[legacy] clean up legacy code (#4743)

* [legacy] remove outdated codes of pipeline (#4692)

* [legacy] remove cli of benchmark and update optim (#4690)

* [legacy] remove cli of benchmark and update optim

* [doc] fix cli doc test

* [legacy] fix engine clip grad norm

* [legacy] remove outdated colo tensor (#4694)

* [legacy] remove outdated colo tensor

* [test] fix test import

* [legacy] move outdated zero to legacy (#4696)

* [legacy] clean up utils (#4700)

* [legacy] clean up utils

* [example] update examples

* [legacy] clean up amp

* [legacy] fix amp module

* [legacy] clean up gpc (#4742)

* [legacy] clean up context

* [legacy] clean core, constants and global vars

* [legacy] refactor initialize

* [example] fix examples ci

* [example] fix examples ci

* [legacy] fix tests

* [example] fix gpt example

* [example] fix examples ci

* [devops] fix ci installation

* [example] fix examples ci

colossalai/legacy/zero/sharded_param/sharded_tensor.py

@@ -0,0 +1,40 @@
+import torch
+
+from colossalai.legacy.zero.gemini.stateful_tensor import StatefulTensor, TensorState
+
+
+class ShardedTensor(StatefulTensor):
+
+    def __init__(self, tensor: torch.Tensor, state: TensorState = TensorState.HOLD) -> None:
+        r"""
+        A tensor sharded in multiple processes. Constructed from an existing torch.Tensor instance.
+        """
+        assert tensor.requires_grad is False
+        super().__init__(tensor, state)
+
+        # kept the shape, numel and dtype of the init tensor.
+        self._origin_shape = tensor.shape
+        self._origin_numel = tensor.numel()
+        self._origin_dtype = tensor.dtype
+        self._is_sharded = False
+
+    @property
+    def dtype(self) -> torch.dtype:
+        assert self._payload.dtype == self._origin_dtype
+        return self._payload.dtype
+
+    @property
+    def origin_numel(self) -> int:
+        return self._origin_numel
+
+    @property
+    def origin_shape(self) -> int:
+        return self._origin_shape
+
+    @property
+    def is_sharded(self):
+        return self._is_sharded
+
+    @is_sharded.setter
+    def is_sharded(self, flag: bool):
+        self._is_sharded = flag