Merge pull request #2311 from opentensor/feat/roman/remove-class-tensor

remove unused code (tensor.py-> class tensor), remove old tests, add new tests

bittensor/core/tensor.py

@@ -133,13 +133,6 @@ def cast_shape(raw: Union[None, List[int], str]) -> Optional[Union[str, list]]:
         )
 
 
-class tensor:
-    def __new__(cls, tensor: Union[list, np.ndarray, "torch.Tensor"]):
-        if isinstance(tensor, list) or isinstance(tensor, np.ndarray):
-            tensor = torch.tensor(tensor) if use_torch() else np.array(tensor)
-        return Tensor.serialize(tensor_=tensor)
-
-
 class Tensor(BaseModel):
     """
     Represents a Tensor object.
@@ -158,7 +151,7 @@ def tensor(self) -> Union[np.ndarray, "torch.Tensor"]:
     def tolist(self) -> List[object]:
         return self.deserialize().tolist()
 
-    def numpy(self) -> "numpy.ndarray":
+    def numpy(self) -> "np.ndarray":
         return (
             self.deserialize().detach().numpy() if use_torch() else self.deserialize()
         )

bittensor/utils/deprecated.py

@@ -96,7 +96,7 @@
 from bittensor.core.stream import StreamingSynapse  # noqa: F401
 from bittensor.core.subtensor import Subtensor
 from bittensor.core.synapse import TerminalInfo, Synapse  # noqa: F401
-from bittensor.core.tensor import tensor, Tensor  # noqa: F401
+from bittensor.core.tensor import Tensor  # noqa: F401
 from bittensor.core.threadpool import (  # noqa: F401
     PriorityThreadPoolExecutor as PriorityThreadPoolExecutor,
 )

tests/unit_tests/test_tensor.py

@@ -20,7 +20,7 @@
 import pytest
 import torch
 
-from bittensor.core.tensor import tensor as tensor_class, Tensor
+from bittensor.core.tensor import Tensor
 
 
 # This is a fixture that creates an example tensor for testing
@@ -30,7 +30,7 @@ def example_tensor():
     data = np.array([1, 2, 3, 4])
 
     # Serialize the tensor into a Tensor instance and return it
-    return tensor_class(data)
+    return Tensor.serialize(data)
 
 
 @pytest.fixture
@@ -39,7 +39,7 @@ def example_tensor_torch(force_legacy_torch_compatible_api):
     data = torch.tensor([1, 2, 3, 4])
 
     # Serialize the tensor into a Tensor instance and return it
-    return tensor_class(data)
+    return Tensor.serialize(data)
 
 
 def test_deserialize(example_tensor):
@@ -175,71 +175,71 @@ def test_shape_field_torch(force_legacy_torch_compatible_api):
 
 
 def test_serialize_all_types():
-    tensor_class(np.array([1], dtype=np.float16))
-    tensor_class(np.array([1], dtype=np.float32))
-    tensor_class(np.array([1], dtype=np.float64))
-    tensor_class(np.array([1], dtype=np.uint8))
-    tensor_class(np.array([1], dtype=np.int32))
-    tensor_class(np.array([1], dtype=np.int64))
-    tensor_class(np.array([1], dtype=bool))
+    Tensor.serialize(np.array([1], dtype=np.float16))
+    Tensor.serialize(np.array([1], dtype=np.float32))
+    Tensor.serialize(np.array([1], dtype=np.float64))
+    Tensor.serialize(np.array([1], dtype=np.uint8))
+    Tensor.serialize(np.array([1], dtype=np.int32))
+    Tensor.serialize(np.array([1], dtype=np.int64))
+    Tensor.serialize(np.array([1], dtype=bool))
 
 
 def test_serialize_all_types_torch(force_legacy_torch_compatible_api):
-    tensor_class(torch.tensor([1], dtype=torch.float16))
-    tensor_class(torch.tensor([1], dtype=torch.float32))
-    tensor_class(torch.tensor([1], dtype=torch.float64))
-    tensor_class(torch.tensor([1], dtype=torch.uint8))
-    tensor_class(torch.tensor([1], dtype=torch.int32))
-    tensor_class(torch.tensor([1], dtype=torch.int64))
-    tensor_class(torch.tensor([1], dtype=torch.bool))
+    Tensor.serialize(torch.tensor([1], dtype=torch.float16))
+    Tensor.serialize(torch.tensor([1], dtype=torch.float32))
+    Tensor.serialize(torch.tensor([1], dtype=torch.float64))
+    Tensor.serialize(torch.tensor([1], dtype=torch.uint8))
+    Tensor.serialize(torch.tensor([1], dtype=torch.int32))
+    Tensor.serialize(torch.tensor([1], dtype=torch.int64))
+    Tensor.serialize(torch.tensor([1], dtype=torch.bool))
 
 
 def test_serialize_all_types_equality():
     rng = np.random.default_rng()
 
     tensor = rng.standard_normal((100,), dtype=np.float32)
-    assert np.all(tensor_class(tensor).tensor() == tensor)
+    assert np.all(Tensor.serialize(tensor).tensor() == tensor)
 
     tensor = rng.standard_normal((100,), dtype=np.float64)
-    assert np.all(tensor_class(tensor).tensor() == tensor)
+    assert np.all(Tensor.serialize(tensor).tensor() == tensor)
 
     tensor = np.random.randint(255, 256, (1000,), dtype=np.uint8)
-    assert np.all(tensor_class(tensor).tensor() == tensor)
+    assert np.all(Tensor.serialize(tensor).tensor() == tensor)
 
     tensor = np.random.randint(2_147_483_646, 2_147_483_647, (1000,), dtype=np.int32)
-    assert np.all(tensor_class(tensor).tensor() == tensor)
+    assert np.all(Tensor.serialize(tensor).tensor() == tensor)
 
     tensor = np.random.randint(
         9_223_372_036_854_775_806, 9_223_372_036_854_775_807, (1000,), dtype=np.int64
     )
-    assert np.all(tensor_class(tensor).tensor() == tensor)
+    assert np.all(Tensor.serialize(tensor).tensor() == tensor)
 
     tensor = rng.standard_normal((100,), dtype=np.float32) < 0.5
-    assert np.all(tensor_class(tensor).tensor() == tensor)
+    assert np.all(Tensor.serialize(tensor).tensor() == tensor)
 
 
 def test_serialize_all_types_equality_torch(force_legacy_torch_compatible_api):
     torchtensor = torch.randn([100], dtype=torch.float16)
-    assert torch.all(tensor_class(torchtensor).tensor() == torchtensor)
+    assert torch.all(Tensor.serialize(torchtensor).tensor() == torchtensor)
 
     torchtensor = torch.randn([100], dtype=torch.float32)
-    assert torch.all(tensor_class(torchtensor).tensor() == torchtensor)
+    assert torch.all(Tensor.serialize(torchtensor).tensor() == torchtensor)
 
     torchtensor = torch.randn([100], dtype=torch.float64)
-    assert torch.all(tensor_class(torchtensor).tensor() == torchtensor)
+    assert torch.all(Tensor.serialize(torchtensor).tensor() == torchtensor)
 
     torchtensor = torch.randint(255, 256, (1000,), dtype=torch.uint8)
-    assert torch.all(tensor_class(torchtensor).tensor() == torchtensor)
+    assert torch.all(Tensor.serialize(torchtensor).tensor() == torchtensor)
 
     torchtensor = torch.randint(
         2_147_483_646, 2_147_483_647, (1000,), dtype=torch.int32
     )
-    assert torch.all(tensor_class(torchtensor).tensor() == torchtensor)
+    assert torch.all(Tensor.serialize(torchtensor).tensor() == torchtensor)
 
     torchtensor = torch.randint(
         9_223_372_036_854_775_806, 9_223_372_036_854_775_807, (1000,), dtype=torch.int64
     )
-    assert torch.all(tensor_class(torchtensor).tensor() == torchtensor)
+    assert torch.all(Tensor.serialize(torchtensor).tensor() == torchtensor)
 
     torchtensor = torch.randn([100], dtype=torch.float32) < 0.5
-    assert torch.all(tensor_class(torchtensor).tensor() == torchtensor)
+    assert torch.all(Tensor.serialize(torchtensor).tensor() == torchtensor)