diff --git a/returnn/tf/util/data.py b/returnn/tf/util/data.py
index 91ec806bde..37dbb1029b 100644
--- a/returnn/tf/util/data.py
+++ b/returnn/tf/util/data.py
@@ -148,6 +148,22 @@ def short_repr(self):
         desc += "{ctx=%s}" % self.control_flow_ctx.repr_inner()
     return desc
 
+  def __getstate__(self):
+    d = vars(self).copy()
+    d["batch"] = None
+    d["_same_as_tb"] = None
+    d["_same_for_batch_ctx"] = {}
+    d["dyn_size_ext"] = None
+    d["kind"] = self.kind.name if self.kind else None
+    return d
+
+  def __setstate__(self, state):
+    self.__dict__.update(state)
+    if self.kind is not None:
+      self.kind = {v.name: v for v in Dim.Types.Types}[self.kind]
+    if self.is_batch_dim():
+      self.same_as = batch_dim
+
   def __copy__(self):
     """
     Normally we would not want to get a new tag with ``tag != copy(tag)``.
@@ -2042,6 +2058,9 @@ def short_repr(self):
     # "x" is the Theano-style shortcut for a broadcast dim.
     return "&".join([dim.short_repr() for dim in self.virtual_dims] or ["Bx"])
 
+  def __getstate__(self):
+    raise Exception("Pickling of BatchInfo is not supported. (%s)" % self)
+
   @property
   def dim(self):
     """
@@ -3016,6 +3035,12 @@ def __repr__(self):
   def __hash__(self):
     return id(self)
 
+  def __getstate__(self):
+    d = vars(self)
+    d["_batch"] = None  # BatchInfo pickling not supported
+    d["_placeholder"] = None  # do not store the TF tensors
+    return d
+
   def _adapt_batch_consistent_dim_tags(self):
     if not self.batch:  # uninitialized
       return
diff --git a/returnn/util/basic.py b/returnn/util/basic.py
index 3d873dafd8..17bd26266f 100644
--- a/returnn/util/basic.py
+++ b/returnn/util/basic.py
@@ -101,6 +101,9 @@ def __str__(self):
   def __repr__(self):
     return "<%s>" % self.name
 
+  def __getstate__(self):
+    raise Exception("Cannot pickle Entity object. (%r)" % self)
+
 
 class OptionalNotImplementedError(NotImplementedError):
   """
@@ -2520,7 +2523,7 @@ def pip_install(*pkg_names):
 
 def pip_check_is_installed(pkg_name):
   """
-  :param str pkg_name: without version, e.g. just "tensorflow" or "tensorflow-gpu"
+  :param str pkg_name: without version, e.g. just "tensorflow", or with version, e.g. "tensorflow==1.2.3"
   :rtype: bool
   """
   if not _pip_installed_packages:
@@ -2528,8 +2531,10 @@ def pip_check_is_installed(pkg_name):
     pip_path = which_pip()
     cmd = [py, pip_path, "freeze"]
     for line in sys_exec_out(*cmd).splitlines():
-      if line:
-        _pip_installed_packages.add(line[:line.index("==")])
+      if line and "==" in line:
+        if "==" not in pkg_name:
+          line = line[:line.index("==")]
+        _pip_installed_packages.add(line)
   return pkg_name in _pip_installed_packages
 
 
diff --git a/tests/pycharm-inspect.py b/tests/pycharm-inspect.py
index b52e2d8d6b..3aa4ca4bdf 100755
--- a/tests/pycharm-inspect.py
+++ b/tests/pycharm-inspect.py
@@ -280,7 +280,7 @@ def setup_pycharm_python_interpreter(pycharm_dir):
     if not pip_check_is_installed("Theano"):
       pip_install("theano==0.9")
     # Note: Horovod will usually fail to install in this env.
-    for pkg in ["typing", "librosa", "PySoundFile", "nltk", "matplotlib", "mpi4py", "pycodestyle"]:
+    for pkg in ["typing", "librosa==0.8.1", "PySoundFile", "nltk", "matplotlib", "mpi4py", "pycodestyle"]:
       if not pip_check_is_installed(pkg):
         try:
           pip_install(pkg)
diff --git a/tests/test_TFNetworkLayer.py b/tests/test_TFNetworkLayer.py
index 0fc4ffde31..ce87049ebe 100644
--- a/tests/test_TFNetworkLayer.py
+++ b/tests/test_TFNetworkLayer.py
@@ -6480,6 +6480,75 @@ def test_batch_norm_args():
   assert batch_norm_args == layer_args  # different arguments in BatchNormLayer and LayerBase.batch_norm()
 
 
+def test_pickle_dim_tags():
+  # Test for pickling net dict and extern data.
+  # https://github.com/rwth-i6/returnn_common/issues/104
+  # What kind of network we pickle here is not really relevant.
+  # I took this from some other test.
+  # It should just involve some dim tags.
+  from returnn.tf.util.data import batch_dim, ImplicitDynSizeDim
+  n_batch, n_time, n_ts, n_in, n_out = 2, 3, 6, 7, 11
+  time_dim = SpatialDim("time")
+  feat_dim = FeatureDim("in-feature", dimension=n_in)
+  ts_dim = SpatialDim("ts", dimension=n_ts)
+  out_dim = FeatureDim("out-feature", dimension=n_out)
+  config = Config({
+    "extern_data": {"data": {"dim_tags": [batch_dim, time_dim, feat_dim]}},
+    "network": {
+      "t": {
+        "class": "range_in_axis", "axis": "t", "from": "data",
+        "out_shape": {time_dim, ImplicitDynSizeDim(batch_dim)}},  # (T,)
+      "range": {"class": "range", "limit": n_ts, "out_spatial_dim": ts_dim},  # (Ts,)
+      "add_t": {
+        "class": "combine", "kind": "add", "from": ["t", "range"],
+        "out_shape": {time_dim, ts_dim, ImplicitDynSizeDim(batch_dim)}},  # (T,Ts)
+      "t_rel_var": {"class": "variable", "shape": (ts_dim, out_dim), "init": "glorot_uniform"},  # (Ts,D)
+      "output": {
+        "class": "scatter_nd", "from": "t_rel_var", "position": "add_t", "position_axis": ts_dim,
+        "out_spatial_dim": time_dim, "filter_invalid_indices": True}  # (T,T,D)
+    }
+  })
+
+  # First test as-is.
+  with make_scope() as session:
+    network = TFNetwork(config=config, train_flag=True)
+    network.construct_from_dict(config.typed_dict["network"])
+    network.initialize_params(session)
+    fetches = network.get_fetches_dict()
+    out_layer = network.get_default_output_layer()
+    print("out layer:", out_layer)
+    session.run((out_layer.output.placeholder, fetches), feed_dict=make_feed_dict(network.extern_data))
+
+  def _debug_dump(tuple_path, obj):
+    print("%s: %s" % (tuple_path, type(obj)))
+    assert isinstance(obj, (Dim, str, set, str, bool, int))
+
+  from tensorflow.python.util import nest
+  nest.map_structure_with_tuple_paths(_debug_dump, config.typed_dict)
+
+  # Now pickle, unpickle and test again.
+  # Use pickle._dumps for easier debugging.
+  # noinspection PyUnresolvedReferences
+  from pickle import _dumps as dumps
+  s = dumps(config.typed_dict)
+  import pickle
+  config_dict = pickle.loads(s)
+  new_dim_tags = config_dict["extern_data"]["data"]["dim_tags"]
+  new_batch, new_time, new_feat = new_dim_tags
+  assert isinstance(new_batch, Dim) and new_batch == batch_dim and new_batch.is_batch_dim()
+  assert isinstance(new_time, Dim) and new_time.is_spatial_dim() and new_time.dimension is None
+  assert isinstance(new_feat, Dim) and new_feat.is_feature_dim() and new_feat.dimension == n_in
+  config = Config(config_dict)
+  with make_scope() as session:
+    network = TFNetwork(config=config, train_flag=True)
+    network.construct_from_dict(config.typed_dict["network"])
+    network.initialize_params(session)
+    fetches = network.get_fetches_dict()
+    out_layer = network.get_default_output_layer()
+    print("out layer:", out_layer)
+    session.run((out_layer.output.placeholder, fetches), feed_dict=make_feed_dict(network.extern_data))
+
+
 def test_contrastive_loss():
   from returnn.tf.util.data import batch_dim
   masked_time_dim = SpatialDim("masked_time")