reverting changes

__init__.py

@@ -1,4 +0,0 @@
-from .modeling_custom import CustomConfig, CustomModel
-from .custom_pipeline import Pipeline_One
-
-__all__ = ["CustomConfig", "CustomModel", "Pipeline_One"]

config.json

@@ -1,9 +1,9 @@
 {
-  "_name_or_path": "Maslionok/pipeline1",
+  "_name_or_path": "Maslionok/pipeline1/LID-40-3-2000000-1-4.bin",
   "architectures": [
-    "CustomModel"
+    "Pipeline_One"
   ],
-  "model_type": "langident",
+  "model_type": "qwen2",
   "num_labels": 3,
   "id2label": {
     "0": "English",
@@ -17,9 +17,9 @@
   },
   "custom_pipelines": {
     "language-detection": {
-      "impl": "custom_pipeline.Pipeline_One",
+      "impl": "impresso_langident_wrapper/Pipeline_One",
       "pt": [],
       "tf": []
     }
   }
-}
+}

configuration_custom.py

@@ -1,23 +0,0 @@
-from transformers import PretrainedConfig
-
-class CustomConfig(PretrainedConfig):
-    model_type = "custom"
-    
-    def __init__(
-        self,
-        num_labels=3,
-        id2label=None,
-        label2id=None,
-        **kwargs
-    ):
-        if id2label is None:
-            id2label = {0: "English", 1: "German", 2: "French"}
-        if label2id is None:
-            label2id = {"English": 0, "German": 1, "French": 2}
-            
-        super().__init__(
-            num_labels=num_labels,
-            id2label=id2label,
-            label2id=label2id,
-            **kwargs
-        )

impresso_langident_wrapper.py

@@ -1,30 +1,27 @@
 from transformers import Pipeline
 from transformers.pipelines import PIPELINE_REGISTRY
-from typing import Dict, List, Union
+
+
 
 class Pipeline_One(Pipeline):
-    def __init__(self, model=None, tokenizer=None, **kwargs):
-        super().__init__(model=model, tokenizer=tokenizer, **kwargs)
-        
     def _sanitize_parameters(self, **kwargs):
-        preprocess_kwargs = {}
-        forward_kwargs = {}
-        postprocess_kwargs = {}
-        return preprocess_kwargs, forward_kwargs, postprocess_kwargs
+        # Add any additional parameter handling if necessary
+        return kwargs, {}, {}
 
-    def preprocess(self, inputs, **kwargs) -> Dict[str, Union[str, List[str]]]:
-        return {"text": inputs if isinstance(inputs, list) else [inputs]}
+    def preprocess(self, text, **kwargs):
+        return text
 
     def _forward(self, inputs):
-        # Implement your model logic here
-        return ["en"]  # placeholder for actual model prediction
+        model_output = self.model.predict(inputs, k=1)
+        return model_output
 
-    def postprocess(self, model_outputs):
-        return model_outputs[0] if isinstance(model_outputs, list) else model_outputs
+    def postprocess(self, outputs, **kwargs):
+        return outputs
+    
 
 # Register the pipeline
 PIPELINE_REGISTRY.register_pipeline(
     "language-detection",
     pipeline_class=Pipeline_One,
-    default={"model": None}
-)
+)
+

modeling_custom.py

@@ -1,18 +0,0 @@
-from transformers import PreTrainedModel
-from transformers.configuration_utils import PretrainedConfig
-
-class CustomConfig(PretrainedConfig):
-    model_type = "langident"  # Changed from "custom" to a specific type
-
-    def __init__(self, **kwargs):
-        super().__init__(**kwargs)
-
-class CustomModel(PreTrainedModel):
-    config_class = CustomConfig
-    
-    def __init__(self, config):
-        super().__init__(config)
-        
-    def forward(self, **kwargs):
-        # Implement your forward pass
-        pass