[KongchangAI]
Concepttensor

Tensor

The most fundamental data structure in machine learning frameworks, a generalization of scalars, vectors and matrices to higher dimensions. In image processing, a color image is represented as a four-dimensional tensor [batch, height, width, channels], while video adds a temporal dimension to become five-dimensional.

Timeline (last 90 days)

Nov 10

Understanding tensors requires grasping three basic concepts: rank (how many axes a tensor has), axis (the dimension along each direction), and shape (the size along each dimension).

Unverified50%
Nov 10

In PyTorch, scalars, vectors, and matrices can all be uniformly called tensors; a vector is a one-dimensional tensor, a matrix is a two-dimensional tensor, and a tensor is a multi-dimensional data container with a shape.

Unverified50%
Nov 3

In deep learning, parameters are stored as tensors in GPU memory or on disk; reading means loading and parsing, while writing involves format, precision and consistency constraints and often requires coordination with the computation graph and optimizer state.

Unverified50%
Oct 18

In Lean, tensors can be represented as indexed finite-dimensional array types, with shape constraints statically checked at compile time via dependent types.

Unverified50%
Oct 13

A tensor is the core data structure in machine learning, and can be understood as a general term for multi-dimensional arrays: a scalar is a 0-dimensional tensor, a vector is 1-dimensional, a matrix is 2-dimensional, and higher dimensions are N-dimensional tensors.

Unverified50%
Oct 13

A forward pass is the process of a neural network computing once completely from the input layer to the output layer; an autoregressive language model needs N forward passes to generate N tokens, with latency growing linearly with output length; a classifier-style model needs only one forward pass.

Unverified60%
Oct 12

张量是机器学习框架中最基础的数据结构,彩色图像通常表示为形状为[batch, height, width, channels]的四维张量,视频则再加一维时间轴。

Unverified50%
Sep 22

Neural network weights are essentially multi-dimensional arrays (tensors) stored in file formats such as .safetensors, .gguf, or .bin, and the numbers themselves are static and cannot execute on their own.

Unverified50%
Sep 21

Stride records the number of memory units that must be skipped to move one step in each dimension, enabling operations such as transposition and slicing to be completed by modifying strides without copying data

Unverified50%
Sep 21

At the implementation level, a tensor is a contiguous block of one-dimensional memory, with two pieces of metadata—shape and stride—describing the multi-dimensional logical view

Unverified50%

1 more timeline events

All Facts (11)

Unverified

A forward pass is the process of a neural network computing once completely from the input layer to the output layer; an autoregressive language model needs N forward passes to generate N tokens, with latency growing linearly with output length; a classifier-style model needs only one forward pass.

60%
Unverified

PyTorch tensors are essentially multi-dimensional arrays, highly similar to NumPy's ndarray, but tensors can run on GPUs and have built-in automatic differentiation support.

60%
Unverified

Understanding tensors requires grasping three basic concepts: rank (how many axes a tensor has), axis (the dimension along each direction), and shape (the size along each dimension).

50%
Unverified

In PyTorch, scalars, vectors, and matrices can all be uniformly called tensors; a vector is a one-dimensional tensor, a matrix is a two-dimensional tensor, and a tensor is a multi-dimensional data container with a shape.

50%
Unverified

In deep learning, parameters are stored as tensors in GPU memory or on disk; reading means loading and parsing, while writing involves format, precision and consistency constraints and often requires coordination with the computation graph and optimizer state.

50%
Unverified

In Lean, tensors can be represented as indexed finite-dimensional array types, with shape constraints statically checked at compile time via dependent types.

50%
Unverified

A tensor is the core data structure in machine learning, and can be understood as a general term for multi-dimensional arrays: a scalar is a 0-dimensional tensor, a vector is 1-dimensional, a matrix is 2-dimensional, and higher dimensions are N-dimensional tensors.

50%
Unverified

张量是机器学习框架中最基础的数据结构,彩色图像通常表示为形状为[batch, height, width, channels]的四维张量,视频则再加一维时间轴。

50%
Unverified

Neural network weights are essentially multi-dimensional arrays (tensors) stored in file formats such as .safetensors, .gguf, or .bin, and the numbers themselves are static and cannot execute on their own.

50%
Unverified

Stride records the number of memory units that must be skipped to move one step in each dimension, enabling operations such as transposition and slicing to be completed by modifying strides without copying data

50%
Unverified

At the implementation level, a tensor is a contiguous block of one-dimensional memory, with two pieces of metadata—shape and stride—describing the multi-dimensional logical view

50%

Source Articles

PyTorch Tensors from Beginner to Practice: A Complete Guide to Core Concepts and OperationsNov 10
本·阿弗莱克谈AI:从CNN到Transformer的电影特效之路Oct 12