Where does transformer collide?

An iron-cored device that moves alternating electrical energy from one circuit to another by electromagnetic induction, raising or lowering the voltage in the process and using no moving parts. Faraday's 1831 induction ring, two coils wound on a shared iron core, first showed that a changing current in one winding induces a current in the other, as reported in his Experimental researches in electricity (1832). The practical closed-core device that engineers now call a transformer was built by the Ganz works team of Zipernowsky, Bláthy, and Déri in 1885 and by William Stanley in 1886, and Coltman's 1988 The Transformer surveys how it works. Here the referent is a piece of power-system hardware rated in volts and turns, not a mythic being who reshapes the world or a neural network that processes text.[1][2]

Unit

One transformer, rated by its voltage ratio, turns ratio, and power in volt-amperes

Recovery means

Not applicable

Measured by

Characterised by its turns ratio, primary and secondary voltages, and efficiency under load

In the study of myth, a Transformer is a culture hero or world-maker who travels through an unfinished world and puts it into its present form, turning people into animals or stone, fixing the courses of rivers, and setting the rules that living things now follow. Boas gathered and named these figures across the Pacific Northwest Coast in his 1916 Tsimshian mythology, where the raven Txamsem acts as the great Transformer, and comparable beings recur among neighbouring peoples. The referent is a narrative character in oral tradition, not an electrical machine or a computing architecture that happen to share the word.[3]

Unit

One Transformer figure within a body of oral tradition

Recovery means

Not applicable

Measured by

Identified by folklorists through recurring motifs of shape-changing and world-ordering in recorded tales

A neural network architecture that processes a whole sequence at once by letting each position attend to every other position through a step called self-attention, rather than reading the sequence one item at a time. Vaswani and colleagues introduced it and coined the name in their 2017 Attention is all you need, showing that attention on its own, with no recurrence or convolution, could model language, and it became the basis for the large language models that followed. The referent is a piece of software defined by learned weights and attention layers, unrelated to a power-system device or a mythic world-shaper that carry the same name.[4]

Unit

One trained model, sized by its parameter count and layer depth

Recovery means

Not applicable

Measured by

Evaluated by accuracy or loss on benchmarks such as machine translation and language modelling