From c48ac5640397d846275b3b5f28abfe4b2a5af787 Mon Sep 17 00:00:00 2001
From: Nanbo Liu <nliu@lambda001.america.striveworks.us>
Date: Tue, 21 Jan 2025 14:00:51 -0600
Subject: [PATCH 01/21] Added chariotoutputcodec for hf image pipelines

---
 mlserver/model.py                             |  1 -
 .../mlserver_huggingface/codecs/__init__.py   |  3 +-
 .../mlserver_huggingface/codecs/chariot.py    | 92 +++++++++++++++++++
 .../mlserver_huggingface/common.py            | 23 +++--
 .../mlserver_huggingface/runtime.py           | 13 ++-
 5 files changed, 113 insertions(+), 19 deletions(-)
 create mode 100644 runtimes/huggingface/mlserver_huggingface/codecs/chariot.py

diff --git a/mlserver/model.py b/mlserver/model.py
index 8296bc19e..c89574f89 100644
--- a/mlserver/model.py
+++ b/mlserver/model.py
@@ -247,7 +247,6 @@ def encode(
         response_output = encode_response_output(
             payload, request_output, self._outputs_index
         )
-
         if response_output:
             return response_output
 
diff --git a/runtimes/huggingface/mlserver_huggingface/codecs/__init__.py b/runtimes/huggingface/mlserver_huggingface/codecs/__init__.py
index 4a1d67025..5388077af 100644
--- a/runtimes/huggingface/mlserver_huggingface/codecs/__init__.py
+++ b/runtimes/huggingface/mlserver_huggingface/codecs/__init__.py
@@ -6,7 +6,7 @@
 from .conversation import HuggingfaceConversationCodec
 from .raw import RawCodec
 from .utils import EqualUtil
-
+from .chariot import ChariotModelOutputCodec
 __all__ = [
     "MultiInputRequestCodec",
     "HuggingfaceRequestCodec",
@@ -14,6 +14,7 @@
     "HuggingfaceSingleJSONCodec",
     "HuggingfaceListJSONCodec",
     "HuggingfaceConversationCodec",
+    "ChariotModelOutputCodec",
     "NumpyListCodec",
     "RawCodec",
     "EqualUtil",
diff --git a/runtimes/huggingface/mlserver_huggingface/codecs/chariot.py b/runtimes/huggingface/mlserver_huggingface/codecs/chariot.py
new file mode 100644
index 000000000..18ff391d1
--- /dev/null
+++ b/runtimes/huggingface/mlserver_huggingface/codecs/chariot.py
@@ -0,0 +1,92 @@
+import numpy as np
+
+def is_list_of_dicts(var):
+    '''Check if a variable is a list of dicts'''
+    if not isinstance(var, list):
+        return False
+    for item in var:
+        if not isinstance(item, dict):
+            return False
+    return True
+def get_det_dict_from_hf_obj_detect(obj_detect):
+    '''Convert hf object detection output to standard chariot object detection output'''
+    det_dict = {
+        "num_detections": 0,
+        "detection_classes": [],
+        "detection_boxes": [],
+        "detection_scores": [],
+    }
+    for det in obj_detect:
+        conf, cls = det["score"],det["label"]
+        x1, y1, x2, y2  = det["box"]["xmin"],det["box"]["ymin"],det["box"]["xmax"],det["box"]["ymax"]
+        det_dict["num_detections"] += 1
+        det_dict["detection_classes"].append(cls)
+        det_dict["detection_scores"].append(conf)
+        det_dict["detection_boxes"].append([x1, y1, x2, y2])
+    return det_dict
+def get_chariot_seg_mask_from_hf_seg_output(seg_pred,class_int_to_str):
+    '''Convert hf segmentation output to standard chariot segmentation output'''
+    img_w,img_h = seg_pred[0]["mask"].size
+    if class_int_to_str[0]!="background":
+        # Let class_int 0 be background and increment all the other class_int by 1
+        class_str_to_int = {v:k+1 for k,v in class_int_to_str.items()}
+    else:
+        class_str_to_int = {v:k for k,v in class_int_to_str.items()}
+    # Create an empty mask
+    combined_mask = np.zeros((img_h, img_w), dtype=np.uint8)
+    for i in seg_pred:
+        # Convert mask from PIL image to numpy array
+        mask = np.array(i["mask"])
+        class_str = i["label"]
+        class_int = class_str_to_int[class_str] 
+        combined_mask[np.where(mask>0)]=class_int
+    predictions= [combined_mask]
+    return predictions
+        
+    
+class ChariotModelOutputCodec():
+    """Encoder that converts HF model output to the standard Chariot model output 
+    """
+    @classmethod
+    def encode_output(cls, predictions,task_type,pipeline):
+        if task_type == "image-classification":
+            if all([is_list_of_dicts(p) for p in predictions]):
+                # get Top-1 predicted class
+                # convert HF output: [[{"label": "tabby, tabby cat", "score": 0.94}, 
+                #                      {"label": "tiger cat", "score": 0.04},
+                #                      {"label": "Egyptian cat", "score": 0.02}]]
+                # to standard Chariot output: ["tabby, tabby cat"]
+                predictions = [p[0]["label"] for p in predictions]
+        
+        elif task_type == "object-detection":
+            if is_list_of_dicts(predictions):
+                # convert HF output: [{'score': 0.9897010326385498,
+                #                       'label': 'cat',
+                #                       'box': {'xmin': 53, 'ymin': 313, 'xmax': 697, 'ymax': 986}},
+                #                       {'score': 0.9896764159202576,
+                #                       'label': 'cat',
+                #                       'box': {'xmin': 974, 'ymin': 221, 'xmax': 1526, 'ymax': 1071}}]
+
+                # to standard Chariot output: [{"num_detections":2,
+                #                             "detection_classes":["cat","cat"],
+                #                             "detection_scores":[0.9897010326385498,0.9896764159202576],
+                #                             "detection_boxes":[[53,313,697,986],
+                #                                                [974,221,1526,1071]]}]
+                predictions = get_det_dict_from_hf_obj_detect(predictions) 
+        
+        elif task_type == "image-segmentation":
+            if is_list_of_dicts(predictions):
+                    # convert HF output: [{"score": None,
+                    #                      "label": "wall",
+                    #                      "mask": <PIL.Image.Image image mode=L size=5362x3016>},
+                    #                     {"score": None,
+                    #                      "label": "floor",
+                    #                      "mask": <PIL.Image.Image image mode=L size=5362x3016>}]
+
+                    # to standard Chariot output: [[0,0,...,0],...,[0,0,0,...,0]] 
+                    # 2d array with size of the original image. Each pixel is a class int
+                    # Background uses class_int 0 
+                    class_int_to_str=pipeline.model.config.id2label
+                    predictions = get_chariot_seg_mask_from_hf_seg_output(predictions,class_int_to_str) 
+        return predictions
+    
diff --git a/runtimes/huggingface/mlserver_huggingface/common.py b/runtimes/huggingface/mlserver_huggingface/common.py
index e53d7f6e2..84fd53cd4 100644
--- a/runtimes/huggingface/mlserver_huggingface/common.py
+++ b/runtimes/huggingface/mlserver_huggingface/common.py
@@ -33,8 +33,6 @@ def load_pipeline_from_settings(
     if not model:
         model = settings.parameters.uri  # type: ignore
     tokenizer = hf_settings.pretrained_tokenizer
-    if not tokenizer:
-        tokenizer = hf_settings.pretrained_model
     if hf_settings.framework == "tf":
         if hf_settings.inter_op_threads is not None:
             tf.config.threading.set_inter_op_parallelism_threads(
@@ -49,7 +47,7 @@ def load_pipeline_from_settings(
             torch.set_num_interop_threads(hf_settings.inter_op_threads)
         if hf_settings.intra_op_threads is not None:
             torch.set_num_threads(hf_settings.intra_op_threads)
-
+    # HF pipeline would automatically load tokenizer from model directory if no tokenizer is provided in the config
     hf_pipeline = pipeline(
         hf_settings.task_name,
         model=model,
@@ -63,15 +61,16 @@ def load_pipeline_from_settings(
     # If max_batch_size > 1 we need to ensure tokens are padded
     if settings.max_batch_size > 1:
         model = hf_pipeline.model
-        if not hf_pipeline.tokenizer.pad_token_id:
-            eos_token_id = model.config.eos_token_id  # type: ignore
-            if eos_token_id:
-                hf_pipeline.tokenizer.pad_token_id = [str(eos_token_id)]  # type: ignore
-            else:
-                logger.warning(
-                    "Model has neither pad_token or eos_token, setting batch size to 1"
-                )
-                hf_pipeline._batch_size = 1
+        if hf_pipeline.tokenizer is not None:
+            if not hf_pipeline.tokenizer.pad_token_id:
+                eos_token_id = model.config.eos_token_id  # type: ignore
+                if eos_token_id:
+                    hf_pipeline.tokenizer.pad_token_id = [str(eos_token_id)]  # type: ignore
+                else:
+                    logger.warning(
+                        "Model has neither pad_token or eos_token, setting batch size to 1"
+                    )
+                    hf_pipeline._batch_size = 1
 
     return hf_pipeline
 
diff --git a/runtimes/huggingface/mlserver_huggingface/runtime.py b/runtimes/huggingface/mlserver_huggingface/runtime.py
index c21d4c141..8f2ebfada 100644
--- a/runtimes/huggingface/mlserver_huggingface/runtime.py
+++ b/runtimes/huggingface/mlserver_huggingface/runtime.py
@@ -1,6 +1,7 @@
 import asyncio
 import torch
 
+from mlserver.codecs.lists import is_list_of
 from mlserver.model import MLModel
 from mlserver.settings import ModelSettings
 from mlserver.logging import logger
@@ -11,7 +12,7 @@
 
 from .settings import get_huggingface_settings
 from .common import load_pipeline_from_settings
-from .codecs import HuggingfaceRequestCodec
+from .codecs import HuggingfaceRequestCodec,ChariotModelOutputCodec
 from .metadata import METADATA
 
 
@@ -44,11 +45,13 @@ async def predict(self, payload: InferenceRequest) -> InferenceResponse:
         array_inputs = kwargs.pop("array_inputs", [])
         if array_inputs:
             args = [list(array_inputs)] + args
-        prediction = self._model(*args, **kwargs)
-
-        return self.encode_response(
-            payload=prediction, default_codec=HuggingfaceRequestCodec
+        predictions = self._model(*args, **kwargs)
+        if self.hf_settings.task in ["image-classification","image-segmentation","object-detection"]:   
+            predictions = ChariotModelOutputCodec.encode_output(predictions,task_type=self.hf_settings.task,pipeline=self._model)
+        response =self.encode_response(
+            payload=predictions, default_codec=HuggingfaceRequestCodec
         )
+        return response
 
     async def unload(self) -> bool:
         # TODO: Free up Tensorflow's GPU memory

From e5442fd422cc1355ac8924f5849a7819633dc37f Mon Sep 17 00:00:00 2001
From: Nanbo Liu <nliu@lambda001.america.striveworks.us>
Date: Tue, 21 Jan 2025 14:04:21 -0600
Subject: [PATCH 02/21] fixed typo

---
 mlserver/model.py | 1 +
 1 file changed, 1 insertion(+)

diff --git a/mlserver/model.py b/mlserver/model.py
index c89574f89..8296bc19e 100644
--- a/mlserver/model.py
+++ b/mlserver/model.py
@@ -247,6 +247,7 @@ def encode(
         response_output = encode_response_output(
             payload, request_output, self._outputs_index
         )
+
         if response_output:
             return response_output
 

From faa363b053eaedd45f618042bb7b724594194d23 Mon Sep 17 00:00:00 2001
From: Nanbo Liu <nliu@lambda001.america.striveworks.us>
Date: Tue, 21 Jan 2025 14:08:04 -0600
Subject: [PATCH 03/21] update codec name

---
 runtimes/huggingface/mlserver_huggingface/codecs/__init__.py | 4 ++--
 runtimes/huggingface/mlserver_huggingface/codecs/chariot.py  | 2 +-
 runtimes/huggingface/mlserver_huggingface/runtime.py         | 4 ++--
 3 files changed, 5 insertions(+), 5 deletions(-)

diff --git a/runtimes/huggingface/mlserver_huggingface/codecs/__init__.py b/runtimes/huggingface/mlserver_huggingface/codecs/__init__.py
index 5388077af..0f88ddad6 100644
--- a/runtimes/huggingface/mlserver_huggingface/codecs/__init__.py
+++ b/runtimes/huggingface/mlserver_huggingface/codecs/__init__.py
@@ -6,7 +6,7 @@
 from .conversation import HuggingfaceConversationCodec
 from .raw import RawCodec
 from .utils import EqualUtil
-from .chariot import ChariotModelOutputCodec
+from .chariot import ChariotImgModelOutputCodec
 __all__ = [
     "MultiInputRequestCodec",
     "HuggingfaceRequestCodec",
@@ -14,7 +14,7 @@
     "HuggingfaceSingleJSONCodec",
     "HuggingfaceListJSONCodec",
     "HuggingfaceConversationCodec",
-    "ChariotModelOutputCodec",
+    "ChariotImgModelOutputCodec",
     "NumpyListCodec",
     "RawCodec",
     "EqualUtil",
diff --git a/runtimes/huggingface/mlserver_huggingface/codecs/chariot.py b/runtimes/huggingface/mlserver_huggingface/codecs/chariot.py
index 18ff391d1..77969e53e 100644
--- a/runtimes/huggingface/mlserver_huggingface/codecs/chariot.py
+++ b/runtimes/huggingface/mlserver_huggingface/codecs/chariot.py
@@ -44,7 +44,7 @@ def get_chariot_seg_mask_from_hf_seg_output(seg_pred,class_int_to_str):
     return predictions
         
     
-class ChariotModelOutputCodec():
+class ChariotImgModelOutputCodec():
     """Encoder that converts HF model output to the standard Chariot model output 
     """
     @classmethod
diff --git a/runtimes/huggingface/mlserver_huggingface/runtime.py b/runtimes/huggingface/mlserver_huggingface/runtime.py
index 8f2ebfada..f500bf804 100644
--- a/runtimes/huggingface/mlserver_huggingface/runtime.py
+++ b/runtimes/huggingface/mlserver_huggingface/runtime.py
@@ -12,7 +12,7 @@
 
 from .settings import get_huggingface_settings
 from .common import load_pipeline_from_settings
-from .codecs import HuggingfaceRequestCodec,ChariotModelOutputCodec
+from .codecs import HuggingfaceRequestCodec,ChariotImgModelOutputCodec
 from .metadata import METADATA
 
 
@@ -47,7 +47,7 @@ async def predict(self, payload: InferenceRequest) -> InferenceResponse:
             args = [list(array_inputs)] + args
         predictions = self._model(*args, **kwargs)
         if self.hf_settings.task in ["image-classification","image-segmentation","object-detection"]:   
-            predictions = ChariotModelOutputCodec.encode_output(predictions,task_type=self.hf_settings.task,pipeline=self._model)
+            predictions = ChariotImgModelOutputCodec.encode_output(predictions,task_type=self.hf_settings.task,pipeline=self._model)
         response =self.encode_response(
             payload=predictions, default_codec=HuggingfaceRequestCodec
         )

From 3361e03d9b1f693b9ca53271423e773715db0b23 Mon Sep 17 00:00:00 2001
From: Nanbo Liu <nliu@lambda001.america.striveworks.us>
Date: Tue, 21 Jan 2025 14:13:18 -0600
Subject: [PATCH 04/21] black them

---
 .../mlserver_huggingface/codecs/__init__.py   |  1 +
 .../mlserver_huggingface/codecs/chariot.py    | 81 +++++++++++--------
 .../mlserver_huggingface/runtime.py           | 14 +++-
 3 files changed, 57 insertions(+), 39 deletions(-)

diff --git a/runtimes/huggingface/mlserver_huggingface/codecs/__init__.py b/runtimes/huggingface/mlserver_huggingface/codecs/__init__.py
index 0f88ddad6..c4004d3d6 100644
--- a/runtimes/huggingface/mlserver_huggingface/codecs/__init__.py
+++ b/runtimes/huggingface/mlserver_huggingface/codecs/__init__.py
@@ -7,6 +7,7 @@
 from .raw import RawCodec
 from .utils import EqualUtil
 from .chariot import ChariotImgModelOutputCodec
+
 __all__ = [
     "MultiInputRequestCodec",
     "HuggingfaceRequestCodec",
diff --git a/runtimes/huggingface/mlserver_huggingface/codecs/chariot.py b/runtimes/huggingface/mlserver_huggingface/codecs/chariot.py
index 77969e53e..cdcbefaaf 100644
--- a/runtimes/huggingface/mlserver_huggingface/codecs/chariot.py
+++ b/runtimes/huggingface/mlserver_huggingface/codecs/chariot.py
@@ -1,15 +1,18 @@
 import numpy as np
 
+
 def is_list_of_dicts(var):
-    '''Check if a variable is a list of dicts'''
+    """Check if a variable is a list of dicts"""
     if not isinstance(var, list):
         return False
     for item in var:
         if not isinstance(item, dict):
             return False
     return True
+
+
 def get_det_dict_from_hf_obj_detect(obj_detect):
-    '''Convert hf object detection output to standard chariot object detection output'''
+    """Convert hf object detection output to standard chariot object detection output"""
     det_dict = {
         "num_detections": 0,
         "detection_classes": [],
@@ -17,47 +20,54 @@ def get_det_dict_from_hf_obj_detect(obj_detect):
         "detection_scores": [],
     }
     for det in obj_detect:
-        conf, cls = det["score"],det["label"]
-        x1, y1, x2, y2  = det["box"]["xmin"],det["box"]["ymin"],det["box"]["xmax"],det["box"]["ymax"]
+        conf, cls = det["score"], det["label"]
+        x1, y1, x2, y2 = (
+            det["box"]["xmin"],
+            det["box"]["ymin"],
+            det["box"]["xmax"],
+            det["box"]["ymax"],
+        )
         det_dict["num_detections"] += 1
         det_dict["detection_classes"].append(cls)
         det_dict["detection_scores"].append(conf)
         det_dict["detection_boxes"].append([x1, y1, x2, y2])
     return det_dict
-def get_chariot_seg_mask_from_hf_seg_output(seg_pred,class_int_to_str):
-    '''Convert hf segmentation output to standard chariot segmentation output'''
-    img_w,img_h = seg_pred[0]["mask"].size
-    if class_int_to_str[0]!="background":
+
+
+def get_chariot_seg_mask_from_hf_seg_output(seg_pred, class_int_to_str):
+    """Convert hf segmentation output to standard chariot segmentation output"""
+    img_w, img_h = seg_pred[0]["mask"].size
+    if class_int_to_str[0] != "background":
         # Let class_int 0 be background and increment all the other class_int by 1
-        class_str_to_int = {v:k+1 for k,v in class_int_to_str.items()}
+        class_str_to_int = {v: k + 1 for k, v in class_int_to_str.items()}
     else:
-        class_str_to_int = {v:k for k,v in class_int_to_str.items()}
+        class_str_to_int = {v: k for k, v in class_int_to_str.items()}
     # Create an empty mask
     combined_mask = np.zeros((img_h, img_w), dtype=np.uint8)
     for i in seg_pred:
         # Convert mask from PIL image to numpy array
         mask = np.array(i["mask"])
         class_str = i["label"]
-        class_int = class_str_to_int[class_str] 
-        combined_mask[np.where(mask>0)]=class_int
-    predictions= [combined_mask]
+        class_int = class_str_to_int[class_str]
+        combined_mask[np.where(mask > 0)] = class_int
+    predictions = [combined_mask]
     return predictions
-        
-    
-class ChariotImgModelOutputCodec():
-    """Encoder that converts HF model output to the standard Chariot model output 
-    """
+
+
+class ChariotImgModelOutputCodec:
+    """Encoder that converts HF model output to the standard Chariot model output"""
+
     @classmethod
-    def encode_output(cls, predictions,task_type,pipeline):
+    def encode_output(cls, predictions, task_type, pipeline):
         if task_type == "image-classification":
             if all([is_list_of_dicts(p) for p in predictions]):
                 # get Top-1 predicted class
-                # convert HF output: [[{"label": "tabby, tabby cat", "score": 0.94}, 
+                # convert HF output: [[{"label": "tabby, tabby cat", "score": 0.94},
                 #                      {"label": "tiger cat", "score": 0.04},
                 #                      {"label": "Egyptian cat", "score": 0.02}]]
                 # to standard Chariot output: ["tabby, tabby cat"]
                 predictions = [p[0]["label"] for p in predictions]
-        
+
         elif task_type == "object-detection":
             if is_list_of_dicts(predictions):
                 # convert HF output: [{'score': 0.9897010326385498,
@@ -72,21 +82,22 @@ def encode_output(cls, predictions,task_type,pipeline):
                 #                             "detection_scores":[0.9897010326385498,0.9896764159202576],
                 #                             "detection_boxes":[[53,313,697,986],
                 #                                                [974,221,1526,1071]]}]
-                predictions = get_det_dict_from_hf_obj_detect(predictions) 
-        
+                predictions = get_det_dict_from_hf_obj_detect(predictions)
+
         elif task_type == "image-segmentation":
             if is_list_of_dicts(predictions):
-                    # convert HF output: [{"score": None,
-                    #                      "label": "wall",
-                    #                      "mask": <PIL.Image.Image image mode=L size=5362x3016>},
-                    #                     {"score": None,
-                    #                      "label": "floor",
-                    #                      "mask": <PIL.Image.Image image mode=L size=5362x3016>}]
+                # convert HF output: [{"score": None,
+                #                      "label": "wall",
+                #                      "mask": <PIL.Image.Image image mode=L size=5362x3016>},
+                #                     {"score": None,
+                #                      "label": "floor",
+                #                      "mask": <PIL.Image.Image image mode=L size=5362x3016>}]
 
-                    # to standard Chariot output: [[0,0,...,0],...,[0,0,0,...,0]] 
-                    # 2d array with size of the original image. Each pixel is a class int
-                    # Background uses class_int 0 
-                    class_int_to_str=pipeline.model.config.id2label
-                    predictions = get_chariot_seg_mask_from_hf_seg_output(predictions,class_int_to_str) 
+                # to standard Chariot output: [[0,0,...,0],...,[0,0,0,...,0]]
+                # 2d array with size of the original image. Each pixel is a class int
+                # Background uses class_int 0
+                class_int_to_str = pipeline.model.config.id2label
+                predictions = get_chariot_seg_mask_from_hf_seg_output(
+                    predictions, class_int_to_str
+                )
         return predictions
-    
diff --git a/runtimes/huggingface/mlserver_huggingface/runtime.py b/runtimes/huggingface/mlserver_huggingface/runtime.py
index f500bf804..415432af2 100644
--- a/runtimes/huggingface/mlserver_huggingface/runtime.py
+++ b/runtimes/huggingface/mlserver_huggingface/runtime.py
@@ -12,7 +12,7 @@
 
 from .settings import get_huggingface_settings
 from .common import load_pipeline_from_settings
-from .codecs import HuggingfaceRequestCodec,ChariotImgModelOutputCodec
+from .codecs import HuggingfaceRequestCodec, ChariotImgModelOutputCodec
 from .metadata import METADATA
 
 
@@ -46,9 +46,15 @@ async def predict(self, payload: InferenceRequest) -> InferenceResponse:
         if array_inputs:
             args = [list(array_inputs)] + args
         predictions = self._model(*args, **kwargs)
-        if self.hf_settings.task in ["image-classification","image-segmentation","object-detection"]:   
-            predictions = ChariotImgModelOutputCodec.encode_output(predictions,task_type=self.hf_settings.task,pipeline=self._model)
-        response =self.encode_response(
+        if self.hf_settings.task in [
+            "image-classification",
+            "image-segmentation",
+            "object-detection",
+        ]:
+            predictions = ChariotImgModelOutputCodec.encode_output(
+                predictions, task_type=self.hf_settings.task, pipeline=self._model
+            )
+        response = self.encode_response(
             payload=predictions, default_codec=HuggingfaceRequestCodec
         )
         return response

From 58dd4544c41fbe60fe7c60ee0afae90a5f4a8ed3 Mon Sep 17 00:00:00 2001
From: Nanbo Liu <nliu@lambda001.america.striveworks.us>
Date: Tue, 21 Jan 2025 14:21:22 -0600
Subject: [PATCH 05/21] fixed lint

---
 .../huggingface/mlserver_huggingface/codecs/chariot.py | 10 ++++++----
 runtimes/huggingface/mlserver_huggingface/common.py    |  8 +++++---
 runtimes/huggingface/mlserver_huggingface/runtime.py   |  1 -
 3 files changed, 11 insertions(+), 8 deletions(-)

diff --git a/runtimes/huggingface/mlserver_huggingface/codecs/chariot.py b/runtimes/huggingface/mlserver_huggingface/codecs/chariot.py
index cdcbefaaf..352c13d3e 100644
--- a/runtimes/huggingface/mlserver_huggingface/codecs/chariot.py
+++ b/runtimes/huggingface/mlserver_huggingface/codecs/chariot.py
@@ -72,10 +72,12 @@ def encode_output(cls, predictions, task_type, pipeline):
             if is_list_of_dicts(predictions):
                 # convert HF output: [{'score': 0.9897010326385498,
                 #                       'label': 'cat',
-                #                       'box': {'xmin': 53, 'ymin': 313, 'xmax': 697, 'ymax': 986}},
+                #                       'box': {'xmin': 53, 'ymin': 313,
+                #                               'xmax': 697, 'ymax': 986}},
                 #                       {'score': 0.9896764159202576,
                 #                       'label': 'cat',
-                #                       'box': {'xmin': 974, 'ymin': 221, 'xmax': 1526, 'ymax': 1071}}]
+                #                       'box': {'xmin': 974, 'ymin': 221,
+                #                               'xmax': 1526, 'ymax': 1071}}]
 
                 # to standard Chariot output: [{"num_detections":2,
                 #                             "detection_classes":["cat","cat"],
@@ -88,10 +90,10 @@ def encode_output(cls, predictions, task_type, pipeline):
             if is_list_of_dicts(predictions):
                 # convert HF output: [{"score": None,
                 #                      "label": "wall",
-                #                      "mask": <PIL.Image.Image image mode=L size=5362x3016>},
+                #                      "mask": <PIL.Image.Image>},
                 #                     {"score": None,
                 #                      "label": "floor",
-                #                      "mask": <PIL.Image.Image image mode=L size=5362x3016>}]
+                #                      "mask": <PIL.Image.Image>}]
 
                 # to standard Chariot output: [[0,0,...,0],...,[0,0,0,...,0]]
                 # 2d array with size of the original image. Each pixel is a class int
diff --git a/runtimes/huggingface/mlserver_huggingface/common.py b/runtimes/huggingface/mlserver_huggingface/common.py
index 84fd53cd4..67ca70e78 100644
--- a/runtimes/huggingface/mlserver_huggingface/common.py
+++ b/runtimes/huggingface/mlserver_huggingface/common.py
@@ -47,7 +47,8 @@ def load_pipeline_from_settings(
             torch.set_num_interop_threads(hf_settings.inter_op_threads)
         if hf_settings.intra_op_threads is not None:
             torch.set_num_threads(hf_settings.intra_op_threads)
-    # HF pipeline would automatically load tokenizer from model directory if no tokenizer is provided in the config
+    # If no tokenizer is provided in the config
+    # HF pipeline would automatically load tokenizer from model directory
     hf_pipeline = pipeline(
         hf_settings.task_name,
         model=model,
@@ -65,10 +66,11 @@ def load_pipeline_from_settings(
             if not hf_pipeline.tokenizer.pad_token_id:
                 eos_token_id = model.config.eos_token_id  # type: ignore
                 if eos_token_id:
-                    hf_pipeline.tokenizer.pad_token_id = [str(eos_token_id)]  # type: ignore
+                    hf_pipeline.tokenizer.pad_token_id = [str(eos_token_id)]
                 else:
                     logger.warning(
-                        "Model has neither pad_token or eos_token, setting batch size to 1"
+                        "Model has neither pad_token or eos_token, \
+                        setting batch size to 1"
                     )
                     hf_pipeline._batch_size = 1
 
diff --git a/runtimes/huggingface/mlserver_huggingface/runtime.py b/runtimes/huggingface/mlserver_huggingface/runtime.py
index 415432af2..24b0d5c0d 100644
--- a/runtimes/huggingface/mlserver_huggingface/runtime.py
+++ b/runtimes/huggingface/mlserver_huggingface/runtime.py
@@ -1,7 +1,6 @@
 import asyncio
 import torch
 
-from mlserver.codecs.lists import is_list_of
 from mlserver.model import MLModel
 from mlserver.settings import ModelSettings
 from mlserver.logging import logger

From 6996e816bfcb0ed8266ce5927f46b4e9b2886f3d Mon Sep 17 00:00:00 2001
From: Nanbo Liu <nliu@lambda001.america.striveworks.us>
Date: Tue, 21 Jan 2025 16:31:38 -0600
Subject: [PATCH 06/21] added chariotoutputcodec unittests

---
 .../mlserver_huggingface/codecs/chariot.py    | 26 +++----
 .../mlserver_huggingface/runtime.py           |  4 +-
 .../tests/test_codecs/test_chariot.py         | 75 +++++++++++++++++++
 3 files changed, 90 insertions(+), 15 deletions(-)
 create mode 100644 runtimes/huggingface/tests/test_codecs/test_chariot.py

diff --git a/runtimes/huggingface/mlserver_huggingface/codecs/chariot.py b/runtimes/huggingface/mlserver_huggingface/codecs/chariot.py
index 352c13d3e..11fd0b00b 100644
--- a/runtimes/huggingface/mlserver_huggingface/codecs/chariot.py
+++ b/runtimes/huggingface/mlserver_huggingface/codecs/chariot.py
@@ -50,7 +50,7 @@ def get_chariot_seg_mask_from_hf_seg_output(seg_pred, class_int_to_str):
         class_str = i["label"]
         class_int = class_str_to_int[class_str]
         combined_mask[np.where(mask > 0)] = class_int
-    predictions = [combined_mask]
+    predictions = [combined_mask.tolist()]
     return predictions
 
 
@@ -58,7 +58,7 @@ class ChariotImgModelOutputCodec:
     """Encoder that converts HF model output to the standard Chariot model output"""
 
     @classmethod
-    def encode_output(cls, predictions, task_type, pipeline):
+    def encode_output(cls, predictions, task_type, class_int_to_str):
         if task_type == "image-classification":
             if all([is_list_of_dicts(p) for p in predictions]):
                 # get Top-1 predicted class
@@ -70,20 +70,20 @@ def encode_output(cls, predictions, task_type, pipeline):
 
         elif task_type == "object-detection":
             if is_list_of_dicts(predictions):
-                # convert HF output: [{'score': 0.9897010326385498,
-                #                       'label': 'cat',
-                #                       'box': {'xmin': 53, 'ymin': 313,
-                #                               'xmax': 697, 'ymax': 986}},
-                #                       {'score': 0.9896764159202576,
-                #                       'label': 'cat',
-                #                       'box': {'xmin': 974, 'ymin': 221,
-                #                               'xmax': 1526, 'ymax': 1071}}]
+                # convert HF output: [{"score": 0.9897010326385498,
+                #                       "label": 'cat',
+                #                       "box": {"xmin": 53, "ymin": 313,
+                #                               "xmax": 697, "ymax": 986}},
+                #                       {"score": 0.9896764159202576,
+                #                       "label": "cat",
+                #                       "box": {"xmin": 974, "ymin": 221,
+                #                               "xmax": 1526, "ymax": 1071}}]
 
-                # to standard Chariot output: [{"num_detections":2,
+                # to standard Chariot output: {"num_detections":2,
                 #                             "detection_classes":["cat","cat"],
                 #                             "detection_scores":[0.9897010326385498,0.9896764159202576],
                 #                             "detection_boxes":[[53,313,697,986],
-                #                                                [974,221,1526,1071]]}]
+                #                                                [974,221,1526,1071]]}
                 predictions = get_det_dict_from_hf_obj_detect(predictions)
 
         elif task_type == "image-segmentation":
@@ -94,11 +94,9 @@ def encode_output(cls, predictions, task_type, pipeline):
                 #                     {"score": None,
                 #                      "label": "floor",
                 #                      "mask": <PIL.Image.Image>}]
-
                 # to standard Chariot output: [[0,0,...,0],...,[0,0,0,...,0]]
                 # 2d array with size of the original image. Each pixel is a class int
                 # Background uses class_int 0
-                class_int_to_str = pipeline.model.config.id2label
                 predictions = get_chariot_seg_mask_from_hf_seg_output(
                     predictions, class_int_to_str
                 )
diff --git a/runtimes/huggingface/mlserver_huggingface/runtime.py b/runtimes/huggingface/mlserver_huggingface/runtime.py
index 24b0d5c0d..35771c7b8 100644
--- a/runtimes/huggingface/mlserver_huggingface/runtime.py
+++ b/runtimes/huggingface/mlserver_huggingface/runtime.py
@@ -51,7 +51,9 @@ async def predict(self, payload: InferenceRequest) -> InferenceResponse:
             "object-detection",
         ]:
             predictions = ChariotImgModelOutputCodec.encode_output(
-                predictions, task_type=self.hf_settings.task, pipeline=self._model
+                predictions,
+                task_type=self.hf_settings.task,
+                class_int_to_str=self._model.model.config.id2label,
             )
         response = self.encode_response(
             payload=predictions, default_codec=HuggingfaceRequestCodec
diff --git a/runtimes/huggingface/tests/test_codecs/test_chariot.py b/runtimes/huggingface/tests/test_codecs/test_chariot.py
new file mode 100644
index 000000000..26dc2ffd7
--- /dev/null
+++ b/runtimes/huggingface/tests/test_codecs/test_chariot.py
@@ -0,0 +1,75 @@
+import pytest
+from mlserver_huggingface.codecs import ChariotImgModelOutputCodec
+from PIL import Image
+import numpy as np
+
+
+@pytest.mark.parametrize(
+    "task_type,class_int_to_str, hf_prediction,expected_chariot_output",
+    [
+        (
+            "image-classification",
+            None,
+            [
+                [
+                    {"label": "tabby, tabby cat", "score": 0.94},
+                    {"label": "tiger cat", "score": 0.04},
+                    {"label": "Egyptian cat", "score": 0.02},
+                ]
+            ],
+            ["tabby, tabby cat"],
+        ),
+        (
+            "object-detection",
+            None,
+            [
+                {
+                    "score": 0.9897010326385498,
+                    "label": "cat",
+                    "box": {"xmin": 53, "ymin": 313, "xmax": 697, "ymax": 986},
+                },
+                {
+                    "score": 0.9896764159202576,
+                    "label": "cat",
+                    "box": {"xmin": 974, "ymin": 221, "xmax": 1526, "ymax": 1071},
+                },
+            ],
+            {
+                "num_detections": 2,
+                "detection_classes": ["cat", "cat"],
+                "detection_scores": [0.9897010326385498, 0.9896764159202576],
+                "detection_boxes": [[53, 313, 697, 986], [974, 221, 1526, 1071]],
+            },
+        ),
+        (
+            "image-segmentation",
+            {0: "class_a", 1: "class_b"},
+            [
+                {
+                    "score": None,
+                    "label": "class_a",
+                    "mask": Image.fromarray(
+                        np.array([[0, 0, 0], [0, 255, 255], [0, 0, 0]]).astype("uint8"),
+                        mode="L",
+                    ),
+                },
+                {
+                    "score": None,
+                    "label": "class_b",
+                    "mask": Image.fromarray(
+                        np.array([[0, 0, 0], [0, 0, 0], [0, 255, 0]]).astype("uint8"),
+                        mode="L",
+                    ),
+                },
+            ],
+            [[[0, 0, 0], [0, 1, 1], [0, 2, 0]]],
+        ),
+    ],
+)
+def test_encode_input(
+    task_type, class_int_to_str, hf_prediction, expected_chariot_output
+):
+    chariot_output = ChariotImgModelOutputCodec.encode_output(
+        hf_prediction, task_type, class_int_to_str
+    )
+    assert chariot_output == expected_chariot_output

From 6921482a7bfc8a05861fad48607d765c6cc251df Mon Sep 17 00:00:00 2001
From: Nanbo Liu <nliu@lambda001.america.striveworks.us>
Date: Wed, 22 Jan 2025 11:13:34 -0600
Subject: [PATCH 07/21] fixed issue george mentioned

---
 .../mlserver_huggingface/codecs/chariot.py    |  22 +--
 .../mlserver_huggingface/test.ipynb           | 173 ++++++++++++++++++
 .../tests/test_codecs/test_chariot.py         |  16 +-
 3 files changed, 194 insertions(+), 17 deletions(-)
 create mode 100644 runtimes/huggingface/mlserver_huggingface/test.ipynb

diff --git a/runtimes/huggingface/mlserver_huggingface/codecs/chariot.py b/runtimes/huggingface/mlserver_huggingface/codecs/chariot.py
index 11fd0b00b..198663d80 100644
--- a/runtimes/huggingface/mlserver_huggingface/codecs/chariot.py
+++ b/runtimes/huggingface/mlserver_huggingface/codecs/chariot.py
@@ -21,29 +21,25 @@ def get_det_dict_from_hf_obj_detect(obj_detect):
     }
     for det in obj_detect:
         conf, cls = det["score"], det["label"]
-        x1, y1, x2, y2 = (
-            det["box"]["xmin"],
+        y1, x1, y2, x2 = (
             det["box"]["ymin"],
-            det["box"]["xmax"],
+            det["box"]["xmin"],
             det["box"]["ymax"],
+            det["box"]["xmax"],
         )
         det_dict["num_detections"] += 1
         det_dict["detection_classes"].append(cls)
         det_dict["detection_scores"].append(conf)
-        det_dict["detection_boxes"].append([x1, y1, x2, y2])
+        det_dict["detection_boxes"].append([y1, x1, y2, x2])
     return det_dict
 
 
 def get_chariot_seg_mask_from_hf_seg_output(seg_pred, class_int_to_str):
     """Convert hf segmentation output to standard chariot segmentation output"""
-    img_w, img_h = seg_pred[0]["mask"].size
-    if class_int_to_str[0] != "background":
-        # Let class_int 0 be background and increment all the other class_int by 1
-        class_str_to_int = {v: k + 1 for k, v in class_int_to_str.items()}
-    else:
-        class_str_to_int = {v: k for k, v in class_int_to_str.items()}
+    mask_shape =np.array(seg_pred[0]["mask"]).shape
+    class_str_to_int = {v: k for k, v in class_int_to_str.items()}
     # Create an empty mask
-    combined_mask = np.zeros((img_h, img_w), dtype=np.uint8)
+    combined_mask = np.full(mask_shape,None)
     for i in seg_pred:
         # Convert mask from PIL image to numpy array
         mask = np.array(i["mask"])
@@ -82,8 +78,8 @@ def encode_output(cls, predictions, task_type, class_int_to_str):
                 # to standard Chariot output: {"num_detections":2,
                 #                             "detection_classes":["cat","cat"],
                 #                             "detection_scores":[0.9897010326385498,0.9896764159202576],
-                #                             "detection_boxes":[[53,313,697,986],
-                #                                                [974,221,1526,1071]]}
+                #                             "detection_boxes":[[313,53,986,697],
+                #                                                [221,974,1071,1562]]}
                 predictions = get_det_dict_from_hf_obj_detect(predictions)
 
         elif task_type == "image-segmentation":
diff --git a/runtimes/huggingface/mlserver_huggingface/test.ipynb b/runtimes/huggingface/mlserver_huggingface/test.ipynb
new file mode 100644
index 000000000..8350de504
--- /dev/null
+++ b/runtimes/huggingface/mlserver_huggingface/test.ipynb
@@ -0,0 +1,173 @@
+{
+ "cells": [
+  {
+   "cell_type": "code",
+   "execution_count": 2,
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "2025-01-22 10:43:47.182194: E external/local_xla/xla/stream_executor/cuda/cuda_fft.cc:479] Unable to register cuFFT factory: Attempting to register factory for plugin cuFFT when one has already been registered\n",
+      "2025-01-22 10:43:47.209070: E external/local_xla/xla/stream_executor/cuda/cuda_dnn.cc:10575] Unable to register cuDNN factory: Attempting to register factory for plugin cuDNN when one has already been registered\n",
+      "2025-01-22 10:43:47.209112: E external/local_xla/xla/stream_executor/cuda/cuda_blas.cc:1442] Unable to register cuBLAS factory: Attempting to register factory for plugin cuBLAS when one has already been registered\n",
+      "2025-01-22 10:43:47.227000: I tensorflow/core/platform/cpu_feature_guard.cc:210] This TensorFlow binary is optimized to use available CPU instructions in performance-critical operations.\n",
+      "To enable the following instructions: AVX2 AVX512F FMA, in other operations, rebuild TensorFlow with the appropriate compiler flags.\n",
+      "2025-01-22 10:43:48.340959: W tensorflow/compiler/tf2tensorrt/utils/py_utils.cc:38] TF-TRT Warning: Could not find TensorRT\n"
+     ]
+    }
+   ],
+   "source": [
+    "import transformers\n",
+    "from transformers import pipeline"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 10,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "p = pipeline(\"image-segmentation\",\"facebook/mask2former-swin-large-cityscapes-semantic\")"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 11,
+   "metadata": {},
+   "outputs": [
+    {
+     "data": {
+      "text/plain": [
+       "{0: 'road',\n",
+       " 1: 'sidewalk',\n",
+       " 2: 'building',\n",
+       " 3: 'wall',\n",
+       " 4: 'fence',\n",
+       " 5: 'pole',\n",
+       " 6: 'traffic light',\n",
+       " 7: 'traffic sign',\n",
+       " 8: 'vegetation',\n",
+       " 9: 'terrain',\n",
+       " 10: 'sky',\n",
+       " 11: 'person',\n",
+       " 12: 'rider',\n",
+       " 13: 'car',\n",
+       " 14: 'truck',\n",
+       " 15: 'bus',\n",
+       " 16: 'train',\n",
+       " 17: 'motorcycle',\n",
+       " 18: 'bicycle'}"
+      ]
+     },
+     "execution_count": 11,
+     "metadata": {},
+     "output_type": "execute_result"
+    }
+   ],
+   "source": [
+    "p.model.config.id2label"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 12,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "import numpy as np"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 16,
+   "metadata": {},
+   "outputs": [
+    {
+     "data": {
+      "text/plain": [
+       "array([[None, None, None],\n",
+       "       [None, None, None]], dtype=object)"
+      ]
+     },
+     "execution_count": 16,
+     "metadata": {},
+     "output_type": "execute_result"
+    }
+   ],
+   "source": [
+    "np.full((2,3),None)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 17,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "from PIL import Image"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 18,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "a=Image.fromarray(\n",
+    "                        np.array([[0, 0, 0], [0, 255, 255], [0, 0, 0],[0,0,0]]).astype(\"uint8\"),\n",
+    "                        mode=\"L\",\n",
+    "                    )"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 19,
+   "metadata": {},
+   "outputs": [
+    {
+     "data": {
+      "text/plain": [
+       "(3, 4)"
+      ]
+     },
+     "execution_count": 19,
+     "metadata": {},
+     "output_type": "execute_result"
+    }
+   ],
+   "source": [
+    "a.size"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": []
+  }
+ ],
+ "metadata": {
+  "kernelspec": {
+   "display_name": "py311",
+   "language": "python",
+   "name": "python3"
+  },
+  "language_info": {
+   "codemirror_mode": {
+    "name": "ipython",
+    "version": 3
+   },
+   "file_extension": ".py",
+   "mimetype": "text/x-python",
+   "name": "python",
+   "nbconvert_exporter": "python",
+   "pygments_lexer": "ipython3",
+   "version": "3.11.11"
+  }
+ },
+ "nbformat": 4,
+ "nbformat_minor": 2
+}
diff --git a/runtimes/huggingface/tests/test_codecs/test_chariot.py b/runtimes/huggingface/tests/test_codecs/test_chariot.py
index 26dc2ffd7..43bd8b785 100644
--- a/runtimes/huggingface/tests/test_codecs/test_chariot.py
+++ b/runtimes/huggingface/tests/test_codecs/test_chariot.py
@@ -38,16 +38,24 @@
                 "num_detections": 2,
                 "detection_classes": ["cat", "cat"],
                 "detection_scores": [0.9897010326385498, 0.9896764159202576],
-                "detection_boxes": [[53, 313, 697, 986], [974, 221, 1526, 1071]],
+                "detection_boxes": [[313, 53, 986, 697], [221, 974, 1071, 1526]],
             },
         ),
         (
             "image-segmentation",
-            {0: "class_a", 1: "class_b"},
+            {0: "class_0", 1: "class_1",2:"class_2"},
             [
                 {
                     "score": None,
-                    "label": "class_a",
+                    "label": "class_0",
+                    "mask": Image.fromarray(
+                        np.array([[255, 255, 255], [255, 0, 0], [255, 0, 255]]).astype("uint8"),
+                        mode="L",
+                    ),
+                },
+                {
+                    "score": None,
+                    "label": "class_1",
                     "mask": Image.fromarray(
                         np.array([[0, 0, 0], [0, 255, 255], [0, 0, 0]]).astype("uint8"),
                         mode="L",
@@ -55,7 +63,7 @@
                 },
                 {
                     "score": None,
-                    "label": "class_b",
+                    "label": "class_2",
                     "mask": Image.fromarray(
                         np.array([[0, 0, 0], [0, 0, 0], [0, 255, 0]]).astype("uint8"),
                         mode="L",

From d342a5d2e68fa4779a6c082c4f56194536be6480 Mon Sep 17 00:00:00 2001
From: Nanbo Liu <nliu@lambda001.america.striveworks.us>
Date: Wed, 22 Jan 2025 15:15:41 -0600
Subject: [PATCH 08/21] added predict_proba for image_classification

---
 .../mlserver_huggingface/codecs/chariot.py    |  90 +++++----
 .../mlserver_huggingface/runtime.py           |  29 ++-
 .../mlserver_huggingface/test.ipynb           | 173 ------------------
 .../tests/test_codecs/test_chariot.py         | 114 +++++++-----
 4 files changed, 151 insertions(+), 255 deletions(-)
 delete mode 100644 runtimes/huggingface/mlserver_huggingface/test.ipynb

diff --git a/runtimes/huggingface/mlserver_huggingface/codecs/chariot.py b/runtimes/huggingface/mlserver_huggingface/codecs/chariot.py
index 198663d80..4a19690f9 100644
--- a/runtimes/huggingface/mlserver_huggingface/codecs/chariot.py
+++ b/runtimes/huggingface/mlserver_huggingface/codecs/chariot.py
@@ -36,17 +36,17 @@ def get_det_dict_from_hf_obj_detect(obj_detect):
 
 def get_chariot_seg_mask_from_hf_seg_output(seg_pred, class_int_to_str):
     """Convert hf segmentation output to standard chariot segmentation output"""
-    mask_shape =np.array(seg_pred[0]["mask"]).shape
+    mask_shape = np.array(seg_pred[0]["mask"]).shape
     class_str_to_int = {v: k for k, v in class_int_to_str.items()}
     # Create an empty mask
-    combined_mask = np.full(mask_shape,None)
+    combined_mask = np.full(mask_shape, None)
     for i in seg_pred:
         # Convert mask from PIL image to numpy array
         mask = np.array(i["mask"])
         class_str = i["label"]
         class_int = class_str_to_int[class_str]
         combined_mask[np.where(mask > 0)] = class_int
-    predictions = [combined_mask.tolist()]
+    predictions = combined_mask.tolist()
     return predictions
 
 
@@ -54,46 +54,68 @@ class ChariotImgModelOutputCodec:
     """Encoder that converts HF model output to the standard Chariot model output"""
 
     @classmethod
-    def encode_output(cls, predictions, task_type, class_int_to_str):
+    def encode_output(
+        cls, predictions, task_type, class_int_to_str, predict_proba=False
+    ):
+        if is_list_of_dicts(predictions):
+            predictions = [predictions]
         if task_type == "image-classification":
-            if all([is_list_of_dicts(p) for p in predictions]):
+
+            if predict_proba:
+                # class_int_to_str: {0:"Egyptian cat",
+                #                    1:"tabby, tabby cat",
+                #                    2:"tiger cat"}
+                # convert HF output: [[{"label": "tabby, tabby cat", "score": 0.94},
+                #                      {"label": "tiger cat", "score": 0.04},
+                #                      {"label": "Egyptian cat", "score": 0.02}]]
+                # to standard Chariot probability output: [[0.02,0.94,0.04]]
+                # The probability scores are ordered by class id
+                num_labels = len(class_int_to_str)
+                class_to_proba = [
+                    {d["label"]: d["score"] for d in p} for p in predictions
+                ]
+                predictions = [
+                    [d.get(class_int_to_str[i]) for i in range(num_labels)]
+                    for d in class_to_proba
+                ]
+            else:
                 # get Top-1 predicted class
                 # convert HF output: [[{"label": "tabby, tabby cat", "score": 0.94},
                 #                      {"label": "tiger cat", "score": 0.04},
                 #                      {"label": "Egyptian cat", "score": 0.02}]]
                 # to standard Chariot output: ["tabby, tabby cat"]
                 predictions = [p[0]["label"] for p in predictions]
-
         elif task_type == "object-detection":
-            if is_list_of_dicts(predictions):
-                # convert HF output: [{"score": 0.9897010326385498,
-                #                       "label": 'cat',
-                #                       "box": {"xmin": 53, "ymin": 313,
-                #                               "xmax": 697, "ymax": 986}},
-                #                       {"score": 0.9896764159202576,
-                #                       "label": "cat",
-                #                       "box": {"xmin": 974, "ymin": 221,
-                #                               "xmax": 1526, "ymax": 1071}}]
 
-                # to standard Chariot output: {"num_detections":2,
-                #                             "detection_classes":["cat","cat"],
-                #                             "detection_scores":[0.9897010326385498,0.9896764159202576],
-                #                             "detection_boxes":[[313,53,986,697],
-                #                                                [221,974,1071,1562]]}
-                predictions = get_det_dict_from_hf_obj_detect(predictions)
+            # convert HF output: [[{"score": 0.9897010326385498,
+            #                       "label": 'cat',
+            #                       "box": {"xmin": 53, "ymin": 313,
+            #                               "xmax": 697, "ymax": 986}},
+            #                       {"score": 0.9896764159202576,
+            #                       "label": "cat",
+            #                       "box": {"xmin": 974, "ymin": 221,
+            #                               "xmax": 1526, "ymax": 1071}}]]
+
+            # to standard Chariot output: [{"num_detections":2,
+            #                             "detection_classes":["cat","cat"],
+            #                             "detection_scores":[0.9897010326385498,0.9896764159202576],
+            #                             "detection_boxes":[[313,53,986,697],
+            #                                                [221,974,1071,1562]]}]
+            predictions = [get_det_dict_from_hf_obj_detect(p) for p in predictions]
 
         elif task_type == "image-segmentation":
-            if is_list_of_dicts(predictions):
-                # convert HF output: [{"score": None,
-                #                      "label": "wall",
-                #                      "mask": <PIL.Image.Image>},
-                #                     {"score": None,
-                #                      "label": "floor",
-                #                      "mask": <PIL.Image.Image>}]
-                # to standard Chariot output: [[0,0,...,0],...,[0,0,0,...,0]]
-                # 2d array with size of the original image. Each pixel is a class int
-                # Background uses class_int 0
-                predictions = get_chariot_seg_mask_from_hf_seg_output(
-                    predictions, class_int_to_str
-                )
+
+            # convert HF output: [[{"score": None,
+            #                      "label": "wall",
+            #                      "mask": <PIL.Image.Image>},
+            #                     {"score": None,
+            #                      "label": "floor",
+            #                      "mask": <PIL.Image.Image>}]]
+            # to standard Chariot output: [[[0,0,...,0],...,[0,0,0,...,0]]]
+            # 2d array with size of the original image. Each pixel is a class int
+            # Background uses class_int 0
+            predictions = [
+                get_chariot_seg_mask_from_hf_seg_output(p, class_int_to_str)
+                for p in predictions
+            ]
         return predictions
diff --git a/runtimes/huggingface/mlserver_huggingface/runtime.py b/runtimes/huggingface/mlserver_huggingface/runtime.py
index 35771c7b8..8b5efc63d 100644
--- a/runtimes/huggingface/mlserver_huggingface/runtime.py
+++ b/runtimes/huggingface/mlserver_huggingface/runtime.py
@@ -14,6 +14,12 @@
 from .codecs import HuggingfaceRequestCodec, ChariotImgModelOutputCodec
 from .metadata import METADATA
 
+CHARIOT_IMAGE_TASK = [
+    "image-classification",
+    "image-segmentation",
+    "object-detection",
+]
+
 
 class HuggingFaceRuntime(MLModel):
     """Runtime class for specific Huggingface models"""
@@ -37,6 +43,20 @@ async def load(self) -> bool:
         return True
 
     async def predict(self, payload: InferenceRequest) -> InferenceResponse:
+        predict_proba = {
+            (
+                request_input.parameters.predict_proba
+                if request_input.parameters
+                else False
+            )
+            for request_input in payload.inputs
+        }
+        if len(predict_proba) > 1:
+            raise ValueError(
+                f"If processing a batch all 'predict_proba' must be the same \
+                but got 'predict_proba': {predict_proba}"
+            )
+        predict_proba = predict_proba.pop()
         # TODO: convert and validate?
         kwargs = HuggingfaceRequestCodec.decode_request(payload)
         args = kwargs.pop("args", [])
@@ -44,16 +64,15 @@ async def predict(self, payload: InferenceRequest) -> InferenceResponse:
         array_inputs = kwargs.pop("array_inputs", [])
         if array_inputs:
             args = [list(array_inputs)] + args
+        if predict_proba and self.hf_settings.task == "image-classification":
+            kwargs["top_k"] = self._model.model.config.num_labels
         predictions = self._model(*args, **kwargs)
-        if self.hf_settings.task in [
-            "image-classification",
-            "image-segmentation",
-            "object-detection",
-        ]:
+        if self.hf_settings.task in CHARIOT_IMAGE_TASK:
             predictions = ChariotImgModelOutputCodec.encode_output(
                 predictions,
                 task_type=self.hf_settings.task,
                 class_int_to_str=self._model.model.config.id2label,
+                predict_proba=predict_proba,
             )
         response = self.encode_response(
             payload=predictions, default_codec=HuggingfaceRequestCodec
diff --git a/runtimes/huggingface/mlserver_huggingface/test.ipynb b/runtimes/huggingface/mlserver_huggingface/test.ipynb
deleted file mode 100644
index 8350de504..000000000
--- a/runtimes/huggingface/mlserver_huggingface/test.ipynb
+++ /dev/null
@@ -1,173 +0,0 @@
-{
- "cells": [
-  {
-   "cell_type": "code",
-   "execution_count": 2,
-   "metadata": {},
-   "outputs": [
-    {
-     "name": "stderr",
-     "output_type": "stream",
-     "text": [
-      "2025-01-22 10:43:47.182194: E external/local_xla/xla/stream_executor/cuda/cuda_fft.cc:479] Unable to register cuFFT factory: Attempting to register factory for plugin cuFFT when one has already been registered\n",
-      "2025-01-22 10:43:47.209070: E external/local_xla/xla/stream_executor/cuda/cuda_dnn.cc:10575] Unable to register cuDNN factory: Attempting to register factory for plugin cuDNN when one has already been registered\n",
-      "2025-01-22 10:43:47.209112: E external/local_xla/xla/stream_executor/cuda/cuda_blas.cc:1442] Unable to register cuBLAS factory: Attempting to register factory for plugin cuBLAS when one has already been registered\n",
-      "2025-01-22 10:43:47.227000: I tensorflow/core/platform/cpu_feature_guard.cc:210] This TensorFlow binary is optimized to use available CPU instructions in performance-critical operations.\n",
-      "To enable the following instructions: AVX2 AVX512F FMA, in other operations, rebuild TensorFlow with the appropriate compiler flags.\n",
-      "2025-01-22 10:43:48.340959: W tensorflow/compiler/tf2tensorrt/utils/py_utils.cc:38] TF-TRT Warning: Could not find TensorRT\n"
-     ]
-    }
-   ],
-   "source": [
-    "import transformers\n",
-    "from transformers import pipeline"
-   ]
-  },
-  {
-   "cell_type": "code",
-   "execution_count": 10,
-   "metadata": {},
-   "outputs": [],
-   "source": [
-    "p = pipeline(\"image-segmentation\",\"facebook/mask2former-swin-large-cityscapes-semantic\")"
-   ]
-  },
-  {
-   "cell_type": "code",
-   "execution_count": 11,
-   "metadata": {},
-   "outputs": [
-    {
-     "data": {
-      "text/plain": [
-       "{0: 'road',\n",
-       " 1: 'sidewalk',\n",
-       " 2: 'building',\n",
-       " 3: 'wall',\n",
-       " 4: 'fence',\n",
-       " 5: 'pole',\n",
-       " 6: 'traffic light',\n",
-       " 7: 'traffic sign',\n",
-       " 8: 'vegetation',\n",
-       " 9: 'terrain',\n",
-       " 10: 'sky',\n",
-       " 11: 'person',\n",
-       " 12: 'rider',\n",
-       " 13: 'car',\n",
-       " 14: 'truck',\n",
-       " 15: 'bus',\n",
-       " 16: 'train',\n",
-       " 17: 'motorcycle',\n",
-       " 18: 'bicycle'}"
-      ]
-     },
-     "execution_count": 11,
-     "metadata": {},
-     "output_type": "execute_result"
-    }
-   ],
-   "source": [
-    "p.model.config.id2label"
-   ]
-  },
-  {
-   "cell_type": "code",
-   "execution_count": 12,
-   "metadata": {},
-   "outputs": [],
-   "source": [
-    "import numpy as np"
-   ]
-  },
-  {
-   "cell_type": "code",
-   "execution_count": 16,
-   "metadata": {},
-   "outputs": [
-    {
-     "data": {
-      "text/plain": [
-       "array([[None, None, None],\n",
-       "       [None, None, None]], dtype=object)"
-      ]
-     },
-     "execution_count": 16,
-     "metadata": {},
-     "output_type": "execute_result"
-    }
-   ],
-   "source": [
-    "np.full((2,3),None)"
-   ]
-  },
-  {
-   "cell_type": "code",
-   "execution_count": 17,
-   "metadata": {},
-   "outputs": [],
-   "source": [
-    "from PIL import Image"
-   ]
-  },
-  {
-   "cell_type": "code",
-   "execution_count": 18,
-   "metadata": {},
-   "outputs": [],
-   "source": [
-    "a=Image.fromarray(\n",
-    "                        np.array([[0, 0, 0], [0, 255, 255], [0, 0, 0],[0,0,0]]).astype(\"uint8\"),\n",
-    "                        mode=\"L\",\n",
-    "                    )"
-   ]
-  },
-  {
-   "cell_type": "code",
-   "execution_count": 19,
-   "metadata": {},
-   "outputs": [
-    {
-     "data": {
-      "text/plain": [
-       "(3, 4)"
-      ]
-     },
-     "execution_count": 19,
-     "metadata": {},
-     "output_type": "execute_result"
-    }
-   ],
-   "source": [
-    "a.size"
-   ]
-  },
-  {
-   "cell_type": "code",
-   "execution_count": null,
-   "metadata": {},
-   "outputs": [],
-   "source": []
-  }
- ],
- "metadata": {
-  "kernelspec": {
-   "display_name": "py311",
-   "language": "python",
-   "name": "python3"
-  },
-  "language_info": {
-   "codemirror_mode": {
-    "name": "ipython",
-    "version": 3
-   },
-   "file_extension": ".py",
-   "mimetype": "text/x-python",
-   "name": "python",
-   "nbconvert_exporter": "python",
-   "pygments_lexer": "ipython3",
-   "version": "3.11.11"
-  }
- },
- "nbformat": 4,
- "nbformat_minor": 2
-}
diff --git a/runtimes/huggingface/tests/test_codecs/test_chariot.py b/runtimes/huggingface/tests/test_codecs/test_chariot.py
index 43bd8b785..601d4676c 100644
--- a/runtimes/huggingface/tests/test_codecs/test_chariot.py
+++ b/runtimes/huggingface/tests/test_codecs/test_chariot.py
@@ -5,11 +5,12 @@
 
 
 @pytest.mark.parametrize(
-    "task_type,class_int_to_str, hf_prediction,expected_chariot_output",
+    "task_type,class_int_to_str,predict_proba, hf_prediction,expected_chariot_output",
     [
         (
             "image-classification",
             None,
+            False,
             [
                 [
                     {"label": "tabby, tabby cat", "score": 0.94},
@@ -19,65 +20,92 @@
             ],
             ["tabby, tabby cat"],
         ),
+        (
+            "image-classification",
+            {0: "Egyptian cat", 1: "tabby, tabby cat", 2: "tiger cat"},
+            True,
+            [
+                [
+                    {"label": "tabby, tabby cat", "score": 0.94},
+                    {"label": "tiger cat", "score": 0.04},
+                    {"label": "Egyptian cat", "score": 0.02},
+                ]
+            ],
+            [[0.02, 0.94, 0.04]],
+        ),
         (
             "object-detection",
             None,
+            False,
+            [
+                [
+                    {
+                        "score": 0.9897010326385498,
+                        "label": "cat",
+                        "box": {"xmin": 53, "ymin": 313, "xmax": 697, "ymax": 986},
+                    },
+                    {
+                        "score": 0.9896764159202576,
+                        "label": "cat",
+                        "box": {"xmin": 974, "ymin": 221, "xmax": 1526, "ymax": 1071},
+                    },
+                ]
+            ],
             [
                 {
-                    "score": 0.9897010326385498,
-                    "label": "cat",
-                    "box": {"xmin": 53, "ymin": 313, "xmax": 697, "ymax": 986},
-                },
-                {
-                    "score": 0.9896764159202576,
-                    "label": "cat",
-                    "box": {"xmin": 974, "ymin": 221, "xmax": 1526, "ymax": 1071},
-                },
+                    "num_detections": 2,
+                    "detection_classes": ["cat", "cat"],
+                    "detection_scores": [0.9897010326385498, 0.9896764159202576],
+                    "detection_boxes": [[313, 53, 986, 697], [221, 974, 1071, 1526]],
+                }
             ],
-            {
-                "num_detections": 2,
-                "detection_classes": ["cat", "cat"],
-                "detection_scores": [0.9897010326385498, 0.9896764159202576],
-                "detection_boxes": [[313, 53, 986, 697], [221, 974, 1071, 1526]],
-            },
         ),
         (
             "image-segmentation",
-            {0: "class_0", 1: "class_1",2:"class_2"},
+            {0: "class_0", 1: "class_1", 2: "class_2"},
+            False,
             [
-                {
-                    "score": None,
-                    "label": "class_0",
-                    "mask": Image.fromarray(
-                        np.array([[255, 255, 255], [255, 0, 0], [255, 0, 255]]).astype("uint8"),
-                        mode="L",
-                    ),
-                },
-                {
-                    "score": None,
-                    "label": "class_1",
-                    "mask": Image.fromarray(
-                        np.array([[0, 0, 0], [0, 255, 255], [0, 0, 0]]).astype("uint8"),
-                        mode="L",
-                    ),
-                },
-                {
-                    "score": None,
-                    "label": "class_2",
-                    "mask": Image.fromarray(
-                        np.array([[0, 0, 0], [0, 0, 0], [0, 255, 0]]).astype("uint8"),
-                        mode="L",
-                    ),
-                },
+                [
+                    {
+                        "score": None,
+                        "label": "class_0",
+                        "mask": Image.fromarray(
+                            np.array(
+                                [[255, 255, 255], [255, 0, 0], [255, 0, 255]]
+                            ).astype("uint8"),
+                            mode="L",
+                        ),
+                    },
+                    {
+                        "score": None,
+                        "label": "class_1",
+                        "mask": Image.fromarray(
+                            np.array([[0, 0, 0], [0, 255, 255], [0, 0, 0]]).astype(
+                                "uint8"
+                            ),
+                            mode="L",
+                        ),
+                    },
+                    {
+                        "score": None,
+                        "label": "class_2",
+                        "mask": Image.fromarray(
+                            np.array([[0, 0, 0], [0, 0, 0], [0, 255, 0]]).astype(
+                                "uint8"
+                            ),
+                            mode="L",
+                        ),
+                    },
+                ]
             ],
             [[[0, 0, 0], [0, 1, 1], [0, 2, 0]]],
         ),
     ],
 )
 def test_encode_input(
-    task_type, class_int_to_str, hf_prediction, expected_chariot_output
+    task_type, class_int_to_str, predict_proba, hf_prediction, expected_chariot_output
 ):
     chariot_output = ChariotImgModelOutputCodec.encode_output(
-        hf_prediction, task_type, class_int_to_str
+        hf_prediction, task_type, class_int_to_str, predict_proba
     )
     assert chariot_output == expected_chariot_output

From dc8d7db4820df85f5cd9214c7145cffd9fb71f9d Mon Sep 17 00:00:00 2001
From: Nanbo Liu <nliu@lambda001.america.striveworks.us>
Date: Wed, 22 Jan 2025 15:28:54 -0600
Subject: [PATCH 09/21] fixed lint error

---
 .../huggingface/mlserver_huggingface/runtime.py    | 14 +++++++-------
 1 file changed, 7 insertions(+), 7 deletions(-)

diff --git a/runtimes/huggingface/mlserver_huggingface/runtime.py b/runtimes/huggingface/mlserver_huggingface/runtime.py
index 8b5efc63d..44782ad87 100644
--- a/runtimes/huggingface/mlserver_huggingface/runtime.py
+++ b/runtimes/huggingface/mlserver_huggingface/runtime.py
@@ -44,13 +44,13 @@ async def load(self) -> bool:
 
     async def predict(self, payload: InferenceRequest) -> InferenceResponse:
         predict_proba = {
-            (
-                request_input.parameters.predict_proba
-                if request_input.parameters
-                else False
-            )
-            for request_input in payload.inputs
-        }
+                    (
+                        getattr(request_input.parameters,"predict_proba")
+                        if request_input.parameters
+                        else False
+                    )
+                    for request_input in payload.inputs
+                } 
         if len(predict_proba) > 1:
             raise ValueError(
                 f"If processing a batch all 'predict_proba' must be the same \

From 08c3a70447f379b37088b670a69c1bf1ed415222 Mon Sep 17 00:00:00 2001
From: Nanbo Liu <nliu@lambda001.america.striveworks.us>
Date: Wed, 22 Jan 2025 15:32:20 -0600
Subject: [PATCH 10/21] fixed lint error

---
 .../huggingface/mlserver_huggingface/runtime.py    | 14 +++++++-------
 1 file changed, 7 insertions(+), 7 deletions(-)

diff --git a/runtimes/huggingface/mlserver_huggingface/runtime.py b/runtimes/huggingface/mlserver_huggingface/runtime.py
index 44782ad87..60b433465 100644
--- a/runtimes/huggingface/mlserver_huggingface/runtime.py
+++ b/runtimes/huggingface/mlserver_huggingface/runtime.py
@@ -44,13 +44,13 @@ async def load(self) -> bool:
 
     async def predict(self, payload: InferenceRequest) -> InferenceResponse:
         predict_proba = {
-                    (
-                        getattr(request_input.parameters,"predict_proba")
-                        if request_input.parameters
-                        else False
-                    )
-                    for request_input in payload.inputs
-                } 
+            (
+                getattr(request_input.parameters, "predict_proba")
+                if request_input.parameters
+                else False
+            )
+            for request_input in payload.inputs
+        }
         if len(predict_proba) > 1:
             raise ValueError(
                 f"If processing a batch all 'predict_proba' must be the same \

From 13fe8488eb1a9dc175b7f8dc71844b2409cfe52f Mon Sep 17 00:00:00 2001
From: Nanbo Liu <nliu@lambda001.america.striveworks.us>
Date: Fri, 24 Jan 2025 14:16:02 -0600
Subject: [PATCH 11/21] update predict_proba inputrequest kwargs

---
 .../mlserver_huggingface/runtime.py           | 41 +++++++++++--------
 1 file changed, 23 insertions(+), 18 deletions(-)

diff --git a/runtimes/huggingface/mlserver_huggingface/runtime.py b/runtimes/huggingface/mlserver_huggingface/runtime.py
index 60b433465..409e86d3c 100644
--- a/runtimes/huggingface/mlserver_huggingface/runtime.py
+++ b/runtimes/huggingface/mlserver_huggingface/runtime.py
@@ -43,30 +43,14 @@ async def load(self) -> bool:
         return True
 
     async def predict(self, payload: InferenceRequest) -> InferenceResponse:
-        predict_proba = {
-            (
-                getattr(request_input.parameters, "predict_proba")
-                if request_input.parameters
-                else False
-            )
-            for request_input in payload.inputs
-        }
-        if len(predict_proba) > 1:
-            raise ValueError(
-                f"If processing a batch all 'predict_proba' must be the same \
-                but got 'predict_proba': {predict_proba}"
-            )
-        predict_proba = predict_proba.pop()
         # TODO: convert and validate?
         kwargs = HuggingfaceRequestCodec.decode_request(payload)
         args = kwargs.pop("args", [])
-
         array_inputs = kwargs.pop("array_inputs", [])
         if array_inputs:
             args = [list(array_inputs)] + args
-        if predict_proba and self.hf_settings.task == "image-classification":
-            kwargs["top_k"] = self._model.model.config.num_labels
-        predictions = self._model(*args, **kwargs)
+        predict_proba, predict_proba_kwargs = self.get_predict_proba_kwargs(payload)
+        predictions = self._model(*args, **kwargs, **predict_proba_kwargs)
         if self.hf_settings.task in CHARIOT_IMAGE_TASK:
             predictions = ChariotImgModelOutputCodec.encode_output(
                 predictions,
@@ -79,6 +63,27 @@ async def predict(self, payload: InferenceRequest) -> InferenceResponse:
         )
         return response
 
+    def get_predict_proba_kwargs(self, payload: InferenceRequest) -> InferenceResponse:
+        actions = {
+            (
+                getattr(request_input.parameters, "action", "predict")
+                if request_input.parameters
+                else "predict"
+            )
+            for request_input in payload.inputs
+        }
+        if len(actions) > 1:
+            raise ValueError(
+                f"If processing a batch all 'actions' must be the same \
+                but got 'actions': {actions}"
+            )
+        action = actions.pop()
+        predict_proba = action == "predict_proba"
+        predict_proba_kwargs = dict()
+        if predict_proba and self.hf_settings.task == "image-classification":
+            predict_proba_kwargs["top_k"] = self._model.model.config.num_labels
+        return predict_proba, predict_proba_kwargs
+
     async def unload(self) -> bool:
         # TODO: Free up Tensorflow's GPU memory
         is_torch = self._model.framework == "pt"

From 566bc202b0ac142b6ba46e01cbed5cdfed62c630 Mon Sep 17 00:00:00 2001
From: Nanbo Liu <nliu@lambda001.america.striveworks.us>
Date: Fri, 24 Jan 2025 14:19:48 -0600
Subject: [PATCH 12/21] fixed typing

---
 runtimes/huggingface/mlserver_huggingface/runtime.py | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/runtimes/huggingface/mlserver_huggingface/runtime.py b/runtimes/huggingface/mlserver_huggingface/runtime.py
index 409e86d3c..1c8fee624 100644
--- a/runtimes/huggingface/mlserver_huggingface/runtime.py
+++ b/runtimes/huggingface/mlserver_huggingface/runtime.py
@@ -1,6 +1,6 @@
 import asyncio
 import torch
-
+from typing import Any
 from mlserver.model import MLModel
 from mlserver.settings import ModelSettings
 from mlserver.logging import logger
@@ -63,7 +63,7 @@ async def predict(self, payload: InferenceRequest) -> InferenceResponse:
         )
         return response
 
-    def get_predict_proba_kwargs(self, payload: InferenceRequest) -> InferenceResponse:
+    def get_predict_proba_kwargs(self, payload: InferenceRequest) -> tuple[bool,dict[str, Any]]:
         actions = {
             (
                 getattr(request_input.parameters, "action", "predict")

From 691f081ee22018448f7d830990f95389c9a15988 Mon Sep 17 00:00:00 2001
From: Nanbo Liu <nliu@lambda001.america.striveworks.us>
Date: Fri, 24 Jan 2025 14:22:10 -0600
Subject: [PATCH 13/21] fixed lint

---
 runtimes/huggingface/mlserver_huggingface/runtime.py | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)

diff --git a/runtimes/huggingface/mlserver_huggingface/runtime.py b/runtimes/huggingface/mlserver_huggingface/runtime.py
index 1c8fee624..9d3094d10 100644
--- a/runtimes/huggingface/mlserver_huggingface/runtime.py
+++ b/runtimes/huggingface/mlserver_huggingface/runtime.py
@@ -63,7 +63,9 @@ async def predict(self, payload: InferenceRequest) -> InferenceResponse:
         )
         return response
 
-    def get_predict_proba_kwargs(self, payload: InferenceRequest) -> tuple[bool,dict[str, Any]]:
+    def get_predict_proba_kwargs(
+        self, payload: InferenceRequest
+    ) -> tuple[bool, dict[str, Any]]:
         actions = {
             (
                 getattr(request_input.parameters, "action", "predict")

From 8f30e0bb73aae9b2e0c7f4435cfdf87902f69b94 Mon Sep 17 00:00:00 2001
From: Nanbo Liu <nliu@lambda001.america.striveworks.us>
Date: Fri, 31 Jan 2025 15:21:20 -0600
Subject: [PATCH 14/21] added hf pytorch image models package: timm

---
 runtimes/huggingface/poetry.lock    | 155 +++++++++++++++++++++++++---
 runtimes/huggingface/pyproject.toml |   1 +
 2 files changed, 144 insertions(+), 12 deletions(-)

diff --git a/runtimes/huggingface/poetry.lock b/runtimes/huggingface/poetry.lock
index 7ffbfcb93..ba8c161c5 100644
--- a/runtimes/huggingface/poetry.lock
+++ b/runtimes/huggingface/poetry.lock
@@ -339,6 +339,10 @@ files = [
     {file = "Brotli-1.1.0-cp310-cp310-musllinux_1_1_i686.whl", hash = "sha256:a37b8f0391212d29b3a91a799c8e4a2855e0576911cdfb2515487e30e322253d"},
     {file = "Brotli-1.1.0-cp310-cp310-musllinux_1_1_ppc64le.whl", hash = "sha256:e84799f09591700a4154154cab9787452925578841a94321d5ee8fb9a9a328f0"},
     {file = "Brotli-1.1.0-cp310-cp310-musllinux_1_1_x86_64.whl", hash = "sha256:f66b5337fa213f1da0d9000bc8dc0cb5b896b726eefd9c6046f699b169c41b9e"},
+    {file = "Brotli-1.1.0-cp310-cp310-musllinux_1_2_aarch64.whl", hash = "sha256:5dab0844f2cf82be357a0eb11a9087f70c5430b2c241493fc122bb6f2bb0917c"},
+    {file = "Brotli-1.1.0-cp310-cp310-musllinux_1_2_i686.whl", hash = "sha256:e4fe605b917c70283db7dfe5ada75e04561479075761a0b3866c081d035b01c1"},
+    {file = "Brotli-1.1.0-cp310-cp310-musllinux_1_2_ppc64le.whl", hash = "sha256:1e9a65b5736232e7a7f91ff3d02277f11d339bf34099a56cdab6a8b3410a02b2"},
+    {file = "Brotli-1.1.0-cp310-cp310-musllinux_1_2_x86_64.whl", hash = "sha256:58d4b711689366d4a03ac7957ab8c28890415e267f9b6589969e74b6e42225ec"},
     {file = "Brotli-1.1.0-cp310-cp310-win32.whl", hash = "sha256:be36e3d172dc816333f33520154d708a2657ea63762ec16b62ece02ab5e4daf2"},
     {file = "Brotli-1.1.0-cp310-cp310-win_amd64.whl", hash = "sha256:0c6244521dda65ea562d5a69b9a26120769b7a9fb3db2fe9545935ed6735b128"},
     {file = "Brotli-1.1.0-cp311-cp311-macosx_10_9_universal2.whl", hash = "sha256:a3daabb76a78f829cafc365531c972016e4aa8d5b4bf60660ad8ecee19df7ccc"},
@@ -351,8 +355,14 @@ files = [
     {file = "Brotli-1.1.0-cp311-cp311-musllinux_1_1_i686.whl", hash = "sha256:19c116e796420b0cee3da1ccec3b764ed2952ccfcc298b55a10e5610ad7885f9"},
     {file = "Brotli-1.1.0-cp311-cp311-musllinux_1_1_ppc64le.whl", hash = "sha256:510b5b1bfbe20e1a7b3baf5fed9e9451873559a976c1a78eebaa3b86c57b4265"},
     {file = "Brotli-1.1.0-cp311-cp311-musllinux_1_1_x86_64.whl", hash = "sha256:a1fd8a29719ccce974d523580987b7f8229aeace506952fa9ce1d53a033873c8"},
+    {file = "Brotli-1.1.0-cp311-cp311-musllinux_1_2_aarch64.whl", hash = "sha256:c247dd99d39e0338a604f8c2b3bc7061d5c2e9e2ac7ba9cc1be5a69cb6cd832f"},
+    {file = "Brotli-1.1.0-cp311-cp311-musllinux_1_2_i686.whl", hash = "sha256:1b2c248cd517c222d89e74669a4adfa5577e06ab68771a529060cf5a156e9757"},
+    {file = "Brotli-1.1.0-cp311-cp311-musllinux_1_2_ppc64le.whl", hash = "sha256:2a24c50840d89ded6c9a8fdc7b6ed3692ed4e86f1c4a4a938e1e92def92933e0"},
+    {file = "Brotli-1.1.0-cp311-cp311-musllinux_1_2_x86_64.whl", hash = "sha256:f31859074d57b4639318523d6ffdca586ace54271a73ad23ad021acd807eb14b"},
     {file = "Brotli-1.1.0-cp311-cp311-win32.whl", hash = "sha256:39da8adedf6942d76dc3e46653e52df937a3c4d6d18fdc94a7c29d263b1f5b50"},
     {file = "Brotli-1.1.0-cp311-cp311-win_amd64.whl", hash = "sha256:aac0411d20e345dc0920bdec5548e438e999ff68d77564d5e9463a7ca9d3e7b1"},
+    {file = "Brotli-1.1.0-cp312-cp312-macosx_10_13_universal2.whl", hash = "sha256:32d95b80260d79926f5fab3c41701dbb818fde1c9da590e77e571eefd14abe28"},
+    {file = "Brotli-1.1.0-cp312-cp312-macosx_10_13_x86_64.whl", hash = "sha256:b760c65308ff1e462f65d69c12e4ae085cff3b332d894637f6273a12a482d09f"},
     {file = "Brotli-1.1.0-cp312-cp312-macosx_10_9_universal2.whl", hash = "sha256:316cc9b17edf613ac76b1f1f305d2a748f1b976b033b049a6ecdfd5612c70409"},
     {file = "Brotli-1.1.0-cp312-cp312-macosx_10_9_x86_64.whl", hash = "sha256:caf9ee9a5775f3111642d33b86237b05808dafcd6268faa492250e9b78046eb2"},
     {file = "Brotli-1.1.0-cp312-cp312-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:70051525001750221daa10907c77830bc889cb6d865cc0b813d9db7fefc21451"},
@@ -363,8 +373,24 @@ files = [
     {file = "Brotli-1.1.0-cp312-cp312-musllinux_1_1_i686.whl", hash = "sha256:4093c631e96fdd49e0377a9c167bfd75b6d0bad2ace734c6eb20b348bc3ea180"},
     {file = "Brotli-1.1.0-cp312-cp312-musllinux_1_1_ppc64le.whl", hash = "sha256:7e4c4629ddad63006efa0ef968c8e4751c5868ff0b1c5c40f76524e894c50248"},
     {file = "Brotli-1.1.0-cp312-cp312-musllinux_1_1_x86_64.whl", hash = "sha256:861bf317735688269936f755fa136a99d1ed526883859f86e41a5d43c61d8966"},
+    {file = "Brotli-1.1.0-cp312-cp312-musllinux_1_2_aarch64.whl", hash = "sha256:87a3044c3a35055527ac75e419dfa9f4f3667a1e887ee80360589eb8c90aabb9"},
+    {file = "Brotli-1.1.0-cp312-cp312-musllinux_1_2_i686.whl", hash = "sha256:c5529b34c1c9d937168297f2c1fde7ebe9ebdd5e121297ff9c043bdb2ae3d6fb"},
+    {file = "Brotli-1.1.0-cp312-cp312-musllinux_1_2_ppc64le.whl", hash = "sha256:ca63e1890ede90b2e4454f9a65135a4d387a4585ff8282bb72964fab893f2111"},
+    {file = "Brotli-1.1.0-cp312-cp312-musllinux_1_2_x86_64.whl", hash = "sha256:e79e6520141d792237c70bcd7a3b122d00f2613769ae0cb61c52e89fd3443839"},
     {file = "Brotli-1.1.0-cp312-cp312-win32.whl", hash = "sha256:5f4d5ea15c9382135076d2fb28dde923352fe02951e66935a9efaac8f10e81b0"},
     {file = "Brotli-1.1.0-cp312-cp312-win_amd64.whl", hash = "sha256:906bc3a79de8c4ae5b86d3d75a8b77e44404b0f4261714306e3ad248d8ab0951"},
+    {file = "Brotli-1.1.0-cp313-cp313-macosx_10_13_universal2.whl", hash = "sha256:8bf32b98b75c13ec7cf774164172683d6e7891088f6316e54425fde1efc276d5"},
+    {file = "Brotli-1.1.0-cp313-cp313-macosx_10_13_x86_64.whl", hash = "sha256:7bc37c4d6b87fb1017ea28c9508b36bbcb0c3d18b4260fcdf08b200c74a6aee8"},
+    {file = "Brotli-1.1.0-cp313-cp313-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:3c0ef38c7a7014ffac184db9e04debe495d317cc9c6fb10071f7fefd93100a4f"},
+    {file = "Brotli-1.1.0-cp313-cp313-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:91d7cc2a76b5567591d12c01f019dd7afce6ba8cba6571187e21e2fc418ae648"},
+    {file = "Brotli-1.1.0-cp313-cp313-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:a93dde851926f4f2678e704fadeb39e16c35d8baebd5252c9fd94ce8ce68c4a0"},
+    {file = "Brotli-1.1.0-cp313-cp313-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:f0db75f47be8b8abc8d9e31bc7aad0547ca26f24a54e6fd10231d623f183d089"},
+    {file = "Brotli-1.1.0-cp313-cp313-musllinux_1_2_aarch64.whl", hash = "sha256:6967ced6730aed543b8673008b5a391c3b1076d834ca438bbd70635c73775368"},
+    {file = "Brotli-1.1.0-cp313-cp313-musllinux_1_2_i686.whl", hash = "sha256:7eedaa5d036d9336c95915035fb57422054014ebdeb6f3b42eac809928e40d0c"},
+    {file = "Brotli-1.1.0-cp313-cp313-musllinux_1_2_ppc64le.whl", hash = "sha256:d487f5432bf35b60ed625d7e1b448e2dc855422e87469e3f450aa5552b0eb284"},
+    {file = "Brotli-1.1.0-cp313-cp313-musllinux_1_2_x86_64.whl", hash = "sha256:832436e59afb93e1836081a20f324cb185836c617659b07b129141a8426973c7"},
+    {file = "Brotli-1.1.0-cp313-cp313-win32.whl", hash = "sha256:43395e90523f9c23a3d5bdf004733246fba087f2948f87ab28015f12359ca6a0"},
+    {file = "Brotli-1.1.0-cp313-cp313-win_amd64.whl", hash = "sha256:9011560a466d2eb3f5a6e4929cf4a09be405c64154e12df0dd72713f6500e32b"},
     {file = "Brotli-1.1.0-cp36-cp36m-macosx_10_9_x86_64.whl", hash = "sha256:a090ca607cbb6a34b0391776f0cb48062081f5f60ddcce5d11838e67a01928d1"},
     {file = "Brotli-1.1.0-cp36-cp36m-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:2de9d02f5bda03d27ede52e8cfe7b865b066fa49258cbab568720aa5be80a47d"},
     {file = "Brotli-1.1.0-cp36-cp36m-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:2333e30a5e00fe0fe55903c8832e08ee9c3b1382aacf4db26664a16528d51b4b"},
@@ -374,6 +400,10 @@ files = [
     {file = "Brotli-1.1.0-cp36-cp36m-musllinux_1_1_i686.whl", hash = "sha256:fd5f17ff8f14003595ab414e45fce13d073e0762394f957182e69035c9f3d7c2"},
     {file = "Brotli-1.1.0-cp36-cp36m-musllinux_1_1_ppc64le.whl", hash = "sha256:069a121ac97412d1fe506da790b3e69f52254b9df4eb665cd42460c837193354"},
     {file = "Brotli-1.1.0-cp36-cp36m-musllinux_1_1_x86_64.whl", hash = "sha256:e93dfc1a1165e385cc8239fab7c036fb2cd8093728cbd85097b284d7b99249a2"},
+    {file = "Brotli-1.1.0-cp36-cp36m-musllinux_1_2_aarch64.whl", hash = "sha256:aea440a510e14e818e67bfc4027880e2fb500c2ccb20ab21c7a7c8b5b4703d75"},
+    {file = "Brotli-1.1.0-cp36-cp36m-musllinux_1_2_i686.whl", hash = "sha256:6974f52a02321b36847cd19d1b8e381bf39939c21efd6ee2fc13a28b0d99348c"},
+    {file = "Brotli-1.1.0-cp36-cp36m-musllinux_1_2_ppc64le.whl", hash = "sha256:a7e53012d2853a07a4a79c00643832161a910674a893d296c9f1259859a289d2"},
+    {file = "Brotli-1.1.0-cp36-cp36m-musllinux_1_2_x86_64.whl", hash = "sha256:d7702622a8b40c49bffb46e1e3ba2e81268d5c04a34f460978c6b5517a34dd52"},
     {file = "Brotli-1.1.0-cp36-cp36m-win32.whl", hash = "sha256:a599669fd7c47233438a56936988a2478685e74854088ef5293802123b5b2460"},
     {file = "Brotli-1.1.0-cp36-cp36m-win_amd64.whl", hash = "sha256:d143fd47fad1db3d7c27a1b1d66162e855b5d50a89666af46e1679c496e8e579"},
     {file = "Brotli-1.1.0-cp37-cp37m-macosx_10_9_x86_64.whl", hash = "sha256:11d00ed0a83fa22d29bc6b64ef636c4552ebafcef57154b4ddd132f5638fbd1c"},
@@ -385,6 +415,10 @@ files = [
     {file = "Brotli-1.1.0-cp37-cp37m-musllinux_1_1_i686.whl", hash = "sha256:919e32f147ae93a09fe064d77d5ebf4e35502a8df75c29fb05788528e330fe74"},
     {file = "Brotli-1.1.0-cp37-cp37m-musllinux_1_1_ppc64le.whl", hash = "sha256:23032ae55523cc7bccb4f6a0bf368cd25ad9bcdcc1990b64a647e7bbcce9cb5b"},
     {file = "Brotli-1.1.0-cp37-cp37m-musllinux_1_1_x86_64.whl", hash = "sha256:224e57f6eac61cc449f498cc5f0e1725ba2071a3d4f48d5d9dffba42db196438"},
+    {file = "Brotli-1.1.0-cp37-cp37m-musllinux_1_2_aarch64.whl", hash = "sha256:cb1dac1770878ade83f2ccdf7d25e494f05c9165f5246b46a621cc849341dc01"},
+    {file = "Brotli-1.1.0-cp37-cp37m-musllinux_1_2_i686.whl", hash = "sha256:3ee8a80d67a4334482d9712b8e83ca6b1d9bc7e351931252ebef5d8f7335a547"},
+    {file = "Brotli-1.1.0-cp37-cp37m-musllinux_1_2_ppc64le.whl", hash = "sha256:5e55da2c8724191e5b557f8e18943b1b4839b8efc3ef60d65985bcf6f587dd38"},
+    {file = "Brotli-1.1.0-cp37-cp37m-musllinux_1_2_x86_64.whl", hash = "sha256:d342778ef319e1026af243ed0a07c97acf3bad33b9f29e7ae6a1f68fd083e90c"},
     {file = "Brotli-1.1.0-cp37-cp37m-win32.whl", hash = "sha256:587ca6d3cef6e4e868102672d3bd9dc9698c309ba56d41c2b9c85bbb903cdb95"},
     {file = "Brotli-1.1.0-cp37-cp37m-win_amd64.whl", hash = "sha256:2954c1c23f81c2eaf0b0717d9380bd348578a94161a65b3a2afc62c86467dd68"},
     {file = "Brotli-1.1.0-cp38-cp38-macosx_10_9_universal2.whl", hash = "sha256:efa8b278894b14d6da122a72fefcebc28445f2d3f880ac59d46c90f4c13be9a3"},
@@ -397,6 +431,10 @@ files = [
     {file = "Brotli-1.1.0-cp38-cp38-musllinux_1_1_i686.whl", hash = "sha256:1ab4fbee0b2d9098c74f3057b2bc055a8bd92ccf02f65944a241b4349229185a"},
     {file = "Brotli-1.1.0-cp38-cp38-musllinux_1_1_ppc64le.whl", hash = "sha256:141bd4d93984070e097521ed07e2575b46f817d08f9fa42b16b9b5f27b5ac088"},
     {file = "Brotli-1.1.0-cp38-cp38-musllinux_1_1_x86_64.whl", hash = "sha256:fce1473f3ccc4187f75b4690cfc922628aed4d3dd013d047f95a9b3919a86596"},
+    {file = "Brotli-1.1.0-cp38-cp38-musllinux_1_2_aarch64.whl", hash = "sha256:d2b35ca2c7f81d173d2fadc2f4f31e88cc5f7a39ae5b6db5513cf3383b0e0ec7"},
+    {file = "Brotli-1.1.0-cp38-cp38-musllinux_1_2_i686.whl", hash = "sha256:af6fa6817889314555aede9a919612b23739395ce767fe7fcbea9a80bf140fe5"},
+    {file = "Brotli-1.1.0-cp38-cp38-musllinux_1_2_ppc64le.whl", hash = "sha256:2feb1d960f760a575dbc5ab3b1c00504b24caaf6986e2dc2b01c09c87866a943"},
+    {file = "Brotli-1.1.0-cp38-cp38-musllinux_1_2_x86_64.whl", hash = "sha256:4410f84b33374409552ac9b6903507cdb31cd30d2501fc5ca13d18f73548444a"},
     {file = "Brotli-1.1.0-cp38-cp38-win32.whl", hash = "sha256:db85ecf4e609a48f4b29055f1e144231b90edc90af7481aa731ba2d059226b1b"},
     {file = "Brotli-1.1.0-cp38-cp38-win_amd64.whl", hash = "sha256:3d7954194c36e304e1523f55d7042c59dc53ec20dd4e9ea9d151f1b62b4415c0"},
     {file = "Brotli-1.1.0-cp39-cp39-macosx_10_9_universal2.whl", hash = "sha256:5fb2ce4b8045c78ebbc7b8f3c15062e435d47e7393cc57c25115cfd49883747a"},
@@ -409,6 +447,10 @@ files = [
     {file = "Brotli-1.1.0-cp39-cp39-musllinux_1_1_i686.whl", hash = "sha256:949f3b7c29912693cee0afcf09acd6ebc04c57af949d9bf77d6101ebb61e388c"},
     {file = "Brotli-1.1.0-cp39-cp39-musllinux_1_1_ppc64le.whl", hash = "sha256:89f4988c7203739d48c6f806f1e87a1d96e0806d44f0fba61dba81392c9e474d"},
     {file = "Brotli-1.1.0-cp39-cp39-musllinux_1_1_x86_64.whl", hash = "sha256:de6551e370ef19f8de1807d0a9aa2cdfdce2e85ce88b122fe9f6b2b076837e59"},
+    {file = "Brotli-1.1.0-cp39-cp39-musllinux_1_2_aarch64.whl", hash = "sha256:0737ddb3068957cf1b054899b0883830bb1fec522ec76b1098f9b6e0f02d9419"},
+    {file = "Brotli-1.1.0-cp39-cp39-musllinux_1_2_i686.whl", hash = "sha256:4f3607b129417e111e30637af1b56f24f7a49e64763253bbc275c75fa887d4b2"},
+    {file = "Brotli-1.1.0-cp39-cp39-musllinux_1_2_ppc64le.whl", hash = "sha256:6c6e0c425f22c1c719c42670d561ad682f7bfeeef918edea971a79ac5252437f"},
+    {file = "Brotli-1.1.0-cp39-cp39-musllinux_1_2_x86_64.whl", hash = "sha256:494994f807ba0b92092a163a0a283961369a65f6cbe01e8891132b7a320e61eb"},
     {file = "Brotli-1.1.0-cp39-cp39-win32.whl", hash = "sha256:f0d8a7a6b5983c2496e364b969f0e526647a06b075d034f3297dc66f3b360c64"},
     {file = "Brotli-1.1.0-cp39-cp39-win_amd64.whl", hash = "sha256:cdad5b9014d83ca68c25d2e9444e28e967ef16e80f6b436918c700c117a85467"},
     {file = "Brotli-1.1.0.tar.gz", hash = "sha256:81de08ac11bcb85841e440c13611c00b67d3bf82698314928d0b676362546724"},
@@ -1681,11 +1723,14 @@ orjson = "*"
 pandas = "*"
 protobuf = "*"
 py-grpc-prometheus = "*"
+pyarrow = ">=14.0.1"
 pydantic = "2.7.1"
 pydantic-settings = "2.2.1"
 python-dotenv = "*"
 python-multipart = "*"
 starlette-exporter = "*"
+tensorflow = ">2.12"
+transformers = ">=4.36.0"
 tritonclient = {version = ">=2.42", extras = ["http"]}
 uvicorn = "*"
 uvloop = {version = "*", markers = "(sys_platform != \"win32\" and sys_platform != \"cygwin\") and platform_python_implementation != \"PyPy\""}
@@ -3123,7 +3168,6 @@ files = [
     {file = "PyYAML-6.0.1-cp310-cp310-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:69b023b2b4daa7548bcfbd4aa3da05b3a74b772db9e23b982788168117739938"},
     {file = "PyYAML-6.0.1-cp310-cp310-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:81e0b275a9ecc9c0c0c07b4b90ba548307583c125f54d5b6946cfee6360c733d"},
     {file = "PyYAML-6.0.1-cp310-cp310-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:ba336e390cd8e4d1739f42dfe9bb83a3cc2e80f567d8805e11b46f4a943f5515"},
-    {file = "PyYAML-6.0.1-cp310-cp310-musllinux_1_1_x86_64.whl", hash = "sha256:326c013efe8048858a6d312ddd31d56e468118ad4cdeda36c719bf5bb6192290"},
     {file = "PyYAML-6.0.1-cp310-cp310-win32.whl", hash = "sha256:bd4af7373a854424dabd882decdc5579653d7868b8fb26dc7d0e99f823aa5924"},
     {file = "PyYAML-6.0.1-cp310-cp310-win_amd64.whl", hash = "sha256:fd1592b3fdf65fff2ad0004b5e363300ef59ced41c2e6b3a99d4089fa8c5435d"},
     {file = "PyYAML-6.0.1-cp311-cp311-macosx_10_9_x86_64.whl", hash = "sha256:6965a7bc3cf88e5a1c3bd2e0b5c22f8d677dc88a455344035f03399034eb3007"},
@@ -3131,16 +3175,8 @@ files = [
     {file = "PyYAML-6.0.1-cp311-cp311-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:42f8152b8dbc4fe7d96729ec2b99c7097d656dc1213a3229ca5383f973a5ed6d"},
     {file = "PyYAML-6.0.1-cp311-cp311-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:062582fca9fabdd2c8b54a3ef1c978d786e0f6b3a1510e0ac93ef59e0ddae2bc"},
     {file = "PyYAML-6.0.1-cp311-cp311-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:d2b04aac4d386b172d5b9692e2d2da8de7bfb6c387fa4f801fbf6fb2e6ba4673"},
-    {file = "PyYAML-6.0.1-cp311-cp311-musllinux_1_1_x86_64.whl", hash = "sha256:e7d73685e87afe9f3b36c799222440d6cf362062f78be1013661b00c5c6f678b"},
     {file = "PyYAML-6.0.1-cp311-cp311-win32.whl", hash = "sha256:1635fd110e8d85d55237ab316b5b011de701ea0f29d07611174a1b42f1444741"},
     {file = "PyYAML-6.0.1-cp311-cp311-win_amd64.whl", hash = "sha256:bf07ee2fef7014951eeb99f56f39c9bb4af143d8aa3c21b1677805985307da34"},
-    {file = "PyYAML-6.0.1-cp312-cp312-macosx_10_9_x86_64.whl", hash = "sha256:855fb52b0dc35af121542a76b9a84f8d1cd886ea97c84703eaa6d88e37a2ad28"},
-    {file = "PyYAML-6.0.1-cp312-cp312-macosx_11_0_arm64.whl", hash = "sha256:40df9b996c2b73138957fe23a16a4f0ba614f4c0efce1e9406a184b6d07fa3a9"},
-    {file = "PyYAML-6.0.1-cp312-cp312-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:a08c6f0fe150303c1c6b71ebcd7213c2858041a7e01975da3a99aed1e7a378ef"},
-    {file = "PyYAML-6.0.1-cp312-cp312-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:6c22bec3fbe2524cde73d7ada88f6566758a8f7227bfbf93a408a9d86bcc12a0"},
-    {file = "PyYAML-6.0.1-cp312-cp312-musllinux_1_1_x86_64.whl", hash = "sha256:8d4e9c88387b0f5c7d5f281e55304de64cf7f9c0021a3525bd3b1c542da3b0e4"},
-    {file = "PyYAML-6.0.1-cp312-cp312-win32.whl", hash = "sha256:d483d2cdf104e7c9fa60c544d92981f12ad66a457afae824d146093b8c294c54"},
-    {file = "PyYAML-6.0.1-cp312-cp312-win_amd64.whl", hash = "sha256:0d3304d8c0adc42be59c5f8a4d9e3d7379e6955ad754aa9d6ab7a398b59dd1df"},
     {file = "PyYAML-6.0.1-cp36-cp36m-macosx_10_9_x86_64.whl", hash = "sha256:50550eb667afee136e9a77d6dc71ae76a44df8b3e51e41b77f6de2932bfe0f47"},
     {file = "PyYAML-6.0.1-cp36-cp36m-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:1fe35611261b29bd1de0070f0b2f47cb6ff71fa6595c077e42bd0c419fa27b98"},
     {file = "PyYAML-6.0.1-cp36-cp36m-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:704219a11b772aea0d8ecd7058d0082713c3562b4e271b849ad7dc4a5c90c13c"},
@@ -3157,7 +3193,6 @@ files = [
     {file = "PyYAML-6.0.1-cp38-cp38-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:a0cd17c15d3bb3fa06978b4e8958dcdc6e0174ccea823003a106c7d4d7899ac5"},
     {file = "PyYAML-6.0.1-cp38-cp38-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:28c119d996beec18c05208a8bd78cbe4007878c6dd15091efb73a30e90539696"},
     {file = "PyYAML-6.0.1-cp38-cp38-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:7e07cbde391ba96ab58e532ff4803f79c4129397514e1413a7dc761ccd755735"},
-    {file = "PyYAML-6.0.1-cp38-cp38-musllinux_1_1_x86_64.whl", hash = "sha256:49a183be227561de579b4a36efbb21b3eab9651dd81b1858589f796549873dd6"},
     {file = "PyYAML-6.0.1-cp38-cp38-win32.whl", hash = "sha256:184c5108a2aca3c5b3d3bf9395d50893a7ab82a38004c8f61c258d4428e80206"},
     {file = "PyYAML-6.0.1-cp38-cp38-win_amd64.whl", hash = "sha256:1e2722cc9fbb45d9b87631ac70924c11d3a401b2d7f410cc0e3bbf249f2dca62"},
     {file = "PyYAML-6.0.1-cp39-cp39-macosx_10_9_x86_64.whl", hash = "sha256:9eb6caa9a297fc2c2fb8862bc5370d0303ddba53ba97e71f08023b6cd73d16a8"},
@@ -3165,7 +3200,6 @@ files = [
     {file = "PyYAML-6.0.1-cp39-cp39-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:5773183b6446b2c99bb77e77595dd486303b4faab2b086e7b17bc6bef28865f6"},
     {file = "PyYAML-6.0.1-cp39-cp39-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:b786eecbdf8499b9ca1d697215862083bd6d2a99965554781d0d8d1ad31e13a0"},
     {file = "PyYAML-6.0.1-cp39-cp39-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:bc1bf2925a1ecd43da378f4db9e4f799775d6367bdb94671027b73b393a7c42c"},
-    {file = "PyYAML-6.0.1-cp39-cp39-musllinux_1_1_x86_64.whl", hash = "sha256:04ac92ad1925b2cff1db0cfebffb6ffc43457495c9b3c39d3fcae417d7125dc5"},
     {file = "PyYAML-6.0.1-cp39-cp39-win32.whl", hash = "sha256:faca3bdcf85b2fc05d06ff3fbc1f83e1391b3e724afa3feba7d13eeab355484c"},
     {file = "PyYAML-6.0.1-cp39-cp39-win_amd64.whl", hash = "sha256:510c9deebc5c0225e8c96813043e62b680ba2f9c50a08d3724c7f28a747d1486"},
     {file = "PyYAML-6.0.1.tar.gz", hash = "sha256:bfdf460b1736c775f2ba9f6a92bca30bc2095067b8a9d77876d1fad6cc3b4a43"},
@@ -3831,6 +3865,24 @@ files = [
     {file = "threadpoolctl-3.3.0.tar.gz", hash = "sha256:5dac632b4fa2d43f42130267929af3ba01399ef4bd1882918e92dbc30365d30c"},
 ]
 
+[[package]]
+name = "timm"
+version = "1.0.14"
+description = "PyTorch Image Models"
+optional = false
+python-versions = ">=3.8"
+files = [
+    {file = "timm-1.0.14-py3-none-any.whl", hash = "sha256:16653695ff420cf4e87e384c8654f2ee6be2070bcb4a0e840b5c6764ee0547ec"},
+    {file = "timm-1.0.14.tar.gz", hash = "sha256:00a7f2cc04ce3ed8f80476bbb7eea27eac8cf6f2d59b5e9aa9cdd375dd6550db"},
+]
+
+[package.dependencies]
+huggingface_hub = "*"
+pyyaml = "*"
+safetensors = "*"
+torch = "*"
+torchvision = "*"
+
 [[package]]
 name = "tokenizers"
 version = "0.19.1"
@@ -4060,6 +4112,85 @@ typing-extensions = ">=4.8.0"
 opt-einsum = ["opt-einsum (>=3.3)"]
 optree = ["optree (>=0.9.1)"]
 
+[[package]]
+name = "torchvision"
+version = "0.17.2"
+description = "image and video datasets and models for torch deep learning"
+optional = false
+python-versions = ">=3.8"
+files = [
+    {file = "torchvision-0.17.2-cp310-cp310-macosx_10_13_x86_64.whl", hash = "sha256:1f2910fe3c21ad6875b2720d46fad835b2e4b336e9553d31ca364d24c90b1d4f"},
+    {file = "torchvision-0.17.2-cp310-cp310-macosx_11_0_arm64.whl", hash = "sha256:ecc1c503fa8a54fbab777e06a7c228032b8ab78efebf35b28bc8f22f544f51f1"},
+    {file = "torchvision-0.17.2-cp310-cp310-manylinux1_x86_64.whl", hash = "sha256:f400145fc108833e7c2fc28486a04989ca742146d7a2a2cc48878ebbb40cdbbd"},
+    {file = "torchvision-0.17.2-cp310-cp310-manylinux2014_aarch64.whl", hash = "sha256:e9e4bed404af33dfc92eecc2b513d21ddc4c242a7fd8708b3b09d3a26aa6f444"},
+    {file = "torchvision-0.17.2-cp310-cp310-win_amd64.whl", hash = "sha256:ba2e62f233eab3d42b648c122a3a29c47cc108ca314dfd5cbb59cd3a143fd623"},
+    {file = "torchvision-0.17.2-cp311-cp311-macosx_10_13_x86_64.whl", hash = "sha256:9b83e55ee7d0a1704f52b9c0ac87388e7a6d1d98a6bde7b0b35f9ab54d7bda54"},
+    {file = "torchvision-0.17.2-cp311-cp311-macosx_11_0_arm64.whl", hash = "sha256:e031004a1bc432c980a7bd642f6c189a3efc316e423fc30b5569837166a4e28d"},
+    {file = "torchvision-0.17.2-cp311-cp311-manylinux1_x86_64.whl", hash = "sha256:3bbc24b7713e8f22766992562547d8b4b10001208d372fe599255af84bfd1a69"},
+    {file = "torchvision-0.17.2-cp311-cp311-manylinux2014_aarch64.whl", hash = "sha256:833fd2e4216ced924c8aca0525733fe727f9a1af66dfad7c5be7257e97c39678"},
+    {file = "torchvision-0.17.2-cp311-cp311-win_amd64.whl", hash = "sha256:6835897df852fad1015e6a106c167c83848114cbcc7d86112384a973404e4431"},
+    {file = "torchvision-0.17.2-cp312-cp312-macosx_10_13_x86_64.whl", hash = "sha256:14fd1d4a033c325bdba2d03a69c3450cab6d3a625f85cc375781d9237ca5d04d"},
+    {file = "torchvision-0.17.2-cp312-cp312-macosx_11_0_arm64.whl", hash = "sha256:9c3acbebbe379af112b62b535820174277b1f3eed30df264a4e458d58ee4e5b2"},
+    {file = "torchvision-0.17.2-cp312-cp312-manylinux1_x86_64.whl", hash = "sha256:77d680adf6ce367166a186d2c7fda3a73807ab9a03b2c31a03fa8812c8c5335b"},
+    {file = "torchvision-0.17.2-cp312-cp312-manylinux2014_aarch64.whl", hash = "sha256:f1c9ab3152cfb27f83aca072cac93a3a4c4e4ab0261cf0f2d516b9868a4e96f3"},
+    {file = "torchvision-0.17.2-cp312-cp312-win_amd64.whl", hash = "sha256:3f784381419f3ed3f2ec2aa42fb4aeec5bf4135e298d1631e41c926e6f1a0dff"},
+    {file = "torchvision-0.17.2-cp38-cp38-macosx_10_13_x86_64.whl", hash = "sha256:b83aac8d78f48981146d582168d75b6c947cfb0a7693f76e219f1926f6e595a3"},
+    {file = "torchvision-0.17.2-cp38-cp38-macosx_11_0_arm64.whl", hash = "sha256:1ece40557e122d79975860a005aa7e2a9e2e6c350a03e78a00ec1450083312fd"},
+    {file = "torchvision-0.17.2-cp38-cp38-manylinux1_x86_64.whl", hash = "sha256:32dbeba3987e20f2dc1bce8d1504139fff582898346dfe8ad98d649f97ca78fa"},
+    {file = "torchvision-0.17.2-cp38-cp38-manylinux2014_aarch64.whl", hash = "sha256:35ba5c1600c3203549d2316422a659bd20c0cfda1b6085eec94fb9f35f55ca43"},
+    {file = "torchvision-0.17.2-cp38-cp38-win_amd64.whl", hash = "sha256:2f69570f50b1d195e51bc03feffb7b7728207bc36efcfb1f0813712b2379d881"},
+    {file = "torchvision-0.17.2-cp39-cp39-macosx_10_13_x86_64.whl", hash = "sha256:4868bbfa55758c8107e69a0e7dd5e77b89056035cd38b767ad5b98cdb71c0f0d"},
+    {file = "torchvision-0.17.2-cp39-cp39-macosx_11_0_arm64.whl", hash = "sha256:efd6d0dd0668e15d01a2cffadc74068433b32cbcf5692e0c4aa15fc5cb250ce7"},
+    {file = "torchvision-0.17.2-cp39-cp39-manylinux1_x86_64.whl", hash = "sha256:7dc85b397f6c6d9ef12716ce0d6e11ac2b803f5cccff6fe3966db248e7774478"},
+    {file = "torchvision-0.17.2-cp39-cp39-manylinux2014_aarch64.whl", hash = "sha256:d506854c5acd69b20a8b6641f01fe841685a21c5406b56813184f1c9fc94279e"},
+    {file = "torchvision-0.17.2-cp39-cp39-win_amd64.whl", hash = "sha256:067095e87a020a7a251ac1d38483aa591c5ccb81e815527c54db88a982fc9267"},
+]
+
+[package.dependencies]
+numpy = "*"
+pillow = ">=5.3.0,<8.3.dev0 || >=8.4.dev0"
+torch = "2.2.2"
+
+[package.extras]
+scipy = ["scipy"]
+
+[[package]]
+name = "torchvision"
+version = "0.18.1"
+description = "image and video datasets and models for torch deep learning"
+optional = false
+python-versions = ">=3.8"
+files = [
+    {file = "torchvision-0.18.1-cp310-cp310-macosx_11_0_arm64.whl", hash = "sha256:3e694e54b0548dad99c12af6bf0c8e4f3350137d391dcd19af22a1c5f89322b3"},
+    {file = "torchvision-0.18.1-cp310-cp310-manylinux1_x86_64.whl", hash = "sha256:0b3bda0aa5b416eeb547143b8eeaf17720bdba9cf516dc991aacb81811aa96a5"},
+    {file = "torchvision-0.18.1-cp310-cp310-manylinux2014_aarch64.whl", hash = "sha256:573ff523c739405edb085f65cb592f482d28a30e29b0be4c4ba08040b3ae785f"},
+    {file = "torchvision-0.18.1-cp310-cp310-win_amd64.whl", hash = "sha256:ef7bbbc60b38e831a75e547c66ca1784f2ac27100f9e4ddbe9614cef6cbcd942"},
+    {file = "torchvision-0.18.1-cp311-cp311-macosx_11_0_arm64.whl", hash = "sha256:80b5d794dd0fdba787adc22f1a367a5ead452327686473cb260dd94364bc56a6"},
+    {file = "torchvision-0.18.1-cp311-cp311-manylinux1_x86_64.whl", hash = "sha256:9077cf590cdb3a5e8fdf5cdb71797f8c67713f974cf0228ecb17fcd670ab42f9"},
+    {file = "torchvision-0.18.1-cp311-cp311-manylinux2014_aarch64.whl", hash = "sha256:ceb993a882f1ae7ae373ed39c28d7e3e802205b0e59a7ed84ef4028f0bba8d7f"},
+    {file = "torchvision-0.18.1-cp311-cp311-win_amd64.whl", hash = "sha256:52f7436140045dc2239cdc502aa76b2bd8bd676d64244ff154d304aa69852046"},
+    {file = "torchvision-0.18.1-cp312-cp312-macosx_11_0_arm64.whl", hash = "sha256:2be6f0bf7c455c89a51a1dbb6f668d36c6edc479f49ac912d745d10df5715657"},
+    {file = "torchvision-0.18.1-cp312-cp312-manylinux1_x86_64.whl", hash = "sha256:f118d887bfde3a948a41d56587525401e5cac1b7db2eaca203324d6ed2b1caca"},
+    {file = "torchvision-0.18.1-cp312-cp312-manylinux2014_aarch64.whl", hash = "sha256:13d24d904f65e62d66a1e0c41faec630bc193867b8a4a01166769e8a8e8df8e9"},
+    {file = "torchvision-0.18.1-cp312-cp312-win_amd64.whl", hash = "sha256:ed6340b69a63a625e512a66127210d412551d9c5f2ad2978130c6a45bf56cd4a"},
+    {file = "torchvision-0.18.1-cp38-cp38-macosx_11_0_arm64.whl", hash = "sha256:b1c3864fa9378c88bce8ad0ef3599f4f25397897ce612e1c245c74b97092f35e"},
+    {file = "torchvision-0.18.1-cp38-cp38-manylinux1_x86_64.whl", hash = "sha256:02085a2ffc7461f5c0edb07d6f3455ee1806561f37736b903da820067eea58c7"},
+    {file = "torchvision-0.18.1-cp38-cp38-manylinux2014_aarch64.whl", hash = "sha256:9726c316a2501df8503e5a5dc46a631afd4c515a958972e5b7f7b9c87d2125c0"},
+    {file = "torchvision-0.18.1-cp38-cp38-win_amd64.whl", hash = "sha256:64a2662dbf30db9055d8b201d6e56f312a504e5ccd9d144c57c41622d3c524cb"},
+    {file = "torchvision-0.18.1-cp39-cp39-macosx_11_0_arm64.whl", hash = "sha256:975b8594c0f5288875408acbb74946eea786c5b008d129c0d045d0ead23742bc"},
+    {file = "torchvision-0.18.1-cp39-cp39-manylinux1_x86_64.whl", hash = "sha256:da83c8bbd34d8bee48bfa1d1b40e0844bc3cba10ed825a5a8cbe3ce7b62264cd"},
+    {file = "torchvision-0.18.1-cp39-cp39-manylinux2014_aarch64.whl", hash = "sha256:54bfcd352abb396d5c9c237d200167c178bd136051b138e1e8ef46ce367c2773"},
+    {file = "torchvision-0.18.1-cp39-cp39-win_amd64.whl", hash = "sha256:5c8366a1aeee49e9ea9e64b30d199debdf06b1bd7610a76165eb5d7869c3bde5"},
+]
+
+[package.dependencies]
+numpy = "*"
+pillow = ">=5.3.0,<8.3.dev0 || >=8.4.dev0"
+torch = "2.3.1"
+
+[package.extras]
+scipy = ["scipy"]
+
 [[package]]
 name = "tqdm"
 version = "4.66.4"
@@ -4740,4 +4871,4 @@ testing = ["coverage (>=5.0.3)", "zope.event", "zope.testing"]
 [metadata]
 lock-version = "2.0"
 python-versions = ">=3.9,<3.12"
-content-hash = "24af455593b387d8ac5e9ee82e55de1dbbd0cb96230453bd17deec76b014439c"
+content-hash = "9861bc587b5154cd19bcaa5fe0c3bafe04e08e2240d6f9d91d3cd13e9e78c466"
diff --git a/runtimes/huggingface/pyproject.toml b/runtimes/huggingface/pyproject.toml
index 44ec8f21e..c3765c6d3 100644
--- a/runtimes/huggingface/pyproject.toml
+++ b/runtimes/huggingface/pyproject.toml
@@ -18,6 +18,7 @@ accelerate = "^0.27.2"
 bitsandbytes = "^0.42.0"
 sentence-transformers = "2.5.1"
 transformers = "4.41.2"
+timm ="^1.0.13"
 
 [tool.poetry.group.dev.dependencies]
 mlserver = {path = "../..", develop = true}

From 09dc4952d5841610641ffb407ff2ccced43518a0 Mon Sep 17 00:00:00 2001
From: Nanbo Liu <nliu@lambda001.america.striveworks.us>
Date: Fri, 31 Jan 2025 15:35:59 -0600
Subject: [PATCH 15/21] boost version

---
 poetry.lock                                   | 205 +++++++++++++++---
 .../mlserver_huggingface/version.py           |   2 +-
 runtimes/huggingface/pyproject.toml           |   2 +-
 3 files changed, 178 insertions(+), 31 deletions(-)

diff --git a/poetry.lock b/poetry.lock
index 9f8570f12..c18524e9d 100644
--- a/poetry.lock
+++ b/poetry.lock
@@ -11,6 +11,36 @@ files = [
     {file = "absl_py-2.1.0-py3-none-any.whl", hash = "sha256:526a04eadab8b4ee719ce68f204172ead1027549089702d99b9059f129ff1308"},
 ]
 
+[[package]]
+name = "accelerate"
+version = "0.27.2"
+description = "Accelerate"
+optional = false
+python-versions = ">=3.8.0"
+files = [
+    {file = "accelerate-0.27.2-py3-none-any.whl", hash = "sha256:a818dd27b9ba24e9eb5030d1b285cf4cdd1b41bbfa675fb4eb2477ddfc097074"},
+    {file = "accelerate-0.27.2.tar.gz", hash = "sha256:cc715fe9a8bc7a286259bfb6d65fb78363badd3371e7cbda4e4a4ef34a0010aa"},
+]
+
+[package.dependencies]
+huggingface-hub = "*"
+numpy = ">=1.17"
+packaging = ">=20.0"
+psutil = "*"
+pyyaml = "*"
+safetensors = ">=0.3.1"
+torch = ">=1.10.0"
+
+[package.extras]
+dev = ["bitsandbytes", "black (>=23.1,<24.0)", "datasets", "deepspeed (<0.13.0)", "evaluate", "hf-doc-builder (>=0.3.0)", "parameterized", "pytest", "pytest-subtests", "pytest-xdist", "rich", "ruff (>=0.1.15,<0.2.0)", "scikit-learn", "scipy", "timm", "torchpippy (>=0.2.0)", "tqdm", "transformers"]
+quality = ["black (>=23.1,<24.0)", "hf-doc-builder (>=0.3.0)", "ruff (>=0.1.15,<0.2.0)"]
+rich = ["rich"]
+sagemaker = ["sagemaker"]
+test-dev = ["bitsandbytes", "datasets", "deepspeed (<0.13.0)", "evaluate", "scikit-learn", "scipy", "timm", "torchpippy (>=0.2.0)", "tqdm", "transformers"]
+test-prod = ["parameterized", "pytest", "pytest-subtests", "pytest-xdist"]
+test-trackers = ["comet-ml", "dvclive", "tensorboard", "wandb"]
+testing = ["bitsandbytes", "datasets", "deepspeed (<0.13.0)", "evaluate", "parameterized", "pytest", "pytest-subtests", "pytest-xdist", "scikit-learn", "scipy", "timm", "torchpippy (>=0.2.0)", "tqdm", "transformers"]
+
 [[package]]
 name = "aiofiles"
 version = "23.2.1"
@@ -485,6 +515,20 @@ charset-normalizer = ["charset-normalizer"]
 html5lib = ["html5lib"]
 lxml = ["lxml"]
 
+[[package]]
+name = "bitsandbytes"
+version = "0.42.0"
+description = "k-bit optimizers and matrix multiplication routines."
+optional = false
+python-versions = "*"
+files = [
+    {file = "bitsandbytes-0.42.0-py3-none-any.whl", hash = "sha256:63798680912cc63bb77b535a2d0860af024e290a52e157f777ad2a52e2585967"},
+    {file = "bitsandbytes-0.42.0.tar.gz", hash = "sha256:fc1505f184f0d275766f2a6c663f1a43b734c1409b5c5a406f3a6073d9f329fd"},
+]
+
+[package.dependencies]
+scipy = "*"
+
 [[package]]
 name = "black"
 version = "24.4.0"
@@ -605,6 +649,10 @@ files = [
     {file = "Brotli-1.1.0-cp310-cp310-musllinux_1_1_i686.whl", hash = "sha256:a37b8f0391212d29b3a91a799c8e4a2855e0576911cdfb2515487e30e322253d"},
     {file = "Brotli-1.1.0-cp310-cp310-musllinux_1_1_ppc64le.whl", hash = "sha256:e84799f09591700a4154154cab9787452925578841a94321d5ee8fb9a9a328f0"},
     {file = "Brotli-1.1.0-cp310-cp310-musllinux_1_1_x86_64.whl", hash = "sha256:f66b5337fa213f1da0d9000bc8dc0cb5b896b726eefd9c6046f699b169c41b9e"},
+    {file = "Brotli-1.1.0-cp310-cp310-musllinux_1_2_aarch64.whl", hash = "sha256:5dab0844f2cf82be357a0eb11a9087f70c5430b2c241493fc122bb6f2bb0917c"},
+    {file = "Brotli-1.1.0-cp310-cp310-musllinux_1_2_i686.whl", hash = "sha256:e4fe605b917c70283db7dfe5ada75e04561479075761a0b3866c081d035b01c1"},
+    {file = "Brotli-1.1.0-cp310-cp310-musllinux_1_2_ppc64le.whl", hash = "sha256:1e9a65b5736232e7a7f91ff3d02277f11d339bf34099a56cdab6a8b3410a02b2"},
+    {file = "Brotli-1.1.0-cp310-cp310-musllinux_1_2_x86_64.whl", hash = "sha256:58d4b711689366d4a03ac7957ab8c28890415e267f9b6589969e74b6e42225ec"},
     {file = "Brotli-1.1.0-cp310-cp310-win32.whl", hash = "sha256:be36e3d172dc816333f33520154d708a2657ea63762ec16b62ece02ab5e4daf2"},
     {file = "Brotli-1.1.0-cp310-cp310-win_amd64.whl", hash = "sha256:0c6244521dda65ea562d5a69b9a26120769b7a9fb3db2fe9545935ed6735b128"},
     {file = "Brotli-1.1.0-cp311-cp311-macosx_10_9_universal2.whl", hash = "sha256:a3daabb76a78f829cafc365531c972016e4aa8d5b4bf60660ad8ecee19df7ccc"},
@@ -617,8 +665,14 @@ files = [
     {file = "Brotli-1.1.0-cp311-cp311-musllinux_1_1_i686.whl", hash = "sha256:19c116e796420b0cee3da1ccec3b764ed2952ccfcc298b55a10e5610ad7885f9"},
     {file = "Brotli-1.1.0-cp311-cp311-musllinux_1_1_ppc64le.whl", hash = "sha256:510b5b1bfbe20e1a7b3baf5fed9e9451873559a976c1a78eebaa3b86c57b4265"},
     {file = "Brotli-1.1.0-cp311-cp311-musllinux_1_1_x86_64.whl", hash = "sha256:a1fd8a29719ccce974d523580987b7f8229aeace506952fa9ce1d53a033873c8"},
+    {file = "Brotli-1.1.0-cp311-cp311-musllinux_1_2_aarch64.whl", hash = "sha256:c247dd99d39e0338a604f8c2b3bc7061d5c2e9e2ac7ba9cc1be5a69cb6cd832f"},
+    {file = "Brotli-1.1.0-cp311-cp311-musllinux_1_2_i686.whl", hash = "sha256:1b2c248cd517c222d89e74669a4adfa5577e06ab68771a529060cf5a156e9757"},
+    {file = "Brotli-1.1.0-cp311-cp311-musllinux_1_2_ppc64le.whl", hash = "sha256:2a24c50840d89ded6c9a8fdc7b6ed3692ed4e86f1c4a4a938e1e92def92933e0"},
+    {file = "Brotli-1.1.0-cp311-cp311-musllinux_1_2_x86_64.whl", hash = "sha256:f31859074d57b4639318523d6ffdca586ace54271a73ad23ad021acd807eb14b"},
     {file = "Brotli-1.1.0-cp311-cp311-win32.whl", hash = "sha256:39da8adedf6942d76dc3e46653e52df937a3c4d6d18fdc94a7c29d263b1f5b50"},
     {file = "Brotli-1.1.0-cp311-cp311-win_amd64.whl", hash = "sha256:aac0411d20e345dc0920bdec5548e438e999ff68d77564d5e9463a7ca9d3e7b1"},
+    {file = "Brotli-1.1.0-cp312-cp312-macosx_10_13_universal2.whl", hash = "sha256:32d95b80260d79926f5fab3c41701dbb818fde1c9da590e77e571eefd14abe28"},
+    {file = "Brotli-1.1.0-cp312-cp312-macosx_10_13_x86_64.whl", hash = "sha256:b760c65308ff1e462f65d69c12e4ae085cff3b332d894637f6273a12a482d09f"},
     {file = "Brotli-1.1.0-cp312-cp312-macosx_10_9_universal2.whl", hash = "sha256:316cc9b17edf613ac76b1f1f305d2a748f1b976b033b049a6ecdfd5612c70409"},
     {file = "Brotli-1.1.0-cp312-cp312-macosx_10_9_x86_64.whl", hash = "sha256:caf9ee9a5775f3111642d33b86237b05808dafcd6268faa492250e9b78046eb2"},
     {file = "Brotli-1.1.0-cp312-cp312-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:70051525001750221daa10907c77830bc889cb6d865cc0b813d9db7fefc21451"},
@@ -629,8 +683,24 @@ files = [
     {file = "Brotli-1.1.0-cp312-cp312-musllinux_1_1_i686.whl", hash = "sha256:4093c631e96fdd49e0377a9c167bfd75b6d0bad2ace734c6eb20b348bc3ea180"},
     {file = "Brotli-1.1.0-cp312-cp312-musllinux_1_1_ppc64le.whl", hash = "sha256:7e4c4629ddad63006efa0ef968c8e4751c5868ff0b1c5c40f76524e894c50248"},
     {file = "Brotli-1.1.0-cp312-cp312-musllinux_1_1_x86_64.whl", hash = "sha256:861bf317735688269936f755fa136a99d1ed526883859f86e41a5d43c61d8966"},
+    {file = "Brotli-1.1.0-cp312-cp312-musllinux_1_2_aarch64.whl", hash = "sha256:87a3044c3a35055527ac75e419dfa9f4f3667a1e887ee80360589eb8c90aabb9"},
+    {file = "Brotli-1.1.0-cp312-cp312-musllinux_1_2_i686.whl", hash = "sha256:c5529b34c1c9d937168297f2c1fde7ebe9ebdd5e121297ff9c043bdb2ae3d6fb"},
+    {file = "Brotli-1.1.0-cp312-cp312-musllinux_1_2_ppc64le.whl", hash = "sha256:ca63e1890ede90b2e4454f9a65135a4d387a4585ff8282bb72964fab893f2111"},
+    {file = "Brotli-1.1.0-cp312-cp312-musllinux_1_2_x86_64.whl", hash = "sha256:e79e6520141d792237c70bcd7a3b122d00f2613769ae0cb61c52e89fd3443839"},
     {file = "Brotli-1.1.0-cp312-cp312-win32.whl", hash = "sha256:5f4d5ea15c9382135076d2fb28dde923352fe02951e66935a9efaac8f10e81b0"},
     {file = "Brotli-1.1.0-cp312-cp312-win_amd64.whl", hash = "sha256:906bc3a79de8c4ae5b86d3d75a8b77e44404b0f4261714306e3ad248d8ab0951"},
+    {file = "Brotli-1.1.0-cp313-cp313-macosx_10_13_universal2.whl", hash = "sha256:8bf32b98b75c13ec7cf774164172683d6e7891088f6316e54425fde1efc276d5"},
+    {file = "Brotli-1.1.0-cp313-cp313-macosx_10_13_x86_64.whl", hash = "sha256:7bc37c4d6b87fb1017ea28c9508b36bbcb0c3d18b4260fcdf08b200c74a6aee8"},
+    {file = "Brotli-1.1.0-cp313-cp313-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:3c0ef38c7a7014ffac184db9e04debe495d317cc9c6fb10071f7fefd93100a4f"},
+    {file = "Brotli-1.1.0-cp313-cp313-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:91d7cc2a76b5567591d12c01f019dd7afce6ba8cba6571187e21e2fc418ae648"},
+    {file = "Brotli-1.1.0-cp313-cp313-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:a93dde851926f4f2678e704fadeb39e16c35d8baebd5252c9fd94ce8ce68c4a0"},
+    {file = "Brotli-1.1.0-cp313-cp313-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:f0db75f47be8b8abc8d9e31bc7aad0547ca26f24a54e6fd10231d623f183d089"},
+    {file = "Brotli-1.1.0-cp313-cp313-musllinux_1_2_aarch64.whl", hash = "sha256:6967ced6730aed543b8673008b5a391c3b1076d834ca438bbd70635c73775368"},
+    {file = "Brotli-1.1.0-cp313-cp313-musllinux_1_2_i686.whl", hash = "sha256:7eedaa5d036d9336c95915035fb57422054014ebdeb6f3b42eac809928e40d0c"},
+    {file = "Brotli-1.1.0-cp313-cp313-musllinux_1_2_ppc64le.whl", hash = "sha256:d487f5432bf35b60ed625d7e1b448e2dc855422e87469e3f450aa5552b0eb284"},
+    {file = "Brotli-1.1.0-cp313-cp313-musllinux_1_2_x86_64.whl", hash = "sha256:832436e59afb93e1836081a20f324cb185836c617659b07b129141a8426973c7"},
+    {file = "Brotli-1.1.0-cp313-cp313-win32.whl", hash = "sha256:43395e90523f9c23a3d5bdf004733246fba087f2948f87ab28015f12359ca6a0"},
+    {file = "Brotli-1.1.0-cp313-cp313-win_amd64.whl", hash = "sha256:9011560a466d2eb3f5a6e4929cf4a09be405c64154e12df0dd72713f6500e32b"},
     {file = "Brotli-1.1.0-cp36-cp36m-macosx_10_9_x86_64.whl", hash = "sha256:a090ca607cbb6a34b0391776f0cb48062081f5f60ddcce5d11838e67a01928d1"},
     {file = "Brotli-1.1.0-cp36-cp36m-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:2de9d02f5bda03d27ede52e8cfe7b865b066fa49258cbab568720aa5be80a47d"},
     {file = "Brotli-1.1.0-cp36-cp36m-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:2333e30a5e00fe0fe55903c8832e08ee9c3b1382aacf4db26664a16528d51b4b"},
@@ -640,6 +710,10 @@ files = [
     {file = "Brotli-1.1.0-cp36-cp36m-musllinux_1_1_i686.whl", hash = "sha256:fd5f17ff8f14003595ab414e45fce13d073e0762394f957182e69035c9f3d7c2"},
     {file = "Brotli-1.1.0-cp36-cp36m-musllinux_1_1_ppc64le.whl", hash = "sha256:069a121ac97412d1fe506da790b3e69f52254b9df4eb665cd42460c837193354"},
     {file = "Brotli-1.1.0-cp36-cp36m-musllinux_1_1_x86_64.whl", hash = "sha256:e93dfc1a1165e385cc8239fab7c036fb2cd8093728cbd85097b284d7b99249a2"},
+    {file = "Brotli-1.1.0-cp36-cp36m-musllinux_1_2_aarch64.whl", hash = "sha256:aea440a510e14e818e67bfc4027880e2fb500c2ccb20ab21c7a7c8b5b4703d75"},
+    {file = "Brotli-1.1.0-cp36-cp36m-musllinux_1_2_i686.whl", hash = "sha256:6974f52a02321b36847cd19d1b8e381bf39939c21efd6ee2fc13a28b0d99348c"},
+    {file = "Brotli-1.1.0-cp36-cp36m-musllinux_1_2_ppc64le.whl", hash = "sha256:a7e53012d2853a07a4a79c00643832161a910674a893d296c9f1259859a289d2"},
+    {file = "Brotli-1.1.0-cp36-cp36m-musllinux_1_2_x86_64.whl", hash = "sha256:d7702622a8b40c49bffb46e1e3ba2e81268d5c04a34f460978c6b5517a34dd52"},
     {file = "Brotli-1.1.0-cp36-cp36m-win32.whl", hash = "sha256:a599669fd7c47233438a56936988a2478685e74854088ef5293802123b5b2460"},
     {file = "Brotli-1.1.0-cp36-cp36m-win_amd64.whl", hash = "sha256:d143fd47fad1db3d7c27a1b1d66162e855b5d50a89666af46e1679c496e8e579"},
     {file = "Brotli-1.1.0-cp37-cp37m-macosx_10_9_x86_64.whl", hash = "sha256:11d00ed0a83fa22d29bc6b64ef636c4552ebafcef57154b4ddd132f5638fbd1c"},
@@ -651,6 +725,10 @@ files = [
     {file = "Brotli-1.1.0-cp37-cp37m-musllinux_1_1_i686.whl", hash = "sha256:919e32f147ae93a09fe064d77d5ebf4e35502a8df75c29fb05788528e330fe74"},
     {file = "Brotli-1.1.0-cp37-cp37m-musllinux_1_1_ppc64le.whl", hash = "sha256:23032ae55523cc7bccb4f6a0bf368cd25ad9bcdcc1990b64a647e7bbcce9cb5b"},
     {file = "Brotli-1.1.0-cp37-cp37m-musllinux_1_1_x86_64.whl", hash = "sha256:224e57f6eac61cc449f498cc5f0e1725ba2071a3d4f48d5d9dffba42db196438"},
+    {file = "Brotli-1.1.0-cp37-cp37m-musllinux_1_2_aarch64.whl", hash = "sha256:cb1dac1770878ade83f2ccdf7d25e494f05c9165f5246b46a621cc849341dc01"},
+    {file = "Brotli-1.1.0-cp37-cp37m-musllinux_1_2_i686.whl", hash = "sha256:3ee8a80d67a4334482d9712b8e83ca6b1d9bc7e351931252ebef5d8f7335a547"},
+    {file = "Brotli-1.1.0-cp37-cp37m-musllinux_1_2_ppc64le.whl", hash = "sha256:5e55da2c8724191e5b557f8e18943b1b4839b8efc3ef60d65985bcf6f587dd38"},
+    {file = "Brotli-1.1.0-cp37-cp37m-musllinux_1_2_x86_64.whl", hash = "sha256:d342778ef319e1026af243ed0a07c97acf3bad33b9f29e7ae6a1f68fd083e90c"},
     {file = "Brotli-1.1.0-cp37-cp37m-win32.whl", hash = "sha256:587ca6d3cef6e4e868102672d3bd9dc9698c309ba56d41c2b9c85bbb903cdb95"},
     {file = "Brotli-1.1.0-cp37-cp37m-win_amd64.whl", hash = "sha256:2954c1c23f81c2eaf0b0717d9380bd348578a94161a65b3a2afc62c86467dd68"},
     {file = "Brotli-1.1.0-cp38-cp38-macosx_10_9_universal2.whl", hash = "sha256:efa8b278894b14d6da122a72fefcebc28445f2d3f880ac59d46c90f4c13be9a3"},
@@ -663,6 +741,10 @@ files = [
     {file = "Brotli-1.1.0-cp38-cp38-musllinux_1_1_i686.whl", hash = "sha256:1ab4fbee0b2d9098c74f3057b2bc055a8bd92ccf02f65944a241b4349229185a"},
     {file = "Brotli-1.1.0-cp38-cp38-musllinux_1_1_ppc64le.whl", hash = "sha256:141bd4d93984070e097521ed07e2575b46f817d08f9fa42b16b9b5f27b5ac088"},
     {file = "Brotli-1.1.0-cp38-cp38-musllinux_1_1_x86_64.whl", hash = "sha256:fce1473f3ccc4187f75b4690cfc922628aed4d3dd013d047f95a9b3919a86596"},
+    {file = "Brotli-1.1.0-cp38-cp38-musllinux_1_2_aarch64.whl", hash = "sha256:d2b35ca2c7f81d173d2fadc2f4f31e88cc5f7a39ae5b6db5513cf3383b0e0ec7"},
+    {file = "Brotli-1.1.0-cp38-cp38-musllinux_1_2_i686.whl", hash = "sha256:af6fa6817889314555aede9a919612b23739395ce767fe7fcbea9a80bf140fe5"},
+    {file = "Brotli-1.1.0-cp38-cp38-musllinux_1_2_ppc64le.whl", hash = "sha256:2feb1d960f760a575dbc5ab3b1c00504b24caaf6986e2dc2b01c09c87866a943"},
+    {file = "Brotli-1.1.0-cp38-cp38-musllinux_1_2_x86_64.whl", hash = "sha256:4410f84b33374409552ac9b6903507cdb31cd30d2501fc5ca13d18f73548444a"},
     {file = "Brotli-1.1.0-cp38-cp38-win32.whl", hash = "sha256:db85ecf4e609a48f4b29055f1e144231b90edc90af7481aa731ba2d059226b1b"},
     {file = "Brotli-1.1.0-cp38-cp38-win_amd64.whl", hash = "sha256:3d7954194c36e304e1523f55d7042c59dc53ec20dd4e9ea9d151f1b62b4415c0"},
     {file = "Brotli-1.1.0-cp39-cp39-macosx_10_9_universal2.whl", hash = "sha256:5fb2ce4b8045c78ebbc7b8f3c15062e435d47e7393cc57c25115cfd49883747a"},
@@ -675,6 +757,10 @@ files = [
     {file = "Brotli-1.1.0-cp39-cp39-musllinux_1_1_i686.whl", hash = "sha256:949f3b7c29912693cee0afcf09acd6ebc04c57af949d9bf77d6101ebb61e388c"},
     {file = "Brotli-1.1.0-cp39-cp39-musllinux_1_1_ppc64le.whl", hash = "sha256:89f4988c7203739d48c6f806f1e87a1d96e0806d44f0fba61dba81392c9e474d"},
     {file = "Brotli-1.1.0-cp39-cp39-musllinux_1_1_x86_64.whl", hash = "sha256:de6551e370ef19f8de1807d0a9aa2cdfdce2e85ce88b122fe9f6b2b076837e59"},
+    {file = "Brotli-1.1.0-cp39-cp39-musllinux_1_2_aarch64.whl", hash = "sha256:0737ddb3068957cf1b054899b0883830bb1fec522ec76b1098f9b6e0f02d9419"},
+    {file = "Brotli-1.1.0-cp39-cp39-musllinux_1_2_i686.whl", hash = "sha256:4f3607b129417e111e30637af1b56f24f7a49e64763253bbc275c75fa887d4b2"},
+    {file = "Brotli-1.1.0-cp39-cp39-musllinux_1_2_ppc64le.whl", hash = "sha256:6c6e0c425f22c1c719c42670d561ad682f7bfeeef918edea971a79ac5252437f"},
+    {file = "Brotli-1.1.0-cp39-cp39-musllinux_1_2_x86_64.whl", hash = "sha256:494994f807ba0b92092a163a0a283961369a65f6cbe01e8891132b7a320e61eb"},
     {file = "Brotli-1.1.0-cp39-cp39-win32.whl", hash = "sha256:f0d8a7a6b5983c2496e364b969f0e526647a06b075d034f3297dc66f3b360c64"},
     {file = "Brotli-1.1.0-cp39-cp39-win_amd64.whl", hash = "sha256:cdad5b9014d83ca68c25d2e9444e28e967ef16e80f6b436918c700c117a85467"},
     {file = "Brotli-1.1.0.tar.gz", hash = "sha256:81de08ac11bcb85841e440c13611c00b67d3bf82698314928d0b676362546724"},
@@ -3623,7 +3709,7 @@ url = "runtimes/catboost"
 
 [[package]]
 name = "mlserver-huggingface"
-version = "1.6.0.dev2"
+version = "2.0.10"
 description = "HuggingFace runtime for MLServer"
 optional = false
 python-versions = ">=3.9,<3.12"
@@ -3631,11 +3717,16 @@ files = []
 develop = true
 
 [package.dependencies]
+accelerate = "^0.27.2"
+bitsandbytes = "^0.42.0"
 mlserver = "*"
 optimum = {version = ">=1.4,<2.0", extras = ["onnxruntime"]}
 pillow = "*"
 pydantic = "2.7.1"
+sentence-transformers = "2.5.1"
 tensorflow = "*"
+timm = "^1.0.13"
+transformers = "4.41.2"
 
 [package.source]
 type = "directory"
@@ -5014,6 +5105,36 @@ files = [
     {file = "protobuf-3.20.3.tar.gz", hash = "sha256:2e3427429c9cffebf259491be0af70189607f365c2f41c7c3764af6f337105f2"},
 ]
 
+[[package]]
+name = "psutil"
+version = "6.1.1"
+description = "Cross-platform lib for process and system monitoring in Python."
+optional = false
+python-versions = "!=3.0.*,!=3.1.*,!=3.2.*,!=3.3.*,!=3.4.*,!=3.5.*,>=2.7"
+files = [
+    {file = "psutil-6.1.1-cp27-cp27m-macosx_10_9_x86_64.whl", hash = "sha256:9ccc4316f24409159897799b83004cb1e24f9819b0dcf9c0b68bdcb6cefee6a8"},
+    {file = "psutil-6.1.1-cp27-cp27m-manylinux2010_i686.whl", hash = "sha256:ca9609c77ea3b8481ab005da74ed894035936223422dc591d6772b147421f777"},
+    {file = "psutil-6.1.1-cp27-cp27m-manylinux2010_x86_64.whl", hash = "sha256:8df0178ba8a9e5bc84fed9cfa61d54601b371fbec5c8eebad27575f1e105c0d4"},
+    {file = "psutil-6.1.1-cp27-cp27mu-manylinux2010_i686.whl", hash = "sha256:1924e659d6c19c647e763e78670a05dbb7feaf44a0e9c94bf9e14dfc6ba50468"},
+    {file = "psutil-6.1.1-cp27-cp27mu-manylinux2010_x86_64.whl", hash = "sha256:018aeae2af92d943fdf1da6b58665124897cfc94faa2ca92098838f83e1b1bca"},
+    {file = "psutil-6.1.1-cp27-none-win32.whl", hash = "sha256:6d4281f5bbca041e2292be3380ec56a9413b790579b8e593b1784499d0005dac"},
+    {file = "psutil-6.1.1-cp27-none-win_amd64.whl", hash = "sha256:c777eb75bb33c47377c9af68f30e9f11bc78e0f07fbf907be4a5d70b2fe5f030"},
+    {file = "psutil-6.1.1-cp36-abi3-macosx_10_9_x86_64.whl", hash = "sha256:fc0ed7fe2231a444fc219b9c42d0376e0a9a1a72f16c5cfa0f68d19f1a0663e8"},
+    {file = "psutil-6.1.1-cp36-abi3-macosx_11_0_arm64.whl", hash = "sha256:0bdd4eab935276290ad3cb718e9809412895ca6b5b334f5a9111ee6d9aff9377"},
+    {file = "psutil-6.1.1-cp36-abi3-manylinux_2_12_i686.manylinux2010_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:b6e06c20c05fe95a3d7302d74e7097756d4ba1247975ad6905441ae1b5b66003"},
+    {file = "psutil-6.1.1-cp36-abi3-manylinux_2_12_x86_64.manylinux2010_x86_64.manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:97f7cb9921fbec4904f522d972f0c0e1f4fabbdd4e0287813b21215074a0f160"},
+    {file = "psutil-6.1.1-cp36-abi3-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:33431e84fee02bc84ea36d9e2c4a6d395d479c9dd9bba2376c1f6ee8f3a4e0b3"},
+    {file = "psutil-6.1.1-cp36-cp36m-win32.whl", hash = "sha256:384636b1a64b47814437d1173be1427a7c83681b17a450bfc309a1953e329603"},
+    {file = "psutil-6.1.1-cp36-cp36m-win_amd64.whl", hash = "sha256:8be07491f6ebe1a693f17d4f11e69d0dc1811fa082736500f649f79df7735303"},
+    {file = "psutil-6.1.1-cp37-abi3-win32.whl", hash = "sha256:eaa912e0b11848c4d9279a93d7e2783df352b082f40111e078388701fd479e53"},
+    {file = "psutil-6.1.1-cp37-abi3-win_amd64.whl", hash = "sha256:f35cfccb065fff93529d2afb4a2e89e363fe63ca1e4a5da22b603a85833c2649"},
+    {file = "psutil-6.1.1.tar.gz", hash = "sha256:cf8496728c18f2d0b45198f06895be52f36611711746b7f30c464b422b50e2f5"},
+]
+
+[package.extras]
+dev = ["abi3audit", "black", "check-manifest", "coverage", "packaging", "pylint", "pyperf", "pypinfo", "pytest-cov", "requests", "rstcheck", "ruff", "sphinx", "sphinx_rtd_theme", "toml-sort", "twine", "virtualenv", "vulture", "wheel"]
+test = ["pytest", "pytest-xdist", "setuptools"]
+
 [[package]]
 name = "py-grpc-prometheus"
 version = "0.8.0"
@@ -5684,7 +5805,6 @@ files = [
     {file = "PyYAML-6.0.1-cp310-cp310-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:69b023b2b4daa7548bcfbd4aa3da05b3a74b772db9e23b982788168117739938"},
     {file = "PyYAML-6.0.1-cp310-cp310-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:81e0b275a9ecc9c0c0c07b4b90ba548307583c125f54d5b6946cfee6360c733d"},
     {file = "PyYAML-6.0.1-cp310-cp310-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:ba336e390cd8e4d1739f42dfe9bb83a3cc2e80f567d8805e11b46f4a943f5515"},
-    {file = "PyYAML-6.0.1-cp310-cp310-musllinux_1_1_x86_64.whl", hash = "sha256:326c013efe8048858a6d312ddd31d56e468118ad4cdeda36c719bf5bb6192290"},
     {file = "PyYAML-6.0.1-cp310-cp310-win32.whl", hash = "sha256:bd4af7373a854424dabd882decdc5579653d7868b8fb26dc7d0e99f823aa5924"},
     {file = "PyYAML-6.0.1-cp310-cp310-win_amd64.whl", hash = "sha256:fd1592b3fdf65fff2ad0004b5e363300ef59ced41c2e6b3a99d4089fa8c5435d"},
     {file = "PyYAML-6.0.1-cp311-cp311-macosx_10_9_x86_64.whl", hash = "sha256:6965a7bc3cf88e5a1c3bd2e0b5c22f8d677dc88a455344035f03399034eb3007"},
@@ -5692,16 +5812,8 @@ files = [
     {file = "PyYAML-6.0.1-cp311-cp311-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:42f8152b8dbc4fe7d96729ec2b99c7097d656dc1213a3229ca5383f973a5ed6d"},
     {file = "PyYAML-6.0.1-cp311-cp311-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:062582fca9fabdd2c8b54a3ef1c978d786e0f6b3a1510e0ac93ef59e0ddae2bc"},
     {file = "PyYAML-6.0.1-cp311-cp311-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:d2b04aac4d386b172d5b9692e2d2da8de7bfb6c387fa4f801fbf6fb2e6ba4673"},
-    {file = "PyYAML-6.0.1-cp311-cp311-musllinux_1_1_x86_64.whl", hash = "sha256:e7d73685e87afe9f3b36c799222440d6cf362062f78be1013661b00c5c6f678b"},
     {file = "PyYAML-6.0.1-cp311-cp311-win32.whl", hash = "sha256:1635fd110e8d85d55237ab316b5b011de701ea0f29d07611174a1b42f1444741"},
     {file = "PyYAML-6.0.1-cp311-cp311-win_amd64.whl", hash = "sha256:bf07ee2fef7014951eeb99f56f39c9bb4af143d8aa3c21b1677805985307da34"},
-    {file = "PyYAML-6.0.1-cp312-cp312-macosx_10_9_x86_64.whl", hash = "sha256:855fb52b0dc35af121542a76b9a84f8d1cd886ea97c84703eaa6d88e37a2ad28"},
-    {file = "PyYAML-6.0.1-cp312-cp312-macosx_11_0_arm64.whl", hash = "sha256:40df9b996c2b73138957fe23a16a4f0ba614f4c0efce1e9406a184b6d07fa3a9"},
-    {file = "PyYAML-6.0.1-cp312-cp312-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:a08c6f0fe150303c1c6b71ebcd7213c2858041a7e01975da3a99aed1e7a378ef"},
-    {file = "PyYAML-6.0.1-cp312-cp312-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:6c22bec3fbe2524cde73d7ada88f6566758a8f7227bfbf93a408a9d86bcc12a0"},
-    {file = "PyYAML-6.0.1-cp312-cp312-musllinux_1_1_x86_64.whl", hash = "sha256:8d4e9c88387b0f5c7d5f281e55304de64cf7f9c0021a3525bd3b1c542da3b0e4"},
-    {file = "PyYAML-6.0.1-cp312-cp312-win32.whl", hash = "sha256:d483d2cdf104e7c9fa60c544d92981f12ad66a457afae824d146093b8c294c54"},
-    {file = "PyYAML-6.0.1-cp312-cp312-win_amd64.whl", hash = "sha256:0d3304d8c0adc42be59c5f8a4d9e3d7379e6955ad754aa9d6ab7a398b59dd1df"},
     {file = "PyYAML-6.0.1-cp36-cp36m-macosx_10_9_x86_64.whl", hash = "sha256:50550eb667afee136e9a77d6dc71ae76a44df8b3e51e41b77f6de2932bfe0f47"},
     {file = "PyYAML-6.0.1-cp36-cp36m-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:1fe35611261b29bd1de0070f0b2f47cb6ff71fa6595c077e42bd0c419fa27b98"},
     {file = "PyYAML-6.0.1-cp36-cp36m-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:704219a11b772aea0d8ecd7058d0082713c3562b4e271b849ad7dc4a5c90c13c"},
@@ -5718,7 +5830,6 @@ files = [
     {file = "PyYAML-6.0.1-cp38-cp38-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:a0cd17c15d3bb3fa06978b4e8958dcdc6e0174ccea823003a106c7d4d7899ac5"},
     {file = "PyYAML-6.0.1-cp38-cp38-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:28c119d996beec18c05208a8bd78cbe4007878c6dd15091efb73a30e90539696"},
     {file = "PyYAML-6.0.1-cp38-cp38-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:7e07cbde391ba96ab58e532ff4803f79c4129397514e1413a7dc761ccd755735"},
-    {file = "PyYAML-6.0.1-cp38-cp38-musllinux_1_1_x86_64.whl", hash = "sha256:49a183be227561de579b4a36efbb21b3eab9651dd81b1858589f796549873dd6"},
     {file = "PyYAML-6.0.1-cp38-cp38-win32.whl", hash = "sha256:184c5108a2aca3c5b3d3bf9395d50893a7ab82a38004c8f61c258d4428e80206"},
     {file = "PyYAML-6.0.1-cp38-cp38-win_amd64.whl", hash = "sha256:1e2722cc9fbb45d9b87631ac70924c11d3a401b2d7f410cc0e3bbf249f2dca62"},
     {file = "PyYAML-6.0.1-cp39-cp39-macosx_10_9_x86_64.whl", hash = "sha256:9eb6caa9a297fc2c2fb8862bc5370d0303ddba53ba97e71f08023b6cd73d16a8"},
@@ -5726,7 +5837,6 @@ files = [
     {file = "PyYAML-6.0.1-cp39-cp39-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:5773183b6446b2c99bb77e77595dd486303b4faab2b086e7b17bc6bef28865f6"},
     {file = "PyYAML-6.0.1-cp39-cp39-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:b786eecbdf8499b9ca1d697215862083bd6d2a99965554781d0d8d1ad31e13a0"},
     {file = "PyYAML-6.0.1-cp39-cp39-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:bc1bf2925a1ecd43da378f4db9e4f799775d6367bdb94671027b73b393a7c42c"},
-    {file = "PyYAML-6.0.1-cp39-cp39-musllinux_1_1_x86_64.whl", hash = "sha256:04ac92ad1925b2cff1db0cfebffb6ffc43457495c9b3c39d3fcae417d7125dc5"},
     {file = "PyYAML-6.0.1-cp39-cp39-win32.whl", hash = "sha256:faca3bdcf85b2fc05d06ff3fbc1f83e1391b3e724afa3feba7d13eeab355484c"},
     {file = "PyYAML-6.0.1-cp39-cp39-win_amd64.whl", hash = "sha256:510c9deebc5c0225e8c96813043e62b680ba2f9c50a08d3724c7f28a747d1486"},
     {file = "PyYAML-6.0.1.tar.gz", hash = "sha256:bfdf460b1736c775f2ba9f6a92bca30bc2095067b8a9d77876d1fad6cc3b4a43"},
@@ -6210,6 +6320,27 @@ dev = ["cython-lint (>=0.12.2)", "doit (>=0.36.0)", "mypy", "pycodestyle", "pyde
 doc = ["jupyterlite-pyodide-kernel", "jupyterlite-sphinx (>=0.12.0)", "jupytext", "matplotlib (>=3.5)", "myst-nb", "numpydoc", "pooch", "pydata-sphinx-theme (>=0.15.2)", "sphinx (>=5.0.0)", "sphinx-design (>=0.4.0)"]
 test = ["array-api-strict", "asv", "gmpy2", "hypothesis (>=6.30)", "mpmath", "pooch", "pytest", "pytest-cov", "pytest-timeout", "pytest-xdist", "scikit-umfpack", "threadpoolctl"]
 
+[[package]]
+name = "sentence-transformers"
+version = "2.5.1"
+description = "Multilingual text embeddings"
+optional = false
+python-versions = ">=3.8.0"
+files = [
+    {file = "sentence-transformers-2.5.1.tar.gz", hash = "sha256:754bf2b2623eb46904fd9c72ff89a0f90200fe141a8d45b03e83bc6d51718153"},
+    {file = "sentence_transformers-2.5.1-py3-none-any.whl", hash = "sha256:f12346f7fca06ed1198d24235cb9114a74665506f7c30044e0a6f12de7eeeb77"},
+]
+
+[package.dependencies]
+huggingface-hub = ">=0.15.1"
+numpy = "*"
+Pillow = "*"
+scikit-learn = "*"
+scipy = "*"
+torch = ">=1.11.0"
+tqdm = "*"
+transformers = ">=4.32.0,<5.0.0"
+
 [[package]]
 name = "sentencepiece"
 version = "0.2.0"
@@ -7308,6 +7439,24 @@ numpy = "*"
 [package.extras]
 all = ["defusedxml", "fsspec", "imagecodecs (>=2023.8.12)", "lxml", "matplotlib", "zarr"]
 
+[[package]]
+name = "timm"
+version = "1.0.14"
+description = "PyTorch Image Models"
+optional = false
+python-versions = ">=3.8"
+files = [
+    {file = "timm-1.0.14-py3-none-any.whl", hash = "sha256:16653695ff420cf4e87e384c8654f2ee6be2070bcb4a0e840b5c6764ee0547ec"},
+    {file = "timm-1.0.14.tar.gz", hash = "sha256:00a7f2cc04ce3ed8f80476bbb7eea27eac8cf6f2d59b5e9aa9cdd375dd6550db"},
+]
+
+[package.dependencies]
+huggingface_hub = "*"
+pyyaml = "*"
+safetensors = "*"
+torch = "*"
+torchvision = "*"
+
 [[package]]
 name = "tokenizers"
 version = "0.19.1"
@@ -7625,19 +7774,19 @@ telegram = ["requests"]
 
 [[package]]
 name = "transformers"
-version = "4.42.4"
+version = "4.41.2"
 description = "State-of-the-art Machine Learning for JAX, PyTorch and TensorFlow"
 optional = false
 python-versions = ">=3.8.0"
 files = [
-    {file = "transformers-4.42.4-py3-none-any.whl", hash = "sha256:6d59061392d0f1da312af29c962df9017ff3c0108c681a56d1bc981004d16d24"},
-    {file = "transformers-4.42.4.tar.gz", hash = "sha256:f956e25e24df851f650cb2c158b6f4352dfae9d702f04c113ed24fc36ce7ae2d"},
+    {file = "transformers-4.41.2-py3-none-any.whl", hash = "sha256:05555d20e43f808de1ef211ab64803cdb513170cef70d29a888b589caebefc67"},
+    {file = "transformers-4.41.2.tar.gz", hash = "sha256:80a4db216533d573e9cc7388646c31ed9480918feb7c55eb211249cb23567f87"},
 ]
 
 [package.dependencies]
 filelock = "*"
-huggingface-hub = ">=0.23.2,<1.0"
-numpy = ">=1.17,<2.0"
+huggingface-hub = ">=0.23.0,<1.0"
+numpy = ">=1.17"
 packaging = ">=20.0"
 protobuf = {version = "*", optional = true, markers = "extra == \"sentencepiece\""}
 pyyaml = ">=5.1"
@@ -7651,15 +7800,14 @@ tqdm = ">=4.27"
 [package.extras]
 accelerate = ["accelerate (>=0.21.0)"]
 agents = ["Pillow (>=10.0.1,<=15.0)", "accelerate (>=0.21.0)", "datasets (!=2.5.0)", "diffusers", "opencv-python", "sentencepiece (>=0.1.91,!=0.1.92)", "torch"]
-all = ["Pillow (>=10.0.1,<=15.0)", "accelerate (>=0.21.0)", "av (==9.2.0)", "codecarbon (==1.2.0)", "decord (==0.6.0)", "flax (>=0.4.1,<=0.7.0)", "jax (>=0.4.1,<=0.4.13)", "jaxlib (>=0.4.1,<=0.4.13)", "kenlm", "keras-nlp (>=0.3.1)", "librosa", "onnxconverter-common", "optax (>=0.0.8,<=0.1.4)", "optuna", "phonemizer", "protobuf", "pyctcdecode (>=0.4.0)", "ray[tune] (>=2.7.0)", "scipy (<1.13.0)", "sentencepiece (>=0.1.91,!=0.1.92)", "sigopt", "tensorflow (>2.9,<2.16)", "tensorflow-text (<2.16)", "tf2onnx", "timm (<=0.9.16)", "tokenizers (>=0.19,<0.20)", "torch", "torchaudio", "torchvision"]
+all = ["Pillow (>=10.0.1,<=15.0)", "accelerate (>=0.21.0)", "av (==9.2.0)", "codecarbon (==1.2.0)", "decord (==0.6.0)", "flax (>=0.4.1,<=0.7.0)", "jax (>=0.4.1,<=0.4.13)", "jaxlib (>=0.4.1,<=0.4.13)", "kenlm", "keras-nlp (>=0.3.1)", "librosa", "onnxconverter-common", "optax (>=0.0.8,<=0.1.4)", "optuna", "phonemizer", "protobuf", "pyctcdecode (>=0.4.0)", "ray[tune] (>=2.7.0)", "scipy (<1.13.0)", "sentencepiece (>=0.1.91,!=0.1.92)", "sigopt", "tensorflow (>2.9,<2.16)", "tensorflow-text (<2.16)", "tf2onnx", "timm", "tokenizers (>=0.19,<0.20)", "torch", "torchaudio", "torchvision"]
 audio = ["kenlm", "librosa", "phonemizer", "pyctcdecode (>=0.4.0)"]
-benchmark = ["optimum-benchmark (>=0.2.0)"]
 codecarbon = ["codecarbon (==1.2.0)"]
 deepspeed = ["accelerate (>=0.21.0)", "deepspeed (>=0.9.3)"]
-deepspeed-testing = ["GitPython (<3.1.19)", "accelerate (>=0.21.0)", "beautifulsoup4", "cookiecutter (==1.7.3)", "datasets (!=2.5.0)", "deepspeed (>=0.9.3)", "dill (<0.3.5)", "evaluate (>=0.2.0)", "faiss-cpu", "nltk", "optuna", "parameterized", "protobuf", "psutil", "pydantic", "pytest (>=7.2.0,<8.0.0)", "pytest-rich", "pytest-timeout", "pytest-xdist", "rjieba", "rouge-score (!=0.0.7,!=0.0.8,!=0.1,!=0.1.1)", "ruff (==0.4.4)", "sacrebleu (>=1.4.12,<2.0.0)", "sacremoses", "sentencepiece (>=0.1.91,!=0.1.92)", "tensorboard", "timeout-decorator"]
-dev = ["GitPython (<3.1.19)", "Pillow (>=10.0.1,<=15.0)", "accelerate (>=0.21.0)", "av (==9.2.0)", "beautifulsoup4", "codecarbon (==1.2.0)", "cookiecutter (==1.7.3)", "datasets (!=2.5.0)", "decord (==0.6.0)", "dill (<0.3.5)", "evaluate (>=0.2.0)", "faiss-cpu", "flax (>=0.4.1,<=0.7.0)", "fugashi (>=1.0)", "ipadic (>=1.0.0,<2.0)", "isort (>=5.5.4)", "jax (>=0.4.1,<=0.4.13)", "jaxlib (>=0.4.1,<=0.4.13)", "kenlm", "keras-nlp (>=0.3.1)", "librosa", "nltk", "onnxconverter-common", "optax (>=0.0.8,<=0.1.4)", "optuna", "parameterized", "phonemizer", "protobuf", "psutil", "pyctcdecode (>=0.4.0)", "pydantic", "pytest (>=7.2.0,<8.0.0)", "pytest-rich", "pytest-timeout", "pytest-xdist", "ray[tune] (>=2.7.0)", "rhoknp (>=1.1.0,<1.3.1)", "rjieba", "rouge-score (!=0.0.7,!=0.0.8,!=0.1,!=0.1.1)", "ruff (==0.4.4)", "sacrebleu (>=1.4.12,<2.0.0)", "sacremoses", "scikit-learn", "scipy (<1.13.0)", "sentencepiece (>=0.1.91,!=0.1.92)", "sigopt", "sudachidict-core (>=20220729)", "sudachipy (>=0.6.6)", "tensorboard", "tensorflow (>2.9,<2.16)", "tensorflow-text (<2.16)", "tf2onnx", "timeout-decorator", "timm (<=0.9.16)", "tokenizers (>=0.19,<0.20)", "torch", "torchaudio", "torchvision", "unidic (>=1.0.2)", "unidic-lite (>=1.0.7)", "urllib3 (<2.0.0)"]
-dev-tensorflow = ["GitPython (<3.1.19)", "Pillow (>=10.0.1,<=15.0)", "beautifulsoup4", "cookiecutter (==1.7.3)", "datasets (!=2.5.0)", "dill (<0.3.5)", "evaluate (>=0.2.0)", "faiss-cpu", "isort (>=5.5.4)", "kenlm", "keras-nlp (>=0.3.1)", "librosa", "nltk", "onnxconverter-common", "onnxruntime (>=1.4.0)", "onnxruntime-tools (>=1.4.2)", "parameterized", "phonemizer", "protobuf", "psutil", "pyctcdecode (>=0.4.0)", "pydantic", "pytest (>=7.2.0,<8.0.0)", "pytest-rich", "pytest-timeout", "pytest-xdist", "rjieba", "rouge-score (!=0.0.7,!=0.0.8,!=0.1,!=0.1.1)", "ruff (==0.4.4)", "sacrebleu (>=1.4.12,<2.0.0)", "sacremoses", "scikit-learn", "sentencepiece (>=0.1.91,!=0.1.92)", "tensorboard", "tensorflow (>2.9,<2.16)", "tensorflow-text (<2.16)", "tf2onnx", "timeout-decorator", "tokenizers (>=0.19,<0.20)", "urllib3 (<2.0.0)"]
-dev-torch = ["GitPython (<3.1.19)", "Pillow (>=10.0.1,<=15.0)", "accelerate (>=0.21.0)", "beautifulsoup4", "codecarbon (==1.2.0)", "cookiecutter (==1.7.3)", "datasets (!=2.5.0)", "dill (<0.3.5)", "evaluate (>=0.2.0)", "faiss-cpu", "fugashi (>=1.0)", "ipadic (>=1.0.0,<2.0)", "isort (>=5.5.4)", "kenlm", "librosa", "nltk", "onnxruntime (>=1.4.0)", "onnxruntime-tools (>=1.4.2)", "optuna", "parameterized", "phonemizer", "protobuf", "psutil", "pyctcdecode (>=0.4.0)", "pydantic", "pytest (>=7.2.0,<8.0.0)", "pytest-rich", "pytest-timeout", "pytest-xdist", "ray[tune] (>=2.7.0)", "rhoknp (>=1.1.0,<1.3.1)", "rjieba", "rouge-score (!=0.0.7,!=0.0.8,!=0.1,!=0.1.1)", "ruff (==0.4.4)", "sacrebleu (>=1.4.12,<2.0.0)", "sacremoses", "scikit-learn", "sentencepiece (>=0.1.91,!=0.1.92)", "sigopt", "sudachidict-core (>=20220729)", "sudachipy (>=0.6.6)", "tensorboard", "timeout-decorator", "timm (<=0.9.16)", "tokenizers (>=0.19,<0.20)", "torch", "torchaudio", "torchvision", "unidic (>=1.0.2)", "unidic-lite (>=1.0.7)", "urllib3 (<2.0.0)"]
+deepspeed-testing = ["GitPython (<3.1.19)", "accelerate (>=0.21.0)", "beautifulsoup4", "cookiecutter (==1.7.3)", "datasets (!=2.5.0)", "deepspeed (>=0.9.3)", "dill (<0.3.5)", "evaluate (>=0.2.0)", "faiss-cpu", "nltk", "optuna", "parameterized", "protobuf", "psutil", "pydantic", "pytest (>=7.2.0,<8.0.0)", "pytest-rich", "pytest-timeout", "pytest-xdist", "rjieba", "rouge-score (!=0.0.7,!=0.0.8,!=0.1,!=0.1.1)", "ruff (==0.1.5)", "sacrebleu (>=1.4.12,<2.0.0)", "sacremoses", "sentencepiece (>=0.1.91,!=0.1.92)", "tensorboard", "timeout-decorator"]
+dev = ["GitPython (<3.1.19)", "Pillow (>=10.0.1,<=15.0)", "accelerate (>=0.21.0)", "av (==9.2.0)", "beautifulsoup4", "codecarbon (==1.2.0)", "cookiecutter (==1.7.3)", "datasets (!=2.5.0)", "decord (==0.6.0)", "dill (<0.3.5)", "evaluate (>=0.2.0)", "faiss-cpu", "flax (>=0.4.1,<=0.7.0)", "fugashi (>=1.0)", "ipadic (>=1.0.0,<2.0)", "isort (>=5.5.4)", "jax (>=0.4.1,<=0.4.13)", "jaxlib (>=0.4.1,<=0.4.13)", "kenlm", "keras-nlp (>=0.3.1)", "librosa", "nltk", "onnxconverter-common", "optax (>=0.0.8,<=0.1.4)", "optuna", "parameterized", "phonemizer", "protobuf", "psutil", "pyctcdecode (>=0.4.0)", "pydantic", "pytest (>=7.2.0,<8.0.0)", "pytest-rich", "pytest-timeout", "pytest-xdist", "ray[tune] (>=2.7.0)", "rhoknp (>=1.1.0,<1.3.1)", "rjieba", "rouge-score (!=0.0.7,!=0.0.8,!=0.1,!=0.1.1)", "ruff (==0.1.5)", "sacrebleu (>=1.4.12,<2.0.0)", "sacremoses", "scikit-learn", "scipy (<1.13.0)", "sentencepiece (>=0.1.91,!=0.1.92)", "sigopt", "sudachidict-core (>=20220729)", "sudachipy (>=0.6.6)", "tensorboard", "tensorflow (>2.9,<2.16)", "tensorflow-text (<2.16)", "tf2onnx", "timeout-decorator", "timm", "tokenizers (>=0.19,<0.20)", "torch", "torchaudio", "torchvision", "unidic (>=1.0.2)", "unidic-lite (>=1.0.7)", "urllib3 (<2.0.0)"]
+dev-tensorflow = ["GitPython (<3.1.19)", "Pillow (>=10.0.1,<=15.0)", "beautifulsoup4", "cookiecutter (==1.7.3)", "datasets (!=2.5.0)", "dill (<0.3.5)", "evaluate (>=0.2.0)", "faiss-cpu", "isort (>=5.5.4)", "kenlm", "keras-nlp (>=0.3.1)", "librosa", "nltk", "onnxconverter-common", "onnxruntime (>=1.4.0)", "onnxruntime-tools (>=1.4.2)", "parameterized", "phonemizer", "protobuf", "psutil", "pyctcdecode (>=0.4.0)", "pydantic", "pytest (>=7.2.0,<8.0.0)", "pytest-rich", "pytest-timeout", "pytest-xdist", "rjieba", "rouge-score (!=0.0.7,!=0.0.8,!=0.1,!=0.1.1)", "ruff (==0.1.5)", "sacrebleu (>=1.4.12,<2.0.0)", "sacremoses", "scikit-learn", "sentencepiece (>=0.1.91,!=0.1.92)", "tensorboard", "tensorflow (>2.9,<2.16)", "tensorflow-text (<2.16)", "tf2onnx", "timeout-decorator", "tokenizers (>=0.19,<0.20)", "urllib3 (<2.0.0)"]
+dev-torch = ["GitPython (<3.1.19)", "Pillow (>=10.0.1,<=15.0)", "accelerate (>=0.21.0)", "beautifulsoup4", "codecarbon (==1.2.0)", "cookiecutter (==1.7.3)", "datasets (!=2.5.0)", "dill (<0.3.5)", "evaluate (>=0.2.0)", "faiss-cpu", "fugashi (>=1.0)", "ipadic (>=1.0.0,<2.0)", "isort (>=5.5.4)", "kenlm", "librosa", "nltk", "onnxruntime (>=1.4.0)", "onnxruntime-tools (>=1.4.2)", "optuna", "parameterized", "phonemizer", "protobuf", "psutil", "pyctcdecode (>=0.4.0)", "pydantic", "pytest (>=7.2.0,<8.0.0)", "pytest-rich", "pytest-timeout", "pytest-xdist", "ray[tune] (>=2.7.0)", "rhoknp (>=1.1.0,<1.3.1)", "rjieba", "rouge-score (!=0.0.7,!=0.0.8,!=0.1,!=0.1.1)", "ruff (==0.1.5)", "sacrebleu (>=1.4.12,<2.0.0)", "sacremoses", "scikit-learn", "sentencepiece (>=0.1.91,!=0.1.92)", "sigopt", "sudachidict-core (>=20220729)", "sudachipy (>=0.6.6)", "tensorboard", "timeout-decorator", "timm", "tokenizers (>=0.19,<0.20)", "torch", "torchaudio", "torchvision", "unidic (>=1.0.2)", "unidic-lite (>=1.0.7)", "urllib3 (<2.0.0)"]
 flax = ["flax (>=0.4.1,<=0.7.0)", "jax (>=0.4.1,<=0.4.13)", "jaxlib (>=0.4.1,<=0.4.13)", "optax (>=0.0.8,<=0.1.4)", "scipy (<1.13.0)"]
 flax-speech = ["kenlm", "librosa", "phonemizer", "pyctcdecode (>=0.4.0)"]
 ftfy = ["ftfy"]
@@ -7670,26 +7818,25 @@ natten = ["natten (>=0.14.6,<0.15.0)"]
 onnx = ["onnxconverter-common", "onnxruntime (>=1.4.0)", "onnxruntime-tools (>=1.4.2)", "tf2onnx"]
 onnxruntime = ["onnxruntime (>=1.4.0)", "onnxruntime-tools (>=1.4.2)"]
 optuna = ["optuna"]
-quality = ["GitPython (<3.1.19)", "datasets (!=2.5.0)", "isort (>=5.5.4)", "ruff (==0.4.4)", "urllib3 (<2.0.0)"]
+quality = ["GitPython (<3.1.19)", "datasets (!=2.5.0)", "isort (>=5.5.4)", "ruff (==0.1.5)", "urllib3 (<2.0.0)"]
 ray = ["ray[tune] (>=2.7.0)"]
 retrieval = ["datasets (!=2.5.0)", "faiss-cpu"]
-ruff = ["ruff (==0.4.4)"]
 sagemaker = ["sagemaker (>=2.31.0)"]
 sentencepiece = ["protobuf", "sentencepiece (>=0.1.91,!=0.1.92)"]
 serving = ["fastapi", "pydantic", "starlette", "uvicorn"]
 sigopt = ["sigopt"]
 sklearn = ["scikit-learn"]
 speech = ["kenlm", "librosa", "phonemizer", "pyctcdecode (>=0.4.0)", "torchaudio"]
-testing = ["GitPython (<3.1.19)", "beautifulsoup4", "cookiecutter (==1.7.3)", "datasets (!=2.5.0)", "dill (<0.3.5)", "evaluate (>=0.2.0)", "faiss-cpu", "nltk", "parameterized", "psutil", "pydantic", "pytest (>=7.2.0,<8.0.0)", "pytest-rich", "pytest-timeout", "pytest-xdist", "rjieba", "rouge-score (!=0.0.7,!=0.0.8,!=0.1,!=0.1.1)", "ruff (==0.4.4)", "sacrebleu (>=1.4.12,<2.0.0)", "sacremoses", "sentencepiece (>=0.1.91,!=0.1.92)", "tensorboard", "timeout-decorator"]
+testing = ["GitPython (<3.1.19)", "beautifulsoup4", "cookiecutter (==1.7.3)", "datasets (!=2.5.0)", "dill (<0.3.5)", "evaluate (>=0.2.0)", "faiss-cpu", "nltk", "parameterized", "psutil", "pydantic", "pytest (>=7.2.0,<8.0.0)", "pytest-rich", "pytest-timeout", "pytest-xdist", "rjieba", "rouge-score (!=0.0.7,!=0.0.8,!=0.1,!=0.1.1)", "ruff (==0.1.5)", "sacrebleu (>=1.4.12,<2.0.0)", "sacremoses", "sentencepiece (>=0.1.91,!=0.1.92)", "tensorboard", "timeout-decorator"]
 tf = ["keras-nlp (>=0.3.1)", "onnxconverter-common", "tensorflow (>2.9,<2.16)", "tensorflow-text (<2.16)", "tf2onnx"]
-tf-cpu = ["keras (>2.9,<2.16)", "keras-nlp (>=0.3.1)", "onnxconverter-common", "tensorflow-cpu (>2.9,<2.16)", "tensorflow-probability (<0.24)", "tensorflow-text (<2.16)", "tf2onnx"]
+tf-cpu = ["keras (>2.9,<2.16)", "keras-nlp (>=0.3.1)", "onnxconverter-common", "tensorflow-cpu (>2.9,<2.16)", "tensorflow-probability (<2.16)", "tensorflow-text (<2.16)", "tf2onnx"]
 tf-speech = ["kenlm", "librosa", "phonemizer", "pyctcdecode (>=0.4.0)"]
-timm = ["timm (<=0.9.16)"]
+timm = ["timm"]
 tokenizers = ["tokenizers (>=0.19,<0.20)"]
 torch = ["accelerate (>=0.21.0)", "torch"]
 torch-speech = ["kenlm", "librosa", "phonemizer", "pyctcdecode (>=0.4.0)", "torchaudio"]
 torch-vision = ["Pillow (>=10.0.1,<=15.0)", "torchvision"]
-torchhub = ["filelock", "huggingface-hub (>=0.23.2,<1.0)", "importlib-metadata", "numpy (>=1.17,<2.0)", "packaging (>=20.0)", "protobuf", "regex (!=2019.12.17)", "requests", "sentencepiece (>=0.1.91,!=0.1.92)", "tokenizers (>=0.19,<0.20)", "torch", "tqdm (>=4.27)"]
+torchhub = ["filelock", "huggingface-hub (>=0.23.0,<1.0)", "importlib-metadata", "numpy (>=1.17)", "packaging (>=20.0)", "protobuf", "regex (!=2019.12.17)", "requests", "sentencepiece (>=0.1.91,!=0.1.92)", "tokenizers (>=0.19,<0.20)", "torch", "tqdm (>=4.27)"]
 video = ["av (==9.2.0)", "decord (==0.6.0)"]
 vision = ["Pillow (>=10.0.1,<=15.0)"]
 
diff --git a/runtimes/huggingface/mlserver_huggingface/version.py b/runtimes/huggingface/mlserver_huggingface/version.py
index 8cb37b588..61663da57 100644
--- a/runtimes/huggingface/mlserver_huggingface/version.py
+++ b/runtimes/huggingface/mlserver_huggingface/version.py
@@ -1 +1 @@
-__version__ = "2.0.8"
+__version__ = "2.0.10"
diff --git a/runtimes/huggingface/pyproject.toml b/runtimes/huggingface/pyproject.toml
index c3765c6d3..aaacc0cd9 100644
--- a/runtimes/huggingface/pyproject.toml
+++ b/runtimes/huggingface/pyproject.toml
@@ -1,6 +1,6 @@
 [tool.poetry]
 name = "mlserver-huggingface"
-version = "2.0.8"
+version = "2.0.10"
 description = "HuggingFace runtime for MLServer"
 authors = ["Seldon Technologies Ltd. <hello@seldon.io>"]
 license = "Apache-2.0"

From 381b46efa25fc44e136adf67f1d440313eeb8a0a Mon Sep 17 00:00:00 2001
From: Nanbo Liu <nliu@lambda001.america.striveworks.us>
Date: Fri, 31 Jan 2025 15:57:26 -0600
Subject: [PATCH 16/21] make the output of image classification json
 serializable

---
 .../huggingface/mlserver_huggingface/codecs/chariot.py    | 8 ++++----
 runtimes/huggingface/tests/test_codecs/test_chariot.py    | 2 +-
 2 files changed, 5 insertions(+), 5 deletions(-)

diff --git a/runtimes/huggingface/mlserver_huggingface/codecs/chariot.py b/runtimes/huggingface/mlserver_huggingface/codecs/chariot.py
index 4a19690f9..4664d7b58 100644
--- a/runtimes/huggingface/mlserver_huggingface/codecs/chariot.py
+++ b/runtimes/huggingface/mlserver_huggingface/codecs/chariot.py
@@ -1,5 +1,5 @@
 import numpy as np
-
+import json
 
 def is_list_of_dicts(var):
     """Check if a variable is a list of dicts"""
@@ -66,7 +66,7 @@ def encode_output(
                 #                    1:"tabby, tabby cat",
                 #                    2:"tiger cat"}
                 # convert HF output: [[{"label": "tabby, tabby cat", "score": 0.94},
-                #                      {"label": "tiger cat", "score": 0.04},
+                #                      {"label": "tiger cat", "score"': 0.04},
                 #                      {"label": "Egyptian cat", "score": 0.02}]]
                 # to standard Chariot probability output: [[0.02,0.94,0.04]]
                 # The probability scores are ordered by class id
@@ -83,8 +83,8 @@ def encode_output(
                 # convert HF output: [[{"label": "tabby, tabby cat", "score": 0.94},
                 #                      {"label": "tiger cat", "score": 0.04},
                 #                      {"label": "Egyptian cat", "score": 0.02}]]
-                # to standard Chariot output: ["tabby, tabby cat"]
-                predictions = [p[0]["label"] for p in predictions]
+                # to standard Chariot output: ['"tabby, tabby cat"']
+                predictions = [json.dumps(p[0]["label"]) for p in predictions]
         elif task_type == "object-detection":
 
             # convert HF output: [[{"score": 0.9897010326385498,
diff --git a/runtimes/huggingface/tests/test_codecs/test_chariot.py b/runtimes/huggingface/tests/test_codecs/test_chariot.py
index 601d4676c..6267dc2c6 100644
--- a/runtimes/huggingface/tests/test_codecs/test_chariot.py
+++ b/runtimes/huggingface/tests/test_codecs/test_chariot.py
@@ -18,7 +18,7 @@
                     {"label": "Egyptian cat", "score": 0.02},
                 ]
             ],
-            ["tabby, tabby cat"],
+            ['"tabby, tabby cat"'],
         ),
         (
             "image-classification",

From db4b26be0a25aba3f0f200a83c230280ddb9170c Mon Sep 17 00:00:00 2001
From: Nanbo Liu <nliu@lambda001.america.striveworks.us>
Date: Fri, 31 Jan 2025 16:01:27 -0600
Subject: [PATCH 17/21] black it

---
 runtimes/huggingface/mlserver_huggingface/codecs/chariot.py | 1 +
 1 file changed, 1 insertion(+)

diff --git a/runtimes/huggingface/mlserver_huggingface/codecs/chariot.py b/runtimes/huggingface/mlserver_huggingface/codecs/chariot.py
index 4664d7b58..5df03d616 100644
--- a/runtimes/huggingface/mlserver_huggingface/codecs/chariot.py
+++ b/runtimes/huggingface/mlserver_huggingface/codecs/chariot.py
@@ -1,6 +1,7 @@
 import numpy as np
 import json
 
+
 def is_list_of_dicts(var):
     """Check if a variable is a list of dicts"""
     if not isinstance(var, list):

From d46943f96ecb11274069f759f09b3ea8c2c6c140 Mon Sep 17 00:00:00 2001
From: Nanbo Liu <nliu@striveworks.us>
Date: Mon, 3 Feb 2025 16:59:41 -0600
Subject: [PATCH 18/21] update chariot codec

---
 .../mlserver_huggingface/codecs/chariot.py        | 15 +++------------
 1 file changed, 3 insertions(+), 12 deletions(-)

diff --git a/runtimes/huggingface/mlserver_huggingface/codecs/chariot.py b/runtimes/huggingface/mlserver_huggingface/codecs/chariot.py
index 5df03d616..18412a06c 100644
--- a/runtimes/huggingface/mlserver_huggingface/codecs/chariot.py
+++ b/runtimes/huggingface/mlserver_huggingface/codecs/chariot.py
@@ -1,17 +1,8 @@
 import numpy as np
+from mlserver.codecs.lists import is_list_of
 import json
 
 
-def is_list_of_dicts(var):
-    """Check if a variable is a list of dicts"""
-    if not isinstance(var, list):
-        return False
-    for item in var:
-        if not isinstance(item, dict):
-            return False
-    return True
-
-
 def get_det_dict_from_hf_obj_detect(obj_detect):
     """Convert hf object detection output to standard chariot object detection output"""
     det_dict = {
@@ -58,7 +49,7 @@ class ChariotImgModelOutputCodec:
     def encode_output(
         cls, predictions, task_type, class_int_to_str, predict_proba=False
     ):
-        if is_list_of_dicts(predictions):
+        if is_list_of(predictions, dict):
             predictions = [predictions]
         if task_type == "image-classification":
 
@@ -84,7 +75,7 @@ def encode_output(
                 # convert HF output: [[{"label": "tabby, tabby cat", "score": 0.94},
                 #                      {"label": "tiger cat", "score": 0.04},
                 #                      {"label": "Egyptian cat", "score": 0.02}]]
-                # to standard Chariot output: ['"tabby, tabby cat"']
+                # to standard Chariot output: ["tabby, tabby cat"]
                 predictions = [json.dumps(p[0]["label"]) for p in predictions]
         elif task_type == "object-detection":
 

From 85afb840cad11b8182910cfb7533f2f76a6a3f2b Mon Sep 17 00:00:00 2001
From: Nanbo Liu <nliu@striveworks.us>
Date: Mon, 3 Feb 2025 17:05:29 -0600
Subject: [PATCH 19/21] boost version

---
 runtimes/huggingface/mlserver_huggingface/version.py | 2 +-
 runtimes/huggingface/pyproject.toml                  | 2 +-
 2 files changed, 2 insertions(+), 2 deletions(-)

diff --git a/runtimes/huggingface/mlserver_huggingface/version.py b/runtimes/huggingface/mlserver_huggingface/version.py
index 61663da57..9d85eb132 100644
--- a/runtimes/huggingface/mlserver_huggingface/version.py
+++ b/runtimes/huggingface/mlserver_huggingface/version.py
@@ -1 +1 @@
-__version__ = "2.0.10"
+__version__ = "2.0.11"
diff --git a/runtimes/huggingface/pyproject.toml b/runtimes/huggingface/pyproject.toml
index aaacc0cd9..95813a623 100644
--- a/runtimes/huggingface/pyproject.toml
+++ b/runtimes/huggingface/pyproject.toml
@@ -1,6 +1,6 @@
 [tool.poetry]
 name = "mlserver-huggingface"
-version = "2.0.10"
+version = "2.0.11"
 description = "HuggingFace runtime for MLServer"
 authors = ["Seldon Technologies Ltd. <hello@seldon.io>"]
 license = "Apache-2.0"

From 4aab03a58e6890fefd0fd9954e5c9472c1597ac8 Mon Sep 17 00:00:00 2001
From: Nanbo Liu <nliu@striveworks.us>
Date: Tue, 4 Feb 2025 12:35:43 -0600
Subject: [PATCH 20/21] update comment for chariot model output codec

---
 runtimes/huggingface/mlserver_huggingface/codecs/chariot.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/runtimes/huggingface/mlserver_huggingface/codecs/chariot.py b/runtimes/huggingface/mlserver_huggingface/codecs/chariot.py
index 18412a06c..3bbb1bfc8 100644
--- a/runtimes/huggingface/mlserver_huggingface/codecs/chariot.py
+++ b/runtimes/huggingface/mlserver_huggingface/codecs/chariot.py
@@ -75,7 +75,7 @@ def encode_output(
                 # convert HF output: [[{"label": "tabby, tabby cat", "score": 0.94},
                 #                      {"label": "tiger cat", "score": 0.04},
                 #                      {"label": "Egyptian cat", "score": 0.02}]]
-                # to standard Chariot output: ["tabby, tabby cat"]
+                # to standard Chariot output: ['"tabby, tabby cat"']
                 predictions = [json.dumps(p[0]["label"]) for p in predictions]
         elif task_type == "object-detection":
 

From 4c7316a6eb10aa75b24ab829629d2461e0207660 Mon Sep 17 00:00:00 2001
From: Nanbo Liu <nliu@striveworks.us>
Date: Tue, 4 Feb 2025 12:37:38 -0600
Subject: [PATCH 21/21] revert poetry.lock change

---
 poetry.lock | 205 ++++++++--------------------------------------------
 1 file changed, 29 insertions(+), 176 deletions(-)

diff --git a/poetry.lock b/poetry.lock
index c18524e9d..9f8570f12 100644
--- a/poetry.lock
+++ b/poetry.lock
@@ -11,36 +11,6 @@ files = [
     {file = "absl_py-2.1.0-py3-none-any.whl", hash = "sha256:526a04eadab8b4ee719ce68f204172ead1027549089702d99b9059f129ff1308"},
 ]
 
-[[package]]
-name = "accelerate"
-version = "0.27.2"
-description = "Accelerate"
-optional = false
-python-versions = ">=3.8.0"
-files = [
-    {file = "accelerate-0.27.2-py3-none-any.whl", hash = "sha256:a818dd27b9ba24e9eb5030d1b285cf4cdd1b41bbfa675fb4eb2477ddfc097074"},
-    {file = "accelerate-0.27.2.tar.gz", hash = "sha256:cc715fe9a8bc7a286259bfb6d65fb78363badd3371e7cbda4e4a4ef34a0010aa"},
-]
-
-[package.dependencies]
-huggingface-hub = "*"
-numpy = ">=1.17"
-packaging = ">=20.0"
-psutil = "*"
-pyyaml = "*"
-safetensors = ">=0.3.1"
-torch = ">=1.10.0"
-
-[package.extras]
-dev = ["bitsandbytes", "black (>=23.1,<24.0)", "datasets", "deepspeed (<0.13.0)", "evaluate", "hf-doc-builder (>=0.3.0)", "parameterized", "pytest", "pytest-subtests", "pytest-xdist", "rich", "ruff (>=0.1.15,<0.2.0)", "scikit-learn", "scipy", "timm", "torchpippy (>=0.2.0)", "tqdm", "transformers"]
-quality = ["black (>=23.1,<24.0)", "hf-doc-builder (>=0.3.0)", "ruff (>=0.1.15,<0.2.0)"]
-rich = ["rich"]
-sagemaker = ["sagemaker"]
-test-dev = ["bitsandbytes", "datasets", "deepspeed (<0.13.0)", "evaluate", "scikit-learn", "scipy", "timm", "torchpippy (>=0.2.0)", "tqdm", "transformers"]
-test-prod = ["parameterized", "pytest", "pytest-subtests", "pytest-xdist"]
-test-trackers = ["comet-ml", "dvclive", "tensorboard", "wandb"]
-testing = ["bitsandbytes", "datasets", "deepspeed (<0.13.0)", "evaluate", "parameterized", "pytest", "pytest-subtests", "pytest-xdist", "scikit-learn", "scipy", "timm", "torchpippy (>=0.2.0)", "tqdm", "transformers"]
-
 [[package]]
 name = "aiofiles"
 version = "23.2.1"
@@ -515,20 +485,6 @@ charset-normalizer = ["charset-normalizer"]
 html5lib = ["html5lib"]
 lxml = ["lxml"]
 
-[[package]]
-name = "bitsandbytes"
-version = "0.42.0"
-description = "k-bit optimizers and matrix multiplication routines."
-optional = false
-python-versions = "*"
-files = [
-    {file = "bitsandbytes-0.42.0-py3-none-any.whl", hash = "sha256:63798680912cc63bb77b535a2d0860af024e290a52e157f777ad2a52e2585967"},
-    {file = "bitsandbytes-0.42.0.tar.gz", hash = "sha256:fc1505f184f0d275766f2a6c663f1a43b734c1409b5c5a406f3a6073d9f329fd"},
-]
-
-[package.dependencies]
-scipy = "*"
-
 [[package]]
 name = "black"
 version = "24.4.0"
@@ -649,10 +605,6 @@ files = [
     {file = "Brotli-1.1.0-cp310-cp310-musllinux_1_1_i686.whl", hash = "sha256:a37b8f0391212d29b3a91a799c8e4a2855e0576911cdfb2515487e30e322253d"},
     {file = "Brotli-1.1.0-cp310-cp310-musllinux_1_1_ppc64le.whl", hash = "sha256:e84799f09591700a4154154cab9787452925578841a94321d5ee8fb9a9a328f0"},
     {file = "Brotli-1.1.0-cp310-cp310-musllinux_1_1_x86_64.whl", hash = "sha256:f66b5337fa213f1da0d9000bc8dc0cb5b896b726eefd9c6046f699b169c41b9e"},
-    {file = "Brotli-1.1.0-cp310-cp310-musllinux_1_2_aarch64.whl", hash = "sha256:5dab0844f2cf82be357a0eb11a9087f70c5430b2c241493fc122bb6f2bb0917c"},
-    {file = "Brotli-1.1.0-cp310-cp310-musllinux_1_2_i686.whl", hash = "sha256:e4fe605b917c70283db7dfe5ada75e04561479075761a0b3866c081d035b01c1"},
-    {file = "Brotli-1.1.0-cp310-cp310-musllinux_1_2_ppc64le.whl", hash = "sha256:1e9a65b5736232e7a7f91ff3d02277f11d339bf34099a56cdab6a8b3410a02b2"},
-    {file = "Brotli-1.1.0-cp310-cp310-musllinux_1_2_x86_64.whl", hash = "sha256:58d4b711689366d4a03ac7957ab8c28890415e267f9b6589969e74b6e42225ec"},
     {file = "Brotli-1.1.0-cp310-cp310-win32.whl", hash = "sha256:be36e3d172dc816333f33520154d708a2657ea63762ec16b62ece02ab5e4daf2"},
     {file = "Brotli-1.1.0-cp310-cp310-win_amd64.whl", hash = "sha256:0c6244521dda65ea562d5a69b9a26120769b7a9fb3db2fe9545935ed6735b128"},
     {file = "Brotli-1.1.0-cp311-cp311-macosx_10_9_universal2.whl", hash = "sha256:a3daabb76a78f829cafc365531c972016e4aa8d5b4bf60660ad8ecee19df7ccc"},
@@ -665,14 +617,8 @@ files = [
     {file = "Brotli-1.1.0-cp311-cp311-musllinux_1_1_i686.whl", hash = "sha256:19c116e796420b0cee3da1ccec3b764ed2952ccfcc298b55a10e5610ad7885f9"},
     {file = "Brotli-1.1.0-cp311-cp311-musllinux_1_1_ppc64le.whl", hash = "sha256:510b5b1bfbe20e1a7b3baf5fed9e9451873559a976c1a78eebaa3b86c57b4265"},
     {file = "Brotli-1.1.0-cp311-cp311-musllinux_1_1_x86_64.whl", hash = "sha256:a1fd8a29719ccce974d523580987b7f8229aeace506952fa9ce1d53a033873c8"},
-    {file = "Brotli-1.1.0-cp311-cp311-musllinux_1_2_aarch64.whl", hash = "sha256:c247dd99d39e0338a604f8c2b3bc7061d5c2e9e2ac7ba9cc1be5a69cb6cd832f"},
-    {file = "Brotli-1.1.0-cp311-cp311-musllinux_1_2_i686.whl", hash = "sha256:1b2c248cd517c222d89e74669a4adfa5577e06ab68771a529060cf5a156e9757"},
-    {file = "Brotli-1.1.0-cp311-cp311-musllinux_1_2_ppc64le.whl", hash = "sha256:2a24c50840d89ded6c9a8fdc7b6ed3692ed4e86f1c4a4a938e1e92def92933e0"},
-    {file = "Brotli-1.1.0-cp311-cp311-musllinux_1_2_x86_64.whl", hash = "sha256:f31859074d57b4639318523d6ffdca586ace54271a73ad23ad021acd807eb14b"},
     {file = "Brotli-1.1.0-cp311-cp311-win32.whl", hash = "sha256:39da8adedf6942d76dc3e46653e52df937a3c4d6d18fdc94a7c29d263b1f5b50"},
     {file = "Brotli-1.1.0-cp311-cp311-win_amd64.whl", hash = "sha256:aac0411d20e345dc0920bdec5548e438e999ff68d77564d5e9463a7ca9d3e7b1"},
-    {file = "Brotli-1.1.0-cp312-cp312-macosx_10_13_universal2.whl", hash = "sha256:32d95b80260d79926f5fab3c41701dbb818fde1c9da590e77e571eefd14abe28"},
-    {file = "Brotli-1.1.0-cp312-cp312-macosx_10_13_x86_64.whl", hash = "sha256:b760c65308ff1e462f65d69c12e4ae085cff3b332d894637f6273a12a482d09f"},
     {file = "Brotli-1.1.0-cp312-cp312-macosx_10_9_universal2.whl", hash = "sha256:316cc9b17edf613ac76b1f1f305d2a748f1b976b033b049a6ecdfd5612c70409"},
     {file = "Brotli-1.1.0-cp312-cp312-macosx_10_9_x86_64.whl", hash = "sha256:caf9ee9a5775f3111642d33b86237b05808dafcd6268faa492250e9b78046eb2"},
     {file = "Brotli-1.1.0-cp312-cp312-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:70051525001750221daa10907c77830bc889cb6d865cc0b813d9db7fefc21451"},
@@ -683,24 +629,8 @@ files = [
     {file = "Brotli-1.1.0-cp312-cp312-musllinux_1_1_i686.whl", hash = "sha256:4093c631e96fdd49e0377a9c167bfd75b6d0bad2ace734c6eb20b348bc3ea180"},
     {file = "Brotli-1.1.0-cp312-cp312-musllinux_1_1_ppc64le.whl", hash = "sha256:7e4c4629ddad63006efa0ef968c8e4751c5868ff0b1c5c40f76524e894c50248"},
     {file = "Brotli-1.1.0-cp312-cp312-musllinux_1_1_x86_64.whl", hash = "sha256:861bf317735688269936f755fa136a99d1ed526883859f86e41a5d43c61d8966"},
-    {file = "Brotli-1.1.0-cp312-cp312-musllinux_1_2_aarch64.whl", hash = "sha256:87a3044c3a35055527ac75e419dfa9f4f3667a1e887ee80360589eb8c90aabb9"},
-    {file = "Brotli-1.1.0-cp312-cp312-musllinux_1_2_i686.whl", hash = "sha256:c5529b34c1c9d937168297f2c1fde7ebe9ebdd5e121297ff9c043bdb2ae3d6fb"},
-    {file = "Brotli-1.1.0-cp312-cp312-musllinux_1_2_ppc64le.whl", hash = "sha256:ca63e1890ede90b2e4454f9a65135a4d387a4585ff8282bb72964fab893f2111"},
-    {file = "Brotli-1.1.0-cp312-cp312-musllinux_1_2_x86_64.whl", hash = "sha256:e79e6520141d792237c70bcd7a3b122d00f2613769ae0cb61c52e89fd3443839"},
     {file = "Brotli-1.1.0-cp312-cp312-win32.whl", hash = "sha256:5f4d5ea15c9382135076d2fb28dde923352fe02951e66935a9efaac8f10e81b0"},
     {file = "Brotli-1.1.0-cp312-cp312-win_amd64.whl", hash = "sha256:906bc3a79de8c4ae5b86d3d75a8b77e44404b0f4261714306e3ad248d8ab0951"},
-    {file = "Brotli-1.1.0-cp313-cp313-macosx_10_13_universal2.whl", hash = "sha256:8bf32b98b75c13ec7cf774164172683d6e7891088f6316e54425fde1efc276d5"},
-    {file = "Brotli-1.1.0-cp313-cp313-macosx_10_13_x86_64.whl", hash = "sha256:7bc37c4d6b87fb1017ea28c9508b36bbcb0c3d18b4260fcdf08b200c74a6aee8"},
-    {file = "Brotli-1.1.0-cp313-cp313-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:3c0ef38c7a7014ffac184db9e04debe495d317cc9c6fb10071f7fefd93100a4f"},
-    {file = "Brotli-1.1.0-cp313-cp313-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:91d7cc2a76b5567591d12c01f019dd7afce6ba8cba6571187e21e2fc418ae648"},
-    {file = "Brotli-1.1.0-cp313-cp313-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:a93dde851926f4f2678e704fadeb39e16c35d8baebd5252c9fd94ce8ce68c4a0"},
-    {file = "Brotli-1.1.0-cp313-cp313-manylinux_2_5_i686.manylinux1_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:f0db75f47be8b8abc8d9e31bc7aad0547ca26f24a54e6fd10231d623f183d089"},
-    {file = "Brotli-1.1.0-cp313-cp313-musllinux_1_2_aarch64.whl", hash = "sha256:6967ced6730aed543b8673008b5a391c3b1076d834ca438bbd70635c73775368"},
-    {file = "Brotli-1.1.0-cp313-cp313-musllinux_1_2_i686.whl", hash = "sha256:7eedaa5d036d9336c95915035fb57422054014ebdeb6f3b42eac809928e40d0c"},
-    {file = "Brotli-1.1.0-cp313-cp313-musllinux_1_2_ppc64le.whl", hash = "sha256:d487f5432bf35b60ed625d7e1b448e2dc855422e87469e3f450aa5552b0eb284"},
-    {file = "Brotli-1.1.0-cp313-cp313-musllinux_1_2_x86_64.whl", hash = "sha256:832436e59afb93e1836081a20f324cb185836c617659b07b129141a8426973c7"},
-    {file = "Brotli-1.1.0-cp313-cp313-win32.whl", hash = "sha256:43395e90523f9c23a3d5bdf004733246fba087f2948f87ab28015f12359ca6a0"},
-    {file = "Brotli-1.1.0-cp313-cp313-win_amd64.whl", hash = "sha256:9011560a466d2eb3f5a6e4929cf4a09be405c64154e12df0dd72713f6500e32b"},
     {file = "Brotli-1.1.0-cp36-cp36m-macosx_10_9_x86_64.whl", hash = "sha256:a090ca607cbb6a34b0391776f0cb48062081f5f60ddcce5d11838e67a01928d1"},
     {file = "Brotli-1.1.0-cp36-cp36m-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:2de9d02f5bda03d27ede52e8cfe7b865b066fa49258cbab568720aa5be80a47d"},
     {file = "Brotli-1.1.0-cp36-cp36m-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:2333e30a5e00fe0fe55903c8832e08ee9c3b1382aacf4db26664a16528d51b4b"},
@@ -710,10 +640,6 @@ files = [
     {file = "Brotli-1.1.0-cp36-cp36m-musllinux_1_1_i686.whl", hash = "sha256:fd5f17ff8f14003595ab414e45fce13d073e0762394f957182e69035c9f3d7c2"},
     {file = "Brotli-1.1.0-cp36-cp36m-musllinux_1_1_ppc64le.whl", hash = "sha256:069a121ac97412d1fe506da790b3e69f52254b9df4eb665cd42460c837193354"},
     {file = "Brotli-1.1.0-cp36-cp36m-musllinux_1_1_x86_64.whl", hash = "sha256:e93dfc1a1165e385cc8239fab7c036fb2cd8093728cbd85097b284d7b99249a2"},
-    {file = "Brotli-1.1.0-cp36-cp36m-musllinux_1_2_aarch64.whl", hash = "sha256:aea440a510e14e818e67bfc4027880e2fb500c2ccb20ab21c7a7c8b5b4703d75"},
-    {file = "Brotli-1.1.0-cp36-cp36m-musllinux_1_2_i686.whl", hash = "sha256:6974f52a02321b36847cd19d1b8e381bf39939c21efd6ee2fc13a28b0d99348c"},
-    {file = "Brotli-1.1.0-cp36-cp36m-musllinux_1_2_ppc64le.whl", hash = "sha256:a7e53012d2853a07a4a79c00643832161a910674a893d296c9f1259859a289d2"},
-    {file = "Brotli-1.1.0-cp36-cp36m-musllinux_1_2_x86_64.whl", hash = "sha256:d7702622a8b40c49bffb46e1e3ba2e81268d5c04a34f460978c6b5517a34dd52"},
     {file = "Brotli-1.1.0-cp36-cp36m-win32.whl", hash = "sha256:a599669fd7c47233438a56936988a2478685e74854088ef5293802123b5b2460"},
     {file = "Brotli-1.1.0-cp36-cp36m-win_amd64.whl", hash = "sha256:d143fd47fad1db3d7c27a1b1d66162e855b5d50a89666af46e1679c496e8e579"},
     {file = "Brotli-1.1.0-cp37-cp37m-macosx_10_9_x86_64.whl", hash = "sha256:11d00ed0a83fa22d29bc6b64ef636c4552ebafcef57154b4ddd132f5638fbd1c"},
@@ -725,10 +651,6 @@ files = [
     {file = "Brotli-1.1.0-cp37-cp37m-musllinux_1_1_i686.whl", hash = "sha256:919e32f147ae93a09fe064d77d5ebf4e35502a8df75c29fb05788528e330fe74"},
     {file = "Brotli-1.1.0-cp37-cp37m-musllinux_1_1_ppc64le.whl", hash = "sha256:23032ae55523cc7bccb4f6a0bf368cd25ad9bcdcc1990b64a647e7bbcce9cb5b"},
     {file = "Brotli-1.1.0-cp37-cp37m-musllinux_1_1_x86_64.whl", hash = "sha256:224e57f6eac61cc449f498cc5f0e1725ba2071a3d4f48d5d9dffba42db196438"},
-    {file = "Brotli-1.1.0-cp37-cp37m-musllinux_1_2_aarch64.whl", hash = "sha256:cb1dac1770878ade83f2ccdf7d25e494f05c9165f5246b46a621cc849341dc01"},
-    {file = "Brotli-1.1.0-cp37-cp37m-musllinux_1_2_i686.whl", hash = "sha256:3ee8a80d67a4334482d9712b8e83ca6b1d9bc7e351931252ebef5d8f7335a547"},
-    {file = "Brotli-1.1.0-cp37-cp37m-musllinux_1_2_ppc64le.whl", hash = "sha256:5e55da2c8724191e5b557f8e18943b1b4839b8efc3ef60d65985bcf6f587dd38"},
-    {file = "Brotli-1.1.0-cp37-cp37m-musllinux_1_2_x86_64.whl", hash = "sha256:d342778ef319e1026af243ed0a07c97acf3bad33b9f29e7ae6a1f68fd083e90c"},
     {file = "Brotli-1.1.0-cp37-cp37m-win32.whl", hash = "sha256:587ca6d3cef6e4e868102672d3bd9dc9698c309ba56d41c2b9c85bbb903cdb95"},
     {file = "Brotli-1.1.0-cp37-cp37m-win_amd64.whl", hash = "sha256:2954c1c23f81c2eaf0b0717d9380bd348578a94161a65b3a2afc62c86467dd68"},
     {file = "Brotli-1.1.0-cp38-cp38-macosx_10_9_universal2.whl", hash = "sha256:efa8b278894b14d6da122a72fefcebc28445f2d3f880ac59d46c90f4c13be9a3"},
@@ -741,10 +663,6 @@ files = [
     {file = "Brotli-1.1.0-cp38-cp38-musllinux_1_1_i686.whl", hash = "sha256:1ab4fbee0b2d9098c74f3057b2bc055a8bd92ccf02f65944a241b4349229185a"},
     {file = "Brotli-1.1.0-cp38-cp38-musllinux_1_1_ppc64le.whl", hash = "sha256:141bd4d93984070e097521ed07e2575b46f817d08f9fa42b16b9b5f27b5ac088"},
     {file = "Brotli-1.1.0-cp38-cp38-musllinux_1_1_x86_64.whl", hash = "sha256:fce1473f3ccc4187f75b4690cfc922628aed4d3dd013d047f95a9b3919a86596"},
-    {file = "Brotli-1.1.0-cp38-cp38-musllinux_1_2_aarch64.whl", hash = "sha256:d2b35ca2c7f81d173d2fadc2f4f31e88cc5f7a39ae5b6db5513cf3383b0e0ec7"},
-    {file = "Brotli-1.1.0-cp38-cp38-musllinux_1_2_i686.whl", hash = "sha256:af6fa6817889314555aede9a919612b23739395ce767fe7fcbea9a80bf140fe5"},
-    {file = "Brotli-1.1.0-cp38-cp38-musllinux_1_2_ppc64le.whl", hash = "sha256:2feb1d960f760a575dbc5ab3b1c00504b24caaf6986e2dc2b01c09c87866a943"},
-    {file = "Brotli-1.1.0-cp38-cp38-musllinux_1_2_x86_64.whl", hash = "sha256:4410f84b33374409552ac9b6903507cdb31cd30d2501fc5ca13d18f73548444a"},
     {file = "Brotli-1.1.0-cp38-cp38-win32.whl", hash = "sha256:db85ecf4e609a48f4b29055f1e144231b90edc90af7481aa731ba2d059226b1b"},
     {file = "Brotli-1.1.0-cp38-cp38-win_amd64.whl", hash = "sha256:3d7954194c36e304e1523f55d7042c59dc53ec20dd4e9ea9d151f1b62b4415c0"},
     {file = "Brotli-1.1.0-cp39-cp39-macosx_10_9_universal2.whl", hash = "sha256:5fb2ce4b8045c78ebbc7b8f3c15062e435d47e7393cc57c25115cfd49883747a"},
@@ -757,10 +675,6 @@ files = [
     {file = "Brotli-1.1.0-cp39-cp39-musllinux_1_1_i686.whl", hash = "sha256:949f3b7c29912693cee0afcf09acd6ebc04c57af949d9bf77d6101ebb61e388c"},
     {file = "Brotli-1.1.0-cp39-cp39-musllinux_1_1_ppc64le.whl", hash = "sha256:89f4988c7203739d48c6f806f1e87a1d96e0806d44f0fba61dba81392c9e474d"},
     {file = "Brotli-1.1.0-cp39-cp39-musllinux_1_1_x86_64.whl", hash = "sha256:de6551e370ef19f8de1807d0a9aa2cdfdce2e85ce88b122fe9f6b2b076837e59"},
-    {file = "Brotli-1.1.0-cp39-cp39-musllinux_1_2_aarch64.whl", hash = "sha256:0737ddb3068957cf1b054899b0883830bb1fec522ec76b1098f9b6e0f02d9419"},
-    {file = "Brotli-1.1.0-cp39-cp39-musllinux_1_2_i686.whl", hash = "sha256:4f3607b129417e111e30637af1b56f24f7a49e64763253bbc275c75fa887d4b2"},
-    {file = "Brotli-1.1.0-cp39-cp39-musllinux_1_2_ppc64le.whl", hash = "sha256:6c6e0c425f22c1c719c42670d561ad682f7bfeeef918edea971a79ac5252437f"},
-    {file = "Brotli-1.1.0-cp39-cp39-musllinux_1_2_x86_64.whl", hash = "sha256:494994f807ba0b92092a163a0a283961369a65f6cbe01e8891132b7a320e61eb"},
     {file = "Brotli-1.1.0-cp39-cp39-win32.whl", hash = "sha256:f0d8a7a6b5983c2496e364b969f0e526647a06b075d034f3297dc66f3b360c64"},
     {file = "Brotli-1.1.0-cp39-cp39-win_amd64.whl", hash = "sha256:cdad5b9014d83ca68c25d2e9444e28e967ef16e80f6b436918c700c117a85467"},
     {file = "Brotli-1.1.0.tar.gz", hash = "sha256:81de08ac11bcb85841e440c13611c00b67d3bf82698314928d0b676362546724"},
@@ -3709,7 +3623,7 @@ url = "runtimes/catboost"
 
 [[package]]
 name = "mlserver-huggingface"
-version = "2.0.10"
+version = "1.6.0.dev2"
 description = "HuggingFace runtime for MLServer"
 optional = false
 python-versions = ">=3.9,<3.12"
@@ -3717,16 +3631,11 @@ files = []
 develop = true
 
 [package.dependencies]
-accelerate = "^0.27.2"
-bitsandbytes = "^0.42.0"
 mlserver = "*"
 optimum = {version = ">=1.4,<2.0", extras = ["onnxruntime"]}
 pillow = "*"
 pydantic = "2.7.1"
-sentence-transformers = "2.5.1"
 tensorflow = "*"
-timm = "^1.0.13"
-transformers = "4.41.2"
 
 [package.source]
 type = "directory"
@@ -5105,36 +5014,6 @@ files = [
     {file = "protobuf-3.20.3.tar.gz", hash = "sha256:2e3427429c9cffebf259491be0af70189607f365c2f41c7c3764af6f337105f2"},
 ]
 
-[[package]]
-name = "psutil"
-version = "6.1.1"
-description = "Cross-platform lib for process and system monitoring in Python."
-optional = false
-python-versions = "!=3.0.*,!=3.1.*,!=3.2.*,!=3.3.*,!=3.4.*,!=3.5.*,>=2.7"
-files = [
-    {file = "psutil-6.1.1-cp27-cp27m-macosx_10_9_x86_64.whl", hash = "sha256:9ccc4316f24409159897799b83004cb1e24f9819b0dcf9c0b68bdcb6cefee6a8"},
-    {file = "psutil-6.1.1-cp27-cp27m-manylinux2010_i686.whl", hash = "sha256:ca9609c77ea3b8481ab005da74ed894035936223422dc591d6772b147421f777"},
-    {file = "psutil-6.1.1-cp27-cp27m-manylinux2010_x86_64.whl", hash = "sha256:8df0178ba8a9e5bc84fed9cfa61d54601b371fbec5c8eebad27575f1e105c0d4"},
-    {file = "psutil-6.1.1-cp27-cp27mu-manylinux2010_i686.whl", hash = "sha256:1924e659d6c19c647e763e78670a05dbb7feaf44a0e9c94bf9e14dfc6ba50468"},
-    {file = "psutil-6.1.1-cp27-cp27mu-manylinux2010_x86_64.whl", hash = "sha256:018aeae2af92d943fdf1da6b58665124897cfc94faa2ca92098838f83e1b1bca"},
-    {file = "psutil-6.1.1-cp27-none-win32.whl", hash = "sha256:6d4281f5bbca041e2292be3380ec56a9413b790579b8e593b1784499d0005dac"},
-    {file = "psutil-6.1.1-cp27-none-win_amd64.whl", hash = "sha256:c777eb75bb33c47377c9af68f30e9f11bc78e0f07fbf907be4a5d70b2fe5f030"},
-    {file = "psutil-6.1.1-cp36-abi3-macosx_10_9_x86_64.whl", hash = "sha256:fc0ed7fe2231a444fc219b9c42d0376e0a9a1a72f16c5cfa0f68d19f1a0663e8"},
-    {file = "psutil-6.1.1-cp36-abi3-macosx_11_0_arm64.whl", hash = "sha256:0bdd4eab935276290ad3cb718e9809412895ca6b5b334f5a9111ee6d9aff9377"},
-    {file = "psutil-6.1.1-cp36-abi3-manylinux_2_12_i686.manylinux2010_i686.manylinux_2_17_i686.manylinux2014_i686.whl", hash = "sha256:b6e06c20c05fe95a3d7302d74e7097756d4ba1247975ad6905441ae1b5b66003"},
-    {file = "psutil-6.1.1-cp36-abi3-manylinux_2_12_x86_64.manylinux2010_x86_64.manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:97f7cb9921fbec4904f522d972f0c0e1f4fabbdd4e0287813b21215074a0f160"},
-    {file = "psutil-6.1.1-cp36-abi3-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:33431e84fee02bc84ea36d9e2c4a6d395d479c9dd9bba2376c1f6ee8f3a4e0b3"},
-    {file = "psutil-6.1.1-cp36-cp36m-win32.whl", hash = "sha256:384636b1a64b47814437d1173be1427a7c83681b17a450bfc309a1953e329603"},
-    {file = "psutil-6.1.1-cp36-cp36m-win_amd64.whl", hash = "sha256:8be07491f6ebe1a693f17d4f11e69d0dc1811fa082736500f649f79df7735303"},
-    {file = "psutil-6.1.1-cp37-abi3-win32.whl", hash = "sha256:eaa912e0b11848c4d9279a93d7e2783df352b082f40111e078388701fd479e53"},
-    {file = "psutil-6.1.1-cp37-abi3-win_amd64.whl", hash = "sha256:f35cfccb065fff93529d2afb4a2e89e363fe63ca1e4a5da22b603a85833c2649"},
-    {file = "psutil-6.1.1.tar.gz", hash = "sha256:cf8496728c18f2d0b45198f06895be52f36611711746b7f30c464b422b50e2f5"},
-]
-
-[package.extras]
-dev = ["abi3audit", "black", "check-manifest", "coverage", "packaging", "pylint", "pyperf", "pypinfo", "pytest-cov", "requests", "rstcheck", "ruff", "sphinx", "sphinx_rtd_theme", "toml-sort", "twine", "virtualenv", "vulture", "wheel"]
-test = ["pytest", "pytest-xdist", "setuptools"]
-
 [[package]]
 name = "py-grpc-prometheus"
 version = "0.8.0"
@@ -5805,6 +5684,7 @@ files = [
     {file = "PyYAML-6.0.1-cp310-cp310-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:69b023b2b4daa7548bcfbd4aa3da05b3a74b772db9e23b982788168117739938"},
     {file = "PyYAML-6.0.1-cp310-cp310-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:81e0b275a9ecc9c0c0c07b4b90ba548307583c125f54d5b6946cfee6360c733d"},
     {file = "PyYAML-6.0.1-cp310-cp310-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:ba336e390cd8e4d1739f42dfe9bb83a3cc2e80f567d8805e11b46f4a943f5515"},
+    {file = "PyYAML-6.0.1-cp310-cp310-musllinux_1_1_x86_64.whl", hash = "sha256:326c013efe8048858a6d312ddd31d56e468118ad4cdeda36c719bf5bb6192290"},
     {file = "PyYAML-6.0.1-cp310-cp310-win32.whl", hash = "sha256:bd4af7373a854424dabd882decdc5579653d7868b8fb26dc7d0e99f823aa5924"},
     {file = "PyYAML-6.0.1-cp310-cp310-win_amd64.whl", hash = "sha256:fd1592b3fdf65fff2ad0004b5e363300ef59ced41c2e6b3a99d4089fa8c5435d"},
     {file = "PyYAML-6.0.1-cp311-cp311-macosx_10_9_x86_64.whl", hash = "sha256:6965a7bc3cf88e5a1c3bd2e0b5c22f8d677dc88a455344035f03399034eb3007"},
@@ -5812,8 +5692,16 @@ files = [
     {file = "PyYAML-6.0.1-cp311-cp311-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:42f8152b8dbc4fe7d96729ec2b99c7097d656dc1213a3229ca5383f973a5ed6d"},
     {file = "PyYAML-6.0.1-cp311-cp311-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:062582fca9fabdd2c8b54a3ef1c978d786e0f6b3a1510e0ac93ef59e0ddae2bc"},
     {file = "PyYAML-6.0.1-cp311-cp311-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:d2b04aac4d386b172d5b9692e2d2da8de7bfb6c387fa4f801fbf6fb2e6ba4673"},
+    {file = "PyYAML-6.0.1-cp311-cp311-musllinux_1_1_x86_64.whl", hash = "sha256:e7d73685e87afe9f3b36c799222440d6cf362062f78be1013661b00c5c6f678b"},
     {file = "PyYAML-6.0.1-cp311-cp311-win32.whl", hash = "sha256:1635fd110e8d85d55237ab316b5b011de701ea0f29d07611174a1b42f1444741"},
     {file = "PyYAML-6.0.1-cp311-cp311-win_amd64.whl", hash = "sha256:bf07ee2fef7014951eeb99f56f39c9bb4af143d8aa3c21b1677805985307da34"},
+    {file = "PyYAML-6.0.1-cp312-cp312-macosx_10_9_x86_64.whl", hash = "sha256:855fb52b0dc35af121542a76b9a84f8d1cd886ea97c84703eaa6d88e37a2ad28"},
+    {file = "PyYAML-6.0.1-cp312-cp312-macosx_11_0_arm64.whl", hash = "sha256:40df9b996c2b73138957fe23a16a4f0ba614f4c0efce1e9406a184b6d07fa3a9"},
+    {file = "PyYAML-6.0.1-cp312-cp312-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:a08c6f0fe150303c1c6b71ebcd7213c2858041a7e01975da3a99aed1e7a378ef"},
+    {file = "PyYAML-6.0.1-cp312-cp312-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:6c22bec3fbe2524cde73d7ada88f6566758a8f7227bfbf93a408a9d86bcc12a0"},
+    {file = "PyYAML-6.0.1-cp312-cp312-musllinux_1_1_x86_64.whl", hash = "sha256:8d4e9c88387b0f5c7d5f281e55304de64cf7f9c0021a3525bd3b1c542da3b0e4"},
+    {file = "PyYAML-6.0.1-cp312-cp312-win32.whl", hash = "sha256:d483d2cdf104e7c9fa60c544d92981f12ad66a457afae824d146093b8c294c54"},
+    {file = "PyYAML-6.0.1-cp312-cp312-win_amd64.whl", hash = "sha256:0d3304d8c0adc42be59c5f8a4d9e3d7379e6955ad754aa9d6ab7a398b59dd1df"},
     {file = "PyYAML-6.0.1-cp36-cp36m-macosx_10_9_x86_64.whl", hash = "sha256:50550eb667afee136e9a77d6dc71ae76a44df8b3e51e41b77f6de2932bfe0f47"},
     {file = "PyYAML-6.0.1-cp36-cp36m-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:1fe35611261b29bd1de0070f0b2f47cb6ff71fa6595c077e42bd0c419fa27b98"},
     {file = "PyYAML-6.0.1-cp36-cp36m-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:704219a11b772aea0d8ecd7058d0082713c3562b4e271b849ad7dc4a5c90c13c"},
@@ -5830,6 +5718,7 @@ files = [
     {file = "PyYAML-6.0.1-cp38-cp38-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:a0cd17c15d3bb3fa06978b4e8958dcdc6e0174ccea823003a106c7d4d7899ac5"},
     {file = "PyYAML-6.0.1-cp38-cp38-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:28c119d996beec18c05208a8bd78cbe4007878c6dd15091efb73a30e90539696"},
     {file = "PyYAML-6.0.1-cp38-cp38-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:7e07cbde391ba96ab58e532ff4803f79c4129397514e1413a7dc761ccd755735"},
+    {file = "PyYAML-6.0.1-cp38-cp38-musllinux_1_1_x86_64.whl", hash = "sha256:49a183be227561de579b4a36efbb21b3eab9651dd81b1858589f796549873dd6"},
     {file = "PyYAML-6.0.1-cp38-cp38-win32.whl", hash = "sha256:184c5108a2aca3c5b3d3bf9395d50893a7ab82a38004c8f61c258d4428e80206"},
     {file = "PyYAML-6.0.1-cp38-cp38-win_amd64.whl", hash = "sha256:1e2722cc9fbb45d9b87631ac70924c11d3a401b2d7f410cc0e3bbf249f2dca62"},
     {file = "PyYAML-6.0.1-cp39-cp39-macosx_10_9_x86_64.whl", hash = "sha256:9eb6caa9a297fc2c2fb8862bc5370d0303ddba53ba97e71f08023b6cd73d16a8"},
@@ -5837,6 +5726,7 @@ files = [
     {file = "PyYAML-6.0.1-cp39-cp39-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:5773183b6446b2c99bb77e77595dd486303b4faab2b086e7b17bc6bef28865f6"},
     {file = "PyYAML-6.0.1-cp39-cp39-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:b786eecbdf8499b9ca1d697215862083bd6d2a99965554781d0d8d1ad31e13a0"},
     {file = "PyYAML-6.0.1-cp39-cp39-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:bc1bf2925a1ecd43da378f4db9e4f799775d6367bdb94671027b73b393a7c42c"},
+    {file = "PyYAML-6.0.1-cp39-cp39-musllinux_1_1_x86_64.whl", hash = "sha256:04ac92ad1925b2cff1db0cfebffb6ffc43457495c9b3c39d3fcae417d7125dc5"},
     {file = "PyYAML-6.0.1-cp39-cp39-win32.whl", hash = "sha256:faca3bdcf85b2fc05d06ff3fbc1f83e1391b3e724afa3feba7d13eeab355484c"},
     {file = "PyYAML-6.0.1-cp39-cp39-win_amd64.whl", hash = "sha256:510c9deebc5c0225e8c96813043e62b680ba2f9c50a08d3724c7f28a747d1486"},
     {file = "PyYAML-6.0.1.tar.gz", hash = "sha256:bfdf460b1736c775f2ba9f6a92bca30bc2095067b8a9d77876d1fad6cc3b4a43"},
@@ -6320,27 +6210,6 @@ dev = ["cython-lint (>=0.12.2)", "doit (>=0.36.0)", "mypy", "pycodestyle", "pyde
 doc = ["jupyterlite-pyodide-kernel", "jupyterlite-sphinx (>=0.12.0)", "jupytext", "matplotlib (>=3.5)", "myst-nb", "numpydoc", "pooch", "pydata-sphinx-theme (>=0.15.2)", "sphinx (>=5.0.0)", "sphinx-design (>=0.4.0)"]
 test = ["array-api-strict", "asv", "gmpy2", "hypothesis (>=6.30)", "mpmath", "pooch", "pytest", "pytest-cov", "pytest-timeout", "pytest-xdist", "scikit-umfpack", "threadpoolctl"]
 
-[[package]]
-name = "sentence-transformers"
-version = "2.5.1"
-description = "Multilingual text embeddings"
-optional = false
-python-versions = ">=3.8.0"
-files = [
-    {file = "sentence-transformers-2.5.1.tar.gz", hash = "sha256:754bf2b2623eb46904fd9c72ff89a0f90200fe141a8d45b03e83bc6d51718153"},
-    {file = "sentence_transformers-2.5.1-py3-none-any.whl", hash = "sha256:f12346f7fca06ed1198d24235cb9114a74665506f7c30044e0a6f12de7eeeb77"},
-]
-
-[package.dependencies]
-huggingface-hub = ">=0.15.1"
-numpy = "*"
-Pillow = "*"
-scikit-learn = "*"
-scipy = "*"
-torch = ">=1.11.0"
-tqdm = "*"
-transformers = ">=4.32.0,<5.0.0"
-
 [[package]]
 name = "sentencepiece"
 version = "0.2.0"
@@ -7439,24 +7308,6 @@ numpy = "*"
 [package.extras]
 all = ["defusedxml", "fsspec", "imagecodecs (>=2023.8.12)", "lxml", "matplotlib", "zarr"]
 
-[[package]]
-name = "timm"
-version = "1.0.14"
-description = "PyTorch Image Models"
-optional = false
-python-versions = ">=3.8"
-files = [
-    {file = "timm-1.0.14-py3-none-any.whl", hash = "sha256:16653695ff420cf4e87e384c8654f2ee6be2070bcb4a0e840b5c6764ee0547ec"},
-    {file = "timm-1.0.14.tar.gz", hash = "sha256:00a7f2cc04ce3ed8f80476bbb7eea27eac8cf6f2d59b5e9aa9cdd375dd6550db"},
-]
-
-[package.dependencies]
-huggingface_hub = "*"
-pyyaml = "*"
-safetensors = "*"
-torch = "*"
-torchvision = "*"
-
 [[package]]
 name = "tokenizers"
 version = "0.19.1"
@@ -7774,19 +7625,19 @@ telegram = ["requests"]
 
 [[package]]
 name = "transformers"
-version = "4.41.2"
+version = "4.42.4"
 description = "State-of-the-art Machine Learning for JAX, PyTorch and TensorFlow"
 optional = false
 python-versions = ">=3.8.0"
 files = [
-    {file = "transformers-4.41.2-py3-none-any.whl", hash = "sha256:05555d20e43f808de1ef211ab64803cdb513170cef70d29a888b589caebefc67"},
-    {file = "transformers-4.41.2.tar.gz", hash = "sha256:80a4db216533d573e9cc7388646c31ed9480918feb7c55eb211249cb23567f87"},
+    {file = "transformers-4.42.4-py3-none-any.whl", hash = "sha256:6d59061392d0f1da312af29c962df9017ff3c0108c681a56d1bc981004d16d24"},
+    {file = "transformers-4.42.4.tar.gz", hash = "sha256:f956e25e24df851f650cb2c158b6f4352dfae9d702f04c113ed24fc36ce7ae2d"},
 ]
 
 [package.dependencies]
 filelock = "*"
-huggingface-hub = ">=0.23.0,<1.0"
-numpy = ">=1.17"
+huggingface-hub = ">=0.23.2,<1.0"
+numpy = ">=1.17,<2.0"
 packaging = ">=20.0"
 protobuf = {version = "*", optional = true, markers = "extra == \"sentencepiece\""}
 pyyaml = ">=5.1"
@@ -7800,14 +7651,15 @@ tqdm = ">=4.27"
 [package.extras]
 accelerate = ["accelerate (>=0.21.0)"]
 agents = ["Pillow (>=10.0.1,<=15.0)", "accelerate (>=0.21.0)", "datasets (!=2.5.0)", "diffusers", "opencv-python", "sentencepiece (>=0.1.91,!=0.1.92)", "torch"]
-all = ["Pillow (>=10.0.1,<=15.0)", "accelerate (>=0.21.0)", "av (==9.2.0)", "codecarbon (==1.2.0)", "decord (==0.6.0)", "flax (>=0.4.1,<=0.7.0)", "jax (>=0.4.1,<=0.4.13)", "jaxlib (>=0.4.1,<=0.4.13)", "kenlm", "keras-nlp (>=0.3.1)", "librosa", "onnxconverter-common", "optax (>=0.0.8,<=0.1.4)", "optuna", "phonemizer", "protobuf", "pyctcdecode (>=0.4.0)", "ray[tune] (>=2.7.0)", "scipy (<1.13.0)", "sentencepiece (>=0.1.91,!=0.1.92)", "sigopt", "tensorflow (>2.9,<2.16)", "tensorflow-text (<2.16)", "tf2onnx", "timm", "tokenizers (>=0.19,<0.20)", "torch", "torchaudio", "torchvision"]
+all = ["Pillow (>=10.0.1,<=15.0)", "accelerate (>=0.21.0)", "av (==9.2.0)", "codecarbon (==1.2.0)", "decord (==0.6.0)", "flax (>=0.4.1,<=0.7.0)", "jax (>=0.4.1,<=0.4.13)", "jaxlib (>=0.4.1,<=0.4.13)", "kenlm", "keras-nlp (>=0.3.1)", "librosa", "onnxconverter-common", "optax (>=0.0.8,<=0.1.4)", "optuna", "phonemizer", "protobuf", "pyctcdecode (>=0.4.0)", "ray[tune] (>=2.7.0)", "scipy (<1.13.0)", "sentencepiece (>=0.1.91,!=0.1.92)", "sigopt", "tensorflow (>2.9,<2.16)", "tensorflow-text (<2.16)", "tf2onnx", "timm (<=0.9.16)", "tokenizers (>=0.19,<0.20)", "torch", "torchaudio", "torchvision"]
 audio = ["kenlm", "librosa", "phonemizer", "pyctcdecode (>=0.4.0)"]
+benchmark = ["optimum-benchmark (>=0.2.0)"]
 codecarbon = ["codecarbon (==1.2.0)"]
 deepspeed = ["accelerate (>=0.21.0)", "deepspeed (>=0.9.3)"]
-deepspeed-testing = ["GitPython (<3.1.19)", "accelerate (>=0.21.0)", "beautifulsoup4", "cookiecutter (==1.7.3)", "datasets (!=2.5.0)", "deepspeed (>=0.9.3)", "dill (<0.3.5)", "evaluate (>=0.2.0)", "faiss-cpu", "nltk", "optuna", "parameterized", "protobuf", "psutil", "pydantic", "pytest (>=7.2.0,<8.0.0)", "pytest-rich", "pytest-timeout", "pytest-xdist", "rjieba", "rouge-score (!=0.0.7,!=0.0.8,!=0.1,!=0.1.1)", "ruff (==0.1.5)", "sacrebleu (>=1.4.12,<2.0.0)", "sacremoses", "sentencepiece (>=0.1.91,!=0.1.92)", "tensorboard", "timeout-decorator"]
-dev = ["GitPython (<3.1.19)", "Pillow (>=10.0.1,<=15.0)", "accelerate (>=0.21.0)", "av (==9.2.0)", "beautifulsoup4", "codecarbon (==1.2.0)", "cookiecutter (==1.7.3)", "datasets (!=2.5.0)", "decord (==0.6.0)", "dill (<0.3.5)", "evaluate (>=0.2.0)", "faiss-cpu", "flax (>=0.4.1,<=0.7.0)", "fugashi (>=1.0)", "ipadic (>=1.0.0,<2.0)", "isort (>=5.5.4)", "jax (>=0.4.1,<=0.4.13)", "jaxlib (>=0.4.1,<=0.4.13)", "kenlm", "keras-nlp (>=0.3.1)", "librosa", "nltk", "onnxconverter-common", "optax (>=0.0.8,<=0.1.4)", "optuna", "parameterized", "phonemizer", "protobuf", "psutil", "pyctcdecode (>=0.4.0)", "pydantic", "pytest (>=7.2.0,<8.0.0)", "pytest-rich", "pytest-timeout", "pytest-xdist", "ray[tune] (>=2.7.0)", "rhoknp (>=1.1.0,<1.3.1)", "rjieba", "rouge-score (!=0.0.7,!=0.0.8,!=0.1,!=0.1.1)", "ruff (==0.1.5)", "sacrebleu (>=1.4.12,<2.0.0)", "sacremoses", "scikit-learn", "scipy (<1.13.0)", "sentencepiece (>=0.1.91,!=0.1.92)", "sigopt", "sudachidict-core (>=20220729)", "sudachipy (>=0.6.6)", "tensorboard", "tensorflow (>2.9,<2.16)", "tensorflow-text (<2.16)", "tf2onnx", "timeout-decorator", "timm", "tokenizers (>=0.19,<0.20)", "torch", "torchaudio", "torchvision", "unidic (>=1.0.2)", "unidic-lite (>=1.0.7)", "urllib3 (<2.0.0)"]
-dev-tensorflow = ["GitPython (<3.1.19)", "Pillow (>=10.0.1,<=15.0)", "beautifulsoup4", "cookiecutter (==1.7.3)", "datasets (!=2.5.0)", "dill (<0.3.5)", "evaluate (>=0.2.0)", "faiss-cpu", "isort (>=5.5.4)", "kenlm", "keras-nlp (>=0.3.1)", "librosa", "nltk", "onnxconverter-common", "onnxruntime (>=1.4.0)", "onnxruntime-tools (>=1.4.2)", "parameterized", "phonemizer", "protobuf", "psutil", "pyctcdecode (>=0.4.0)", "pydantic", "pytest (>=7.2.0,<8.0.0)", "pytest-rich", "pytest-timeout", "pytest-xdist", "rjieba", "rouge-score (!=0.0.7,!=0.0.8,!=0.1,!=0.1.1)", "ruff (==0.1.5)", "sacrebleu (>=1.4.12,<2.0.0)", "sacremoses", "scikit-learn", "sentencepiece (>=0.1.91,!=0.1.92)", "tensorboard", "tensorflow (>2.9,<2.16)", "tensorflow-text (<2.16)", "tf2onnx", "timeout-decorator", "tokenizers (>=0.19,<0.20)", "urllib3 (<2.0.0)"]
-dev-torch = ["GitPython (<3.1.19)", "Pillow (>=10.0.1,<=15.0)", "accelerate (>=0.21.0)", "beautifulsoup4", "codecarbon (==1.2.0)", "cookiecutter (==1.7.3)", "datasets (!=2.5.0)", "dill (<0.3.5)", "evaluate (>=0.2.0)", "faiss-cpu", "fugashi (>=1.0)", "ipadic (>=1.0.0,<2.0)", "isort (>=5.5.4)", "kenlm", "librosa", "nltk", "onnxruntime (>=1.4.0)", "onnxruntime-tools (>=1.4.2)", "optuna", "parameterized", "phonemizer", "protobuf", "psutil", "pyctcdecode (>=0.4.0)", "pydantic", "pytest (>=7.2.0,<8.0.0)", "pytest-rich", "pytest-timeout", "pytest-xdist", "ray[tune] (>=2.7.0)", "rhoknp (>=1.1.0,<1.3.1)", "rjieba", "rouge-score (!=0.0.7,!=0.0.8,!=0.1,!=0.1.1)", "ruff (==0.1.5)", "sacrebleu (>=1.4.12,<2.0.0)", "sacremoses", "scikit-learn", "sentencepiece (>=0.1.91,!=0.1.92)", "sigopt", "sudachidict-core (>=20220729)", "sudachipy (>=0.6.6)", "tensorboard", "timeout-decorator", "timm", "tokenizers (>=0.19,<0.20)", "torch", "torchaudio", "torchvision", "unidic (>=1.0.2)", "unidic-lite (>=1.0.7)", "urllib3 (<2.0.0)"]
+deepspeed-testing = ["GitPython (<3.1.19)", "accelerate (>=0.21.0)", "beautifulsoup4", "cookiecutter (==1.7.3)", "datasets (!=2.5.0)", "deepspeed (>=0.9.3)", "dill (<0.3.5)", "evaluate (>=0.2.0)", "faiss-cpu", "nltk", "optuna", "parameterized", "protobuf", "psutil", "pydantic", "pytest (>=7.2.0,<8.0.0)", "pytest-rich", "pytest-timeout", "pytest-xdist", "rjieba", "rouge-score (!=0.0.7,!=0.0.8,!=0.1,!=0.1.1)", "ruff (==0.4.4)", "sacrebleu (>=1.4.12,<2.0.0)", "sacremoses", "sentencepiece (>=0.1.91,!=0.1.92)", "tensorboard", "timeout-decorator"]
+dev = ["GitPython (<3.1.19)", "Pillow (>=10.0.1,<=15.0)", "accelerate (>=0.21.0)", "av (==9.2.0)", "beautifulsoup4", "codecarbon (==1.2.0)", "cookiecutter (==1.7.3)", "datasets (!=2.5.0)", "decord (==0.6.0)", "dill (<0.3.5)", "evaluate (>=0.2.0)", "faiss-cpu", "flax (>=0.4.1,<=0.7.0)", "fugashi (>=1.0)", "ipadic (>=1.0.0,<2.0)", "isort (>=5.5.4)", "jax (>=0.4.1,<=0.4.13)", "jaxlib (>=0.4.1,<=0.4.13)", "kenlm", "keras-nlp (>=0.3.1)", "librosa", "nltk", "onnxconverter-common", "optax (>=0.0.8,<=0.1.4)", "optuna", "parameterized", "phonemizer", "protobuf", "psutil", "pyctcdecode (>=0.4.0)", "pydantic", "pytest (>=7.2.0,<8.0.0)", "pytest-rich", "pytest-timeout", "pytest-xdist", "ray[tune] (>=2.7.0)", "rhoknp (>=1.1.0,<1.3.1)", "rjieba", "rouge-score (!=0.0.7,!=0.0.8,!=0.1,!=0.1.1)", "ruff (==0.4.4)", "sacrebleu (>=1.4.12,<2.0.0)", "sacremoses", "scikit-learn", "scipy (<1.13.0)", "sentencepiece (>=0.1.91,!=0.1.92)", "sigopt", "sudachidict-core (>=20220729)", "sudachipy (>=0.6.6)", "tensorboard", "tensorflow (>2.9,<2.16)", "tensorflow-text (<2.16)", "tf2onnx", "timeout-decorator", "timm (<=0.9.16)", "tokenizers (>=0.19,<0.20)", "torch", "torchaudio", "torchvision", "unidic (>=1.0.2)", "unidic-lite (>=1.0.7)", "urllib3 (<2.0.0)"]
+dev-tensorflow = ["GitPython (<3.1.19)", "Pillow (>=10.0.1,<=15.0)", "beautifulsoup4", "cookiecutter (==1.7.3)", "datasets (!=2.5.0)", "dill (<0.3.5)", "evaluate (>=0.2.0)", "faiss-cpu", "isort (>=5.5.4)", "kenlm", "keras-nlp (>=0.3.1)", "librosa", "nltk", "onnxconverter-common", "onnxruntime (>=1.4.0)", "onnxruntime-tools (>=1.4.2)", "parameterized", "phonemizer", "protobuf", "psutil", "pyctcdecode (>=0.4.0)", "pydantic", "pytest (>=7.2.0,<8.0.0)", "pytest-rich", "pytest-timeout", "pytest-xdist", "rjieba", "rouge-score (!=0.0.7,!=0.0.8,!=0.1,!=0.1.1)", "ruff (==0.4.4)", "sacrebleu (>=1.4.12,<2.0.0)", "sacremoses", "scikit-learn", "sentencepiece (>=0.1.91,!=0.1.92)", "tensorboard", "tensorflow (>2.9,<2.16)", "tensorflow-text (<2.16)", "tf2onnx", "timeout-decorator", "tokenizers (>=0.19,<0.20)", "urllib3 (<2.0.0)"]
+dev-torch = ["GitPython (<3.1.19)", "Pillow (>=10.0.1,<=15.0)", "accelerate (>=0.21.0)", "beautifulsoup4", "codecarbon (==1.2.0)", "cookiecutter (==1.7.3)", "datasets (!=2.5.0)", "dill (<0.3.5)", "evaluate (>=0.2.0)", "faiss-cpu", "fugashi (>=1.0)", "ipadic (>=1.0.0,<2.0)", "isort (>=5.5.4)", "kenlm", "librosa", "nltk", "onnxruntime (>=1.4.0)", "onnxruntime-tools (>=1.4.2)", "optuna", "parameterized", "phonemizer", "protobuf", "psutil", "pyctcdecode (>=0.4.0)", "pydantic", "pytest (>=7.2.0,<8.0.0)", "pytest-rich", "pytest-timeout", "pytest-xdist", "ray[tune] (>=2.7.0)", "rhoknp (>=1.1.0,<1.3.1)", "rjieba", "rouge-score (!=0.0.7,!=0.0.8,!=0.1,!=0.1.1)", "ruff (==0.4.4)", "sacrebleu (>=1.4.12,<2.0.0)", "sacremoses", "scikit-learn", "sentencepiece (>=0.1.91,!=0.1.92)", "sigopt", "sudachidict-core (>=20220729)", "sudachipy (>=0.6.6)", "tensorboard", "timeout-decorator", "timm (<=0.9.16)", "tokenizers (>=0.19,<0.20)", "torch", "torchaudio", "torchvision", "unidic (>=1.0.2)", "unidic-lite (>=1.0.7)", "urllib3 (<2.0.0)"]
 flax = ["flax (>=0.4.1,<=0.7.0)", "jax (>=0.4.1,<=0.4.13)", "jaxlib (>=0.4.1,<=0.4.13)", "optax (>=0.0.8,<=0.1.4)", "scipy (<1.13.0)"]
 flax-speech = ["kenlm", "librosa", "phonemizer", "pyctcdecode (>=0.4.0)"]
 ftfy = ["ftfy"]
@@ -7818,25 +7670,26 @@ natten = ["natten (>=0.14.6,<0.15.0)"]
 onnx = ["onnxconverter-common", "onnxruntime (>=1.4.0)", "onnxruntime-tools (>=1.4.2)", "tf2onnx"]
 onnxruntime = ["onnxruntime (>=1.4.0)", "onnxruntime-tools (>=1.4.2)"]
 optuna = ["optuna"]
-quality = ["GitPython (<3.1.19)", "datasets (!=2.5.0)", "isort (>=5.5.4)", "ruff (==0.1.5)", "urllib3 (<2.0.0)"]
+quality = ["GitPython (<3.1.19)", "datasets (!=2.5.0)", "isort (>=5.5.4)", "ruff (==0.4.4)", "urllib3 (<2.0.0)"]
 ray = ["ray[tune] (>=2.7.0)"]
 retrieval = ["datasets (!=2.5.0)", "faiss-cpu"]
+ruff = ["ruff (==0.4.4)"]
 sagemaker = ["sagemaker (>=2.31.0)"]
 sentencepiece = ["protobuf", "sentencepiece (>=0.1.91,!=0.1.92)"]
 serving = ["fastapi", "pydantic", "starlette", "uvicorn"]
 sigopt = ["sigopt"]
 sklearn = ["scikit-learn"]
 speech = ["kenlm", "librosa", "phonemizer", "pyctcdecode (>=0.4.0)", "torchaudio"]
-testing = ["GitPython (<3.1.19)", "beautifulsoup4", "cookiecutter (==1.7.3)", "datasets (!=2.5.0)", "dill (<0.3.5)", "evaluate (>=0.2.0)", "faiss-cpu", "nltk", "parameterized", "psutil", "pydantic", "pytest (>=7.2.0,<8.0.0)", "pytest-rich", "pytest-timeout", "pytest-xdist", "rjieba", "rouge-score (!=0.0.7,!=0.0.8,!=0.1,!=0.1.1)", "ruff (==0.1.5)", "sacrebleu (>=1.4.12,<2.0.0)", "sacremoses", "sentencepiece (>=0.1.91,!=0.1.92)", "tensorboard", "timeout-decorator"]
+testing = ["GitPython (<3.1.19)", "beautifulsoup4", "cookiecutter (==1.7.3)", "datasets (!=2.5.0)", "dill (<0.3.5)", "evaluate (>=0.2.0)", "faiss-cpu", "nltk", "parameterized", "psutil", "pydantic", "pytest (>=7.2.0,<8.0.0)", "pytest-rich", "pytest-timeout", "pytest-xdist", "rjieba", "rouge-score (!=0.0.7,!=0.0.8,!=0.1,!=0.1.1)", "ruff (==0.4.4)", "sacrebleu (>=1.4.12,<2.0.0)", "sacremoses", "sentencepiece (>=0.1.91,!=0.1.92)", "tensorboard", "timeout-decorator"]
 tf = ["keras-nlp (>=0.3.1)", "onnxconverter-common", "tensorflow (>2.9,<2.16)", "tensorflow-text (<2.16)", "tf2onnx"]
-tf-cpu = ["keras (>2.9,<2.16)", "keras-nlp (>=0.3.1)", "onnxconverter-common", "tensorflow-cpu (>2.9,<2.16)", "tensorflow-probability (<2.16)", "tensorflow-text (<2.16)", "tf2onnx"]
+tf-cpu = ["keras (>2.9,<2.16)", "keras-nlp (>=0.3.1)", "onnxconverter-common", "tensorflow-cpu (>2.9,<2.16)", "tensorflow-probability (<0.24)", "tensorflow-text (<2.16)", "tf2onnx"]
 tf-speech = ["kenlm", "librosa", "phonemizer", "pyctcdecode (>=0.4.0)"]
-timm = ["timm"]
+timm = ["timm (<=0.9.16)"]
 tokenizers = ["tokenizers (>=0.19,<0.20)"]
 torch = ["accelerate (>=0.21.0)", "torch"]
 torch-speech = ["kenlm", "librosa", "phonemizer", "pyctcdecode (>=0.4.0)", "torchaudio"]
 torch-vision = ["Pillow (>=10.0.1,<=15.0)", "torchvision"]
-torchhub = ["filelock", "huggingface-hub (>=0.23.0,<1.0)", "importlib-metadata", "numpy (>=1.17)", "packaging (>=20.0)", "protobuf", "regex (!=2019.12.17)", "requests", "sentencepiece (>=0.1.91,!=0.1.92)", "tokenizers (>=0.19,<0.20)", "torch", "tqdm (>=4.27)"]
+torchhub = ["filelock", "huggingface-hub (>=0.23.2,<1.0)", "importlib-metadata", "numpy (>=1.17,<2.0)", "packaging (>=20.0)", "protobuf", "regex (!=2019.12.17)", "requests", "sentencepiece (>=0.1.91,!=0.1.92)", "tokenizers (>=0.19,<0.20)", "torch", "tqdm (>=4.27)"]
 video = ["av (==9.2.0)", "decord (==0.6.0)"]
 vision = ["Pillow (>=10.0.1,<=15.0)"]