Browse Source
fix(imperative): remove convert_inputs from concat
GitOrigin-RevId: 1511cb4b43
tags/v1.9.0
Megvii Engine Team
3 years ago
1 changed files with 8 additions and 8 deletions
Unified View
Diff Options
+ 8
- 8
imperative/python/megengine/functional/tensor.py
View File
| @@ -399,8 +399,6 @@ def concat(inps: Iterable[Tensor], axis: int = 0, device=None) -> Tensor: | |||||
| if len(inps) == 1: | if len(inps) == 1: | ||||
| return inps[0] | return inps[0] | ||||
| # FIXME: remove this convert_inputs | |||||
| inps = convert_inputs(*inps, device=device) | |||||
| if device is None: | if device is None: | ||||
| device = get_device(inps) | device = get_device(inps) | ||||
| device = as_device(device) | device = as_device(device) | ||||
| @@ -1168,9 +1166,10 @@ def repeat(inp: Tensor, repeats: int, axis: Optional[int] = None): | |||||
| bcast_shape.append(shape[axis + 1 :]) | bcast_shape.append(shape[axis + 1 :]) | ||||
| target_shape.append(shape[axis + 1 :]) | target_shape.append(shape[axis + 1 :]) | ||||
| out = broadcast_to(inp.reshape(concat(base_shape)), concat(bcast_shape)).reshape( | |||||
| concat(target_shape) | |||||
| ) | |||||
| base_shape = astensor1d(base_shape) | |||||
| bcast_shape = astensor1d(bcast_shape) | |||||
| target_shape = astensor1d(target_shape) | |||||
| out = broadcast_to(inp.reshape(base_shape), bcast_shape).reshape(target_shape) | |||||
| return out | return out | ||||
| @@ -1191,9 +1190,10 @@ def _tile_one_dim(inp, rep, axis): | |||||
| if axis + 1 <= max_axis: | if axis + 1 <= max_axis: | ||||
| target_shape.append(shape[axis + 1 :]) | target_shape.append(shape[axis + 1 :]) | ||||
| out = broadcast_to(inp.reshape(concat(base_shape)), concat(bcast_shape)).reshape( | |||||
| concat(target_shape) | |||||
| ) | |||||
| base_shape = astensor1d(base_shape) | |||||
| bcast_shape = astensor1d(bcast_shape) | |||||
| target_shape = astensor1d(target_shape) | |||||
| out = broadcast_to(inp.reshape(base_shape), bcast_shape).reshape(target_shape) | |||||
| return out | return out | ||||