Return a function that formats the value given to it with the given label.

Determine the quantization statistics for input matrix `A` in accordance to the `LLM.int8()` algorithm. The statistics are determined both row-wi

Performs an 8-bit integer matrix multiplication. A linear transformation is applied such that `out = A @ B.T`. When possible, integer tensor core

Performs dequantization on the result of a quantized int8 matrix multiplication. Args: A (`torch.Tensor` with dtype `torch.int32`): The r

Dequantizes a tensor with dtype `torch.int8` to `torch.float32`. Args: A (`torch.Tensor` with dtype `torch.int8`): The quantized int8 ten

Quantizes a tensor with dtype `torch.float16` to `torch.int8` in accordance to the `LLM.int8()` algorithm. For more information, see the [LLM.int

Efficient NEON float LUT lookup using indexed lane extraction The LUT has 16 float entries stored as a flat array for direct indexing

Performs an inplace optimizer update with one or two optimizer states. Universal optimizer update for 32-bit state and 32/16-bit gradients/w

32-bit optimizer implemented by Triton

Pure PyTorch implementation of the 8-bit block-wise optimizer update step. This version ensures high-precision updates for float16 parameters

Pure PyTorch implementation of the 8-bit block-wise optimizer update step. This version ensures high-precision updates for float16 parameters

Replace linear modules with a new Linear module. Parameters: model (`torch.nn.Module`): Input model or `torch.nn.Module`

Simulates torch+rocm7.0.

Perform a single optimization step. Arguments: closure (`Callable`, *optional*, defaults to `None`): A closure th