diff --git a/relik/common/utils.py b/relik/common/utils.py
index b0a9463..9be7012 100644
--- a/relik/common/utils.py
+++ b/relik/common/utils.py
@@ -339,7 +339,7 @@ def download_from_hf(
             downloaded_paths.append(downloaded_path)
         except OSError:
             if ignore_failure:
-                logger.error(
+                logger.warn(
                     f"Couldn't download {filename} from {path_or_repo_id}, ignoring"
                 )
             else:
diff --git a/relik/inference/annotator.py b/relik/inference/annotator.py
index 6fc51fd..44db8b0 100644
--- a/relik/inference/annotator.py
+++ b/relik/inference/annotator.py
@@ -551,8 +551,6 @@ def __call__(
         windows = windows + blank_windows
         windows.sort(key=lambda x: (x.doc_id, x.offset))
 
-        print(windows)
-
         # if there is no reader, just return the windows
         if self.reader is None:
             # normalize window candidates to be a list of lists, like when the reader is used
@@ -563,7 +561,6 @@ def __call__(
             merged_windows = self.window_manager.merge_windows(windows)
 
         # transform predictions into RelikOutput objects
-        print(merged_windows)
         output = []
         for w in merged_windows:
             span_labels = []
diff --git a/relik/retriever/indexers/base.py b/relik/retriever/indexers/base.py
index e48e707..ff7a621 100644
--- a/relik/retriever/indexers/base.py
+++ b/relik/retriever/indexers/base.py
@@ -445,7 +445,7 @@ def from_pretrained(
         config = OmegaConf.load(config_path)
         # add the actual cls class to the config in place of the _target_ if cls is not BaseDocumentIndex
 
-        target = config.pop("_target_", None)
+        target = config.get("_target_", None)
         if cls.__name__ != "BaseDocumentIndex" and target is None:
             kwargs["_target_"] = f"{cls.__module__}.{cls.__name__}"
 
@@ -483,7 +483,6 @@ def from_pretrained(
         # boolean variables to check if the index and embeddings exist
         index_exists = index_path.exists()
         embeddings_exists = embedding_path.exists()
-        # use_faiss = "FaissDocumentIndex" in cls.__name__
 
         if use_faiss:
             if index_exists:
diff --git a/requirements.txt b/requirements.txt
index 004bc31..7ac7f93 100644
--- a/requirements.txt
+++ b/requirements.txt
@@ -4,7 +4,7 @@ torch==2.3.1
 
 transformers[sentencepiece]>=4.41,<4.42
 rich>=13.0.0,<14.0.0
-scikit-learn>=1.3,<1.4
+scikit-learn>=1.5,<1.6
 overrides>=7.4,<7.9
 art==6.2
 pprintpp==0.4.0
@@ -28,12 +28,12 @@ faiss-cpu==1.8.0 # needed by: faiss
 # serve
 fastapi>=0.112,<0.113 # needed by: serve, ray
 uvicorn[standard]==0.23.2 # needed by: serve, ray
-gunicorn==21.2.0 # needed by: serve, ray
+gunicorn==22.0.0 # needed by: serve, ray
 streamlit>=1.28,<1.29 # needed by: serve, ray
 streamlit_extras>=0.3,<0.4 # needed by: serve, ray
 gradio>=4.37,<4.38 # needed by: serve, ray
 pyvis # needed by: serve, ray
-ray[serve]>=2.7,<=2.8 # needed by: ray
+ray[serve]>=2.34,<=2.35 # needed by: ray
 
 # dev
 pre-commit # needed by: dev