adam_step

std.optim.adam_step · Level L2

One step of Adam (Kingma and Ba) on the state [params; m; v] at step t (1, 2, …): m ← β₁m + (1 − β₁)g, v ← β₂v + (1 − β₂)g², then params ← params − lr·m̂/(√v̂ + ε) with the bias corrections m̂ = m/(1 − β₁ᵗ), v̂ = v/(1 − β₂ᵗ).

m ← β₁m + (1−β₁)g, v ← β₂v + (1−β₂)g², θ ← θ − η·(m/(1−β₁ᵗ)) / (√(v/(1−β₂ᵗ)) + ε)

Signature

adam_step(state: f64[3, n], grad: f64[n], t: f64[], lr: f64[], beta1: f64[], beta2: f64[], eps: f64[]) → f64[3, n]

Structure

The function as NOVA stores it: one box per input, operation and output, and arrows that carry values. A double border marks another library function this one runs — called once, or by Scan once per element; select it to open that function.

beta1f64[]statef64[3, n]tf64[]1.0gradf64[n]epsf64[]beta2f64[]lrf64[]SlicepSlicemSubtractc1Logl1Logl2SlicevMultiplyg2Subtractc2MultiplybmMultiplycgMultiplytl1Multiplytl2MultiplybvMultiplycg2Addm2Expp1Expp2Addv2Subtractk1Subtractk2DividemhDividevhSqrtrvAdddenDividedirMultiplystepSubtractpnConcatnextnextf64[3, n]
  • input
  • operation
  • constant
  • call
  • output

Verification

  • Signature proven by NOVA’s shape solver, for every size.
  • Agrees with the reference Adam's update written out element by element to 80 digits (100-digit arithmetic), on all 40 test cases.
  • All 513 float64 results inside the running error bound; the closest uses 86% of it.
  • Interpreter and NumPy backend return bit-identical results.
  • At float32 too: the 40 test cases rounded to float32, and 40 of them inside the running error bound at u = 2⁻²⁴; the closest uses 93% of it.
Accuracy in detail
correctly rounded (the float64 nearest the exact value)
69%
bit-equal to the NumPy formula in float64
89%
largest error, in units in the last place
377

Large ulp counts appear where a result is tiny next to the numbers it is computed from (after cancellation, for example), so one unit in the last place is tiny too; the absolute error is still inside the bound. Results within their own error of zero are not counted.

Identity

sha256:431082dfc67e29472c4dbac508698ff2758564edc7ec4aba775c329cba7e39ec

The semantic hash of the graph. It changes when the program changes, and never when only its documentation does.

Control handle

Symbol
Ω:std.optim.adam_step · Ω:adam_step
Pins
sha256:7e6924652bd08f16a88d2e2763dfc3ba155c430961c561da553c88996ec6e618this graph alone
Evidence
sha256:95444dc8a04a01a5b488b44d9f80f0f3083d11f42d3501051c2e1165b8e2a719the hash of its verification record
Needs
no capability: a pure function

Through NOVA’s control layer, the symbol launches this function only while the program still matches what it pins: a change to this graph, or to any graph it reaches, needs a migration first.

ΣEML 2026

Efficient New Language — a high-density semantic overlay for humans and AI agents.

This site’s demos run on the real EML toolchain, in your browser.

© 2026 EveMissLab (一言諾科技有限公司) / Neo.K. All rights reserved.