diff --git a/doc/releases/0.100.3.rst b/doc/releases/0.100.3.rst
new file mode 100644
index 0000000000..f43d0a8034
--- /dev/null
+++ b/doc/releases/0.100.3.rst
@@ -0,0 +1,12 @@
+.. _release0.100.3:
+
+SpikeInterface 0.100.3 release notes
+------------------------------------
+
+28th March 2024
+
+Minor release with bug fixes for Zarr compressor and NWB in container
+
+* Remove lazy typing in nwb (#2635)
+* Add extra_requirements for nwb extractors (#2637)
+* Fix compressor propagation to Zarr save function (#2639)
diff --git a/doc/whatisnew.rst b/doc/whatisnew.rst
index 48606467fb..3063db51f5 100644
--- a/doc/whatisnew.rst
+++ b/doc/whatisnew.rst
@@ -8,6 +8,7 @@ Release notes
 .. toctree::
   :maxdepth: 1
 
+  releases/0.100.3.rst
   releases/0.100.2.rst
   releases/0.100.1.rst
   releases/0.100.0.rst
@@ -36,6 +37,12 @@ Release notes
   releases/0.9.1.rst
 
 
+Version 0.100.3
+==============
+
+* Minor release with bug fixes for Zarr compressor and NWB in container
+
+
 Version 0.100.2
 ==============
 
diff --git a/pyproject.toml b/pyproject.toml
index d36c3b9e3c..2a2a072fc8 100644
--- a/pyproject.toml
+++ b/pyproject.toml
@@ -1,6 +1,6 @@
 [project]
 name = "spikeinterface"
-version = "0.100.2"
+version = "0.100.3"
 authors = [
   { name="Alessio Buccino", email="alessiop.buccino@gmail.com" },
   { name="Samuel Garcia", email="sam.garcia.die@gmail.com" },
diff --git a/src/spikeinterface/core/zarrextractors.py b/src/spikeinterface/core/zarrextractors.py
index 106f8ccc1e..a8a23b5863 100644
--- a/src/spikeinterface/core/zarrextractors.py
+++ b/src/spikeinterface/core/zarrextractors.py
@@ -381,7 +381,7 @@ def add_recording_to_zarr_group(recording: BaseRecording, zarr_group: zarr.hiera
     dataset_paths = [f"traces_seg{i}" for i in range(recording.get_num_segments())]
 
     zarr_kwargs["dtype"] = kwargs.get("dtype", None) or recording.get_dtype()
-    if "compressor" not in zarr_group:
+    if "compressor" not in zarr_kwargs:
         zarr_kwargs["compressor"] = get_default_zarr_compressor()
 
     add_traces_to_zarr(
@@ -392,7 +392,7 @@ def add_recording_to_zarr_group(recording: BaseRecording, zarr_group: zarr.hiera
         **job_kwargs,
     )
 
-    # # save probe
+    # save probe
     if recording.get_property("contact_vector") is not None:
         probegroup = recording.get_probegroup()
         zarr_group.attrs["probe"] = check_json(probegroup.to_dict(array_as_list=True))
diff --git a/src/spikeinterface/extractors/nwbextractors.py b/src/spikeinterface/extractors/nwbextractors.py
index 287522a870..09b462d5ae 100644
--- a/src/spikeinterface/extractors/nwbextractors.py
+++ b/src/spikeinterface/extractors/nwbextractors.py
@@ -10,15 +10,6 @@
 from spikeinterface.core.core_tools import define_function_from_class
 
 
-def import_lazily():
-    "Makes annotations / typing available lazily"
-    global NWBFile, ElectricalSeries, Units, NWBHDF5IO
-    from pynwb import NWBFile
-    from pynwb.ecephys import ElectricalSeries
-    from pynwb.misc import Units
-    from pynwb import NWBHDF5IO
-
-
 def read_file_from_backend(
     *,
     file_path: str | Path | None,
@@ -111,7 +102,7 @@ def read_nwbfile(
     cache: bool = False,
     stream_cache_path: str | Path | None = None,
     storage_options: dict | None = None,
-) -> NWBFile:
+) -> "NWBFile":
     """
     Read an NWB file and return the NWBFile object.
 
@@ -176,8 +167,8 @@ def read_nwbfile(
 
 
 def _retrieve_electrical_series_pynwb(
-    nwbfile: NWBFile, electrical_series_path: Optional[str] = None
-) -> ElectricalSeries:
+    nwbfile: "NWBFile", electrical_series_path: Optional[str] = None
+) -> "ElectricalSeries":
     """
     Get an ElectricalSeries object from an NWBFile.
 
@@ -230,7 +221,7 @@ def _retrieve_electrical_series_pynwb(
     return electrical_series
 
 
-def _retrieve_unit_table_pynwb(nwbfile: NWBFile, unit_table_path: Optional[str] = None) -> Units:
+def _retrieve_unit_table_pynwb(nwbfile: "NWBFile", unit_table_path: Optional[str] = None) -> "Units":
     """
     Get an Units object from an NWBFile.
     Units tables can be either the main unit table (nwbfile.units) or in the processing module.
@@ -568,8 +559,10 @@ def __init__(
         # fetch and add main recording properties
         if use_pynwb:
             gains, offsets, locations, groups = self._fetch_main_properties_pynwb()
+            self.extra_requirements.append("pynwb")
         else:
             gains, offsets, locations, groups = self._fetch_main_properties_backend()
+            self.extra_requirements.append("h5py")
         self.set_channel_gains(gains)
         self.set_channel_offsets(offsets)
         if locations is not None:
@@ -781,8 +774,6 @@ def _fetch_other_properties(self, electrodes_table, electrodes_indices, columns)
         #########
         # Extract and re-name properties from nwbfile TODO: Should be a function
         ########
-        from pynwb.ecephys import ElectrodeGroup
-
         properties = dict()
         properties_to_skip = [
             "id",
@@ -798,9 +789,7 @@ def _fetch_other_properties(self, electrodes_table, electrodes_indices, columns)
 
         for column in columns:
             first_value = electrodes_table[column][0]
-            if isinstance(first_value, ElectrodeGroup):
-                continue
-            elif column in properties_to_skip:
+            if column in properties_to_skip:
                 continue
             else:
                 column_name = rename_properties.get(column, column)
@@ -1024,8 +1013,10 @@ def __init__(
         if load_unit_properties:
             if use_pynwb:
                 columns = [c.name for c in self.units_table.columns]
+                self.extra_requirements.append("pynwb")
             else:
                 columns = list(self.units_table.keys())
+                self.extra_requirements.append("h5py")
             properties = self._fetch_properties(columns)
             for property_name, property_values in properties.items():
                 values = [x.decode("utf-8") if isinstance(x, bytes) else x for x in property_values]