Tensors & Neural Plumbing

Shapes, matmuls, forward and backward passes, parameter counts, memory footprints.

18concepts
211flashcards
138minutes of reading
  1. 01 Broadcasting and Vectorisation Broadcasting is the rule that lets a bias vector add to every row of a batch without an explicit loop, and understanding its shape-matching logic prevents the class of bugs that produce wrong answers without an error. beginner 6m 5 cards
  2. 02 Matrix Multiplication: The Core Operation Nearly every FLOP a transformer spends is a matrix multiplication; understanding its shape rule and its cost is the single most load-bearing piece of maths in deep learning. beginner 7m 4 cards
  3. 03 Tensors, Shapes, and Batching Every number a transformer touches lives inside a tensor with a fixed shape; learning to read and predict those shapes is the fastest way to stop being confused by model code. beginner 7m 4 cards
  4. 04 Why Non-Linearity Matters Stack any number of linear layers and the result is still one linear layer; the humble activation function is the only thing standing between a transformer and a glorified matrix multiplication. beginner 6m 4 cards
  5. 05 Checkpoint Formats and State Dicts A checkpoint is a dictionary from parameter names to tensors plus a pile of implicit assumptions about dtype, sharding, and key naming, and every one of those assumptions is somewhere a load goes wrong. intermediate 7m 24 cards
  6. 06 Counting Transformer Parameters A transformer's parameter count is not a mysterious headline number, it is the sum of a handful of matrix shapes multiplied out, and knowing the formula lets you sanity-check any model card in seconds. intermediate 8m 4 cards
  7. 07 Einsum and Tensor Contractions Einstein summation notation expresses matmuls, batched matmuls, and attention itself as one uniform pattern, and reading it fluently is the fastest way to understand what a line of unfamiliar model code actually computes. intermediate 8m 4 cards
  8. 08 Forward-Mode and Reverse-Mode Autodiff Backpropagation is one of two ways to apply the chain rule mechanically, and the choice between them is decided by a single number, the ratio of inputs to outputs, which is why training uses reverse mode and Hessian-vector products use both. intermediate 8m 24 cards
  9. 09 Padding, Masks, and Variable-Length Batching Real batches contain sequences of different lengths, and the three ways of handling that difference, padding, packing, and varlen kernels, have wildly different costs and each has a signature bug. intermediate 8m 24 cards
  10. 10 Tensor Memory Layout and Contiguity A tensor is a pointer, a shape, and a stride tuple; transposes and slices change only the metadata, which is why some reshapes are free, some silently copy gigabytes, and a permute in the wrong place can halve your throughput. intermediate 7m 15 cards
  11. 11 The Backward Pass and Gradient Flow Backpropagation through a transformer is the forward pass run in reverse with the chain rule attached; seeing which operations preserve gradient magnitude and which shrink it explains why architecture choices exist. intermediate 8m 5 cards
  12. 12 The Forward Pass End to End Trace one token's numbers from embedding table to output logits and every "mysterious" transformer component turns out to be a shape-preserving or shape-mixing step in a fixed pipeline. intermediate 9m 4 cards
  13. 13 Views, Aliasing, and In-Place Operations A transpose returns a new tensor object that shares its old storage, which is why in-place edits leak across variables you thought were separate and why autograd raises "a variable needed for gradient computation has been modified". intermediate 7m 24 cards
  14. 14 Accumulation Precision and Mixed-Dtype Plumbing Storing weights in 16 bits is safe because almost nothing is actually computed in 16 bits — matmuls accumulate in fp32, reductions run in fp32, and the master copy is fp32, and every documented failure of low-precision training is one of these rules being broken. advanced 8m 15 cards
  15. 15 Batch Invariance and Numerical Reproducibility Temperature zero does not give the same answer twice, and the reason is neither random seeds nor GPU atomics; it is that reduction kernels change their split strategy with batch shape, so your logits depend on which other requests happened to be in flight. advanced 8m 28 cards
  16. 16 Signal Propagation and Initialisation How a network's weights are initialised decides, before a single gradient step, whether activations and gradients stay in a trainable range or collapse to zero or explode across depth. advanced 9m 4 cards
  17. 17 Training Memory Footprint A model's weights are the smallest part of its training memory bill; optimiser state, gradients, and activations usually cost several times more, and knowing the breakdown explains why training needs far more memory than inference. advanced 9m 4 cards
  18. 18 muP and Hyperparameter Transfer Under standard parametrisation the best learning rate drifts as a model gets wider, so every scale-up is a fresh search; muP rescales initialisation, learning rates, and multipliers per layer so the optimum stays put and can be tuned on a small proxy model. advanced 8m 15 cards