diff --git a/.github/workflows/tests-xgboost.yaml b/.github/workflows/tests-xgboost.yaml
index f5989d5fbd..3a3c2c5044 100644
--- a/.github/workflows/tests-xgboost.yaml
+++ b/.github/workflows/tests-xgboost.yaml
@@ -22,6 +22,7 @@ jobs:
     - name: Install other dependencies
       run: |
         python -m pip install -U pip setuptools wheel
+        python -m pip install -U "Cython<3.0"
         pip install .
         pip install -r requirements/requirements-test.txt
         pip install -r requirements/requirements-rotbaum.txt
diff --git a/requirements/requirements-arrow.txt b/requirements/requirements-arrow.txt
index 6e5a693686..19d8363071 100644
--- a/requirements/requirements-arrow.txt
+++ b/requirements/requirements-arrow.txt
@@ -1,2 +1 @@
-pyarrow>=6.0; python_version=='3.6.*'
-pyarrow~=8.0; python_version>='3.7'
+pyarrow
diff --git a/src/gluonts/core/serde/_base.py b/src/gluonts/core/serde/_base.py
index 6f51e02a70..5432bd8105 100644
--- a/src/gluonts/core/serde/_base.py
+++ b/src/gluonts/core/serde/_base.py
@@ -309,7 +309,7 @@ def decode(r: Any) -> Any:
     """
 
     # structural recursion over the possible shapes of r
-    if type(r) == dict and "__kind__" in r:
+    if isinstance(r, dict) and "__kind__" in r:
         kind = r["__kind__"]
         cls = cast(Any, locate(r["class"]))
 
@@ -331,10 +331,10 @@ def decode(r: Any) -> Any:
 
         raise ValueError(f"Unknown kind {kind}.")
 
-    if type(r) == dict:
+    if isinstance(r, dict):
         return valmap(decode, r)
 
-    if type(r) == list:
+    if isinstance(r, list):
         return list(map(decode, r))
 
     return r
diff --git a/src/gluonts/dataset/arrow/dec.py b/src/gluonts/dataset/arrow/dec.py
index 3082b70790..148d5311c7 100644
--- a/src/gluonts/dataset/arrow/dec.py
+++ b/src/gluonts/dataset/arrow/dec.py
@@ -32,7 +32,7 @@ def from_schema(cls, schema):
         )
 
     def decode(self, batch, row_number: int):
-        yield from self.decode_batch(batch.slice(row_number, row_number + 1))
+        return next(self.decode_batch(batch.slice(row_number, row_number + 1)))
 
     def decode_batch(self, batch):
         for row in batch.to_pandas().to_dict("records"):
diff --git a/src/gluonts/model/predictor.py b/src/gluonts/model/predictor.py
index 69ff23ed84..e0d490c851 100644
--- a/src/gluonts/model/predictor.py
+++ b/src/gluonts/model/predictor.py
@@ -26,7 +26,7 @@
 
 import gluonts
 from gluonts.core import fqname_for
-from gluonts.core.component import equals, from_hyperparameters, validated
+from gluonts.core.component import equals, from_hyperparameters
 from gluonts.core.serde import dump_json, load_json
 from gluonts.dataset.common import DataEntry, Dataset
 from gluonts.model.forecast import Forecast
@@ -135,23 +135,18 @@ def from_inputs(cls, train_iter, **params):
 
 class RepresentablePredictor(Predictor):
     """
-    An abstract predictor that can be subclassed by models that are not based
-    on Gluon. Subclasses should have @validated() constructors.
-    (De)serialization and value equality are all implemented on top of the.
+    An abstract predictor that can be subclassed by framework-specific models.
+    Subclasses should have ``@validated()`` constructors:
+    (de)serialization and equality test are all implemented on top of its logic.
 
-    @validated() logic.
     Parameters
     ----------
     prediction_length
         Prediction horizon.
+    lead_time
+        Prediction lead time.
     """
 
-    @validated()
-    def __init__(self, prediction_length: int, lead_time: int = 0) -> None:
-        super().__init__(
-            lead_time=lead_time, prediction_length=prediction_length
-        )
-
     def predict(self, dataset: Dataset, **kwargs) -> Iterator[Forecast]:
         for item in dataset:
             yield self.predict_item(item)
diff --git a/src/gluonts/mx/model/seq2seq/_mq_dnn_estimator.py b/src/gluonts/mx/model/seq2seq/_mq_dnn_estimator.py
index 7359beb3a7..5864d37931 100644
--- a/src/gluonts/mx/model/seq2seq/_mq_dnn_estimator.py
+++ b/src/gluonts/mx/model/seq2seq/_mq_dnn_estimator.py
@@ -317,7 +317,7 @@ def from_inputs(cls, train_iter, **params):
             if field in params.keys():
                 is_params_field = (
                     params[field]
-                    if type(params[field]) == bool
+                    if isinstance(params[field], bool)
                     else strtobool(params[field])
                 )
                 if is_params_field and not auto_params[field]:
diff --git a/test/dataset/test_arrow.py b/test/dataset/test_arrow.py
index 3c12818165..9aba14f7bb 100644
--- a/test/dataset/test_arrow.py
+++ b/test/dataset/test_arrow.py
@@ -21,6 +21,7 @@
 
 from gluonts.dataset.arrow import (
     File,
+    ArrowFile,
     ArrowWriter,
     ParquetWriter,
 )
@@ -78,3 +79,6 @@ def test_arrow(writer, flatten_arrays):
 
         for orig, arrow_value in zip(data, dataset):
             assert_equal(orig, arrow_value)
+
+        if isinstance(dataset, ArrowFile):
+            assert_equal(dataset[4], data[4])