[Frontend] Add torch.tensor binding

python/hidet/graph/frontend/torch/register_functions.py

@@ -10,10 +10,10 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 from __future__ import annotations
-from typing import Optional, Union, Sequence
+from typing import Optional, Union, Sequence, Any
 import operator
 import torch
-from hidet.graph.tensor import Tensor, full_like
+from hidet.graph.tensor import Tensor, full_like, from_torch
 from hidet.graph import ops
 from hidet.utils import same_list
 from hidet.ir.type import DataType
@@ -383,3 +383,17 @@ def bmm(input: Tensor, mat2: Tensor, *, out: Optional[Tensor] = None) -> Tensor:
     if out is not None:
         raise NotImplementedError("hidet: does not support torch.bmm(..., out=...)")
     return ops.matmul(input, mat2)
+
+
+@register_function(torch.tensor)
+def torch_tensor(
+    data: Any, dtype: Optional[torch.dtype] = None, device: Optional[torch.device] = None, requires_grad: bool = False
+) -> Tensor:
+    if requires_grad and torch.is_grad_enabled():
+        warnings.warn_once("hidet: requires_grad=True when torch.is_grad_enabled(), treating as requires_grad=False")
+    if isinstance(data, Tensor):
+        device = device_from_torch(torch_device=device) if device is not None else device
+        return data.to(device=device, dtype=dtype_from_torch(dtype))
+    else:
+        tt = torch.tensor(data, dtype=dtype, device=device)
+        return from_torch(tt)