Remove redundant imports

Author: minhqngo

minitorch/autodiff.py

@@ -1,5 +1,5 @@
 from dataclasses import dataclass
-from typing import Any, Iterable, List, Tuple
+from typing import Any, Iterable, Tuple
 
 from typing_extensions import Protocol
 

minitorch/nn/init.py

@@ -1,7 +1,7 @@
 import math
 
 
-def kaiming_uniform(tensor, fan_in, **kwargs):
+def kaiming_uniform(tensor, fan_in):
     bound = math.sqrt(6 / fan_in)
     tensor.uniform_(-bound, bound)
 
@@ -11,14 +11,14 @@ def glorot_uniform(tensor, fan_in, fan_out):
     tensor.uniform_(-bound, bound)
 
 
-def lecun_uniform(tensor, fan_in, **kwargs):
+def lecun_uniform(tensor, fan_in):
     bound = math.sqrt(3 / fan_in)
     tensor.uniform_(-bound, bound)
 
 
-def zero(tensor, **kwargs):
+def zero(tensor):
     tensor.fill_(0.0)
 
 
-def one(tensor, **kwargs):
+def one(tensor):
     tensor.fill_(1.0)

minitorch/nn/layers.py

@@ -1,8 +1,7 @@
 from ..tensor.functions import rand, zeros
 from .module import Module, Parameter
-from ..backends import fast_conv, fast_ops
+from ..backends import fast_conv
 from . import init
-from ..tensor.tensor import Tensor
 
 try:
     from ..backends import cuda_conv
@@ -37,24 +36,19 @@ def __init__(self, in_channels, out_channels, kernel_width, backend, stride=1, i
         self.kernel_width = kernel_width
 
     def forward(self, input):
-        # Perform strided convolution manually if stride != 1
         batch, in_channels, w = input.shape
         kw = self.kernel_width
         stride = self.stride
         out_channels = self.weights.value.shape[0]
 
-        # Calculate output width with stride
         out_w = (w - kw) // stride + 1
 
-        # Create output tensor
         output = input.zeros((batch, out_channels, out_w))
 
-        # Perform convolution with stride
         for b in range(batch):
             for oc in range(out_channels):
                 for ow in range(out_w):
                     start_w = ow * stride
-                    # Compute convolution for this window
                     total = 0.0
                     for ic in range(in_channels):
                         for k in range(kw):
@@ -83,7 +77,6 @@ def __init__(self, in_channels, out_channels, kernel, backend, stride=1, initial
         self.backend = backend
 
     def forward(self, input):
-        # Use CUDA conv if backend is CUDA, otherwise use fast conv
         if self.backend.cuda and cuda_conv is not None:
             out = cuda_conv.conv2d(input, self.weights.value, self.stride) + self.bias.value
         else:

minitorch/nn/loss.py

@@ -1,4 +1,3 @@
-from ..tensor import tensor
 from .nn import logsoftmax