[bug] `linearBlockMinifloat` backward has arguments which are not passed

[johanjino]

Hi, I was working on the adls labs when I found out that the linearBlockMinifloat has backward which takes arguments same as the forward pass:

mase/src/chop/nn/quantizers/block_minifloat.py

Lines 84 to 92 in 488d49e

	def backward(
	ctx,
	grad_output: Tensor,
	width: int,
	exponent_width: int,
	exponent_bias_width: int,
	block_size: list[int] | int = [16],
	skip_first_dim: bool = False,
	):

Other precision backward passes do not have this arguments. Further, torch when calculating backprop wont pass these arguments, hence causing Missing 3 positional arguments error (Block size has default). This is either a bug, or have not been implemented yet, this precision hence cannot be used for QAT (only backward is broken).

[johanjino]

If these args need removal, I am happy to create a PR for it 😄.

[Aaron-Zhao123]

@ChengZhang-98 do you want to take a look?