promptdojo_promptdojo_promptdojo_promptdojo_promptdojo_promptdojo_promptdojo_promptdojo_promptdojo_promptdojo_promptdojo_

CNNs and local patterns — step 7 of 7

ch 44 · cnns, transformers, and useful llm internalsch 44 · cnns, transformers, and useful llm internalsch 44 · cnns, transformers, and useful llm internalsch 44 · cnns, transformers, and useful llm internalsch 44 · cnns, transformers, and useful llm internalsch 44 · cnns, transformers, and useful llm internalsch 44 · cnns, transformers, and useful llm internalsch 44 · cnns, transformers, and useful llm internalsch 44 · cnns, transformers, and useful llm internalsch 44 · cnns, transformers, and useful llm internalsch 44 · cnns, transformers, and useful llm internals7/7
Checkpoint

One last thing before we move on. pass this to mark the lesson done, or skip and keep moving. hop to the next when you're ready.

Demonstrate BOTH properties that make CNNs work. (1) Translation: run the [-1, 1] kernel over two signals whose jump sits at different positions and print f"{name}: up-jump found at window {i}" for each — same kernel, both found. (2) Weight sharing: an 8-input dense layer mapping to 8 outputs needs 8*8 weights; the kernel needs len(kernel). Print f"dense layer params: {d} conv kernel params: {c}" and f"parameter ratio: {d // c}x".

full-screen editor opens — close anytime to keep reading.

open the editor to run it.