Delete pipeline.py

pipeline.py

@@ -1,70 +0,0 @@
-from __future__ import annotations
-
-from typing import Any, Dict
-
-from transformers import Pipeline
-
-try:
-    from .runtime import JNUTSBRuntime
-except ImportError:  # pragma: no cover - local execution fallback
-    from runtime import JNUTSBRuntime
-
-
-class JNUTSBPipeline(Pipeline):
-    """Transformers custom pipeline for JNU-TSB.
-
-    Example:
-        from transformers import pipeline
-        pipe = pipeline(
-            task="jnu-tsb",
-            model="HONGRIZON/JNU-TSB",
-            trust_remote_code=True,
-            device=-1,
-        )
-        pipe({"stock": [...], "news": [...]}, prediction_length=5)
-    """
-
-    def __init__(self, *args: Any, **kwargs: Any) -> None:
-        self._runtime = None
-        super().__init__(*args, **kwargs)
-
-    @property
-    def runtime(self) -> JNUTSBRuntime:
-        if self._runtime is None:
-            device = getattr(self, "device", None)
-            if device is None:
-                device_str = None
-            elif getattr(device, "type", "cpu") == "cuda":
-                idx = getattr(device, "index", None)
-                device_str = "cuda" if idx is None else f"cuda:{idx}"
-            else:
-                device_str = "cpu"
-            self._runtime = JNUTSBRuntime.from_config(self.model.config, device=device_str)
-        return self._runtime
-
-    def _sanitize_parameters(self, **kwargs: Any):
-        allowed = {
-            "news",
-            "stock",
-            "future_news",
-            "future_covariates",
-            "prediction_length",
-            "quantile_levels",
-            "timestamp_column",
-            "target",
-            "id_column",
-            "use_llm_extractor",
-        }
-        forward_kwargs = {k: v for k, v in kwargs.items() if k in allowed}
-        return {}, forward_kwargs, {}
-
-    def preprocess(self, inputs: Dict[str, Any]) -> Dict[str, Any]:
-        if not isinstance(inputs, dict):
-            raise TypeError("inputs는 {'stock': ..., 'news': ...} 형태의 dict여야 합니다.")
-        return inputs
-
-    def _forward(self, model_inputs: Dict[str, Any], **forward_kwargs: Any) -> Dict[str, Any]:
-        return self.runtime.predict(inputs=model_inputs, **forward_kwargs)
-
-    def postprocess(self, model_outputs: Dict[str, Any]) -> Dict[str, Any]:
-        return model_outputs