfm" != dirpath
-    and "finetuned" not in dirpath
-    and "figures" not in dirpath
+    if ".git" not in dirpath and "./mytest-tsfm" != dirpath and "finetuned" not in dirpath and "figures" not in dirpath
 ]
 
 
 def min_context_length(model_id):
     model_path: Path = resolve_model_path(TSFM_MODEL_DIR, model_id)
     assert model_path.exists(), f"{model_path} does not exist!"
-    handler, e = ForecastingServiceHandler.load(
-        model_id=model_id, model_path=model_path
-    )
+    handler, e = ForecastingServiceHandler.load(model_id=model_id, model_path=model_path)
     return handler.handler_config.minimum_context_length
 
 
@@ -80,9 +75,7 @@ def forecasting_input_base(request: type[FixtureRequest]) -> ForecastingInferenc
     schema: ForecastingMetadataInput = ForecastingMetadataInput(
         timestamp_column="date", id_columns=["ID"], target_columns=["VAL"]
     )
-    parameters: ForecastingParameters = ForecastingParameters(
-        prediction_length=FORECAST_LENGTH
-    )
+    parameters: ForecastingParameters = ForecastingParameters(prediction_length=FORECAST_LENGTH)
     input: ForecastingInferenceInput = ForecastingInferenceInput(
         model_id=request.param,
         schema=schema,
@@ -108,16 +101,12 @@ def _basic_result_checks(results: PredictOutput, df: pd.DataFrame):
     # expected start time
     assert results["date"].iloc[0] - df["date"].iloc[-1] == timedelta(hours=1)
     # expected end time
-    assert results["date"].iloc[-1] - df["date"].iloc[-1] == timedelta(
-        hours=FORECAST_LENGTH
-    )
+    assert results["date"].iloc[-1] - df["date"].iloc[-1] == timedelta(hours=FORECAST_LENGTH)
 
 
 @pytest.mark.parametrize("forecasting_input_base", MODEL_IDS, indirect=True)
 @pytest.mark.parametrize("ts_data_base", MODEL_IDS, indirect=True)
-def test_forecast_with_good_data(
-    ts_data_base: pd.DataFrame, forecasting_input_base: ForecastingInferenceInput
-):
+def test_forecast_with_good_data(ts_data_base: pd.DataFrame, forecasting_input_base: ForecastingInferenceInput):
     input = forecasting_input_base
     data, model_id = ts_data_base
     # since we're sometimes generating non-sensible combinations
@@ -183,18 +172,13 @@ def test_forecast_with_integer_timestamps(
     po: PredictOutput = runtime.forecast(input=input)
     results = pd.DataFrame.from_dict(po.results[0])
     assert results[timestamp_column].iloc[0] == series_length + 1
-    assert (
-        results[timestamp_column].iloc[-1] - df[timestamp_column].iloc[-1]
-        == FORECAST_LENGTH
-    )
+    assert results[timestamp_column].iloc[-1] - df[timestamp_column].iloc[-1] == FORECAST_LENGTH
     assert results.dtypes[timestamp_column] == df.dtypes[timestamp_column]
 
 
 @pytest.mark.parametrize("forecasting_input_base", MODEL_IDS, indirect=True)
 @pytest.mark.parametrize("ts_data_base", MODEL_IDS, indirect=True)
-def test_forecast_with_bogus_timestamps(
-    ts_data_base: pd.DataFrame, forecasting_input_base: ForecastingInferenceInput
-):
+def test_forecast_with_bogus_timestamps(ts_data_base: pd.DataFrame, forecasting_input_base: ForecastingInferenceInput):
     input: ForecastingInferenceInput = copy.deepcopy(forecasting_input_base)
     data, model_id = ts_data_base
     # since we're sometimes generating non-sensible combinations
@@ -214,9 +198,7 @@ def test_forecast_with_bogus_timestamps(
 
 @pytest.mark.parametrize("forecasting_input_base", MODEL_IDS, indirect=True)
 @pytest.mark.parametrize("ts_data_base", MODEL_IDS, indirect=True)
-def test_forecast_with_bogus_values(
-    ts_data_base: pd.DataFrame, forecasting_input_base: ForecastingInferenceInput
-):
+def test_forecast_with_bogus_values(ts_data_base: pd.DataFrame, forecasting_input_base: ForecastingInferenceInput):
     input: ForecastingInferenceInput = copy.deepcopy(forecasting_input_base)
     data, model_id = ts_data_base
     # since we're sometimes generating non-sensible combinations
@@ -235,9 +217,7 @@ def test_forecast_with_bogus_values(
 
 @pytest.mark.parametrize("forecasting_input_base", MODEL_IDS, indirect=True)
 @pytest.mark.parametrize("ts_data_base", MODEL_IDS, indirect=True)
-def test_forecast_with_bogus_model_id(
-    ts_data_base: pd.DataFrame, forecasting_input_base: ForecastingInferenceInput
-):
+def test_forecast_with_bogus_model_id(ts_data_base: pd.DataFrame, forecasting_input_base: ForecastingInferenceInput):
     input: ForecastingInferenceInput = copy.deepcopy(forecasting_input_base)
     data, model_id = ts_data_base
     # since we're sometimes generating non-sensible combinations
@@ -277,9 +257,7 @@ def test_forecast_with_insufficient_context_length(
 @pytest.mark.skip
 @pytest.mark.parametrize("forecasting_input_base", MODEL_IDS, indirect=True)
 @pytest.mark.parametrize("ts_data_base", MODEL_IDS, indirect=True)
-def test_forecast_with_nan_data(
-    ts_data_base: pd.DataFrame, forecasting_input_base: ForecastingInferenceInput
-):
+def test_forecast_with_nan_data(ts_data_base: pd.DataFrame, forecasting_input_base: ForecastingInferenceInput):
     input: ForecastingInferenceInput = copy.deepcopy(forecasting_input_base)
     data, model_id = ts_data_base
     # since we're sometimes generating non-sensible combinations
@@ -299,9 +277,7 @@ def test_forecast_with_nan_data(
 # @pytest.mark.skip
 @pytest.mark.parametrize("forecasting_input_base", MODEL_IDS, indirect=True)
 @pytest.mark.parametrize("ts_data_base", MODEL_IDS, indirect=True)
-def test_forecast_with_missing_row(
-    ts_data_base: pd.DataFrame, forecasting_input_base: ForecastingInferenceInput
-):
+def test_forecast_with_missing_row(ts_data_base: pd.DataFrame, forecasting_input_base: ForecastingInferenceInput):
     input: ForecastingInferenceInput = copy.deepcopy(forecasting_input_base)
     data, model_id = ts_data_base
     # since we're sometimes generating non-sensible combinations