← All parts of this equation

Equation 6 · Part 10 · How Constitutional AI Actually Constrains a Model's Behavior

subtraction

LPM(θ)=− E(x, yw, yl) ∼ DH∪DAI[log⁡σ(rθ(x,yw)−rθ(x,yl))].\mathcal{L}_{PM}(\theta) = -\,\mathbb{E}_{(x,\,y_w,\,y_l)\,\sim\, D_H \cup D_{AI}}\Big[\log \sigma\big(r_\theta(x,y_w) - r_\theta(x,y_l)\big)\Big].
subtraction

What this part means

Subtract the following term or group from the preceding one. A leading minus marks a negative quantity.

Its job in the formula

Subtract the following term or group from the preceding one. A leading minus marks a negative quantity.

The passage around this formula

Crucially, this AI-generated harmlessness comparison data did not replace human comparison data outright; it was combined with it. The paper reports 135,296 human helpfulness comparisons and 182,831 constitutionally generated harmlessness comparisons feeding one preference model, trained on the union of both, in the authors’ words: “we use human labels for helpfulness, but only AI labels for harmlessness.” A single Bradley-Terry-style preference loss is fit across both sources at once. Writing DHD_H for the human comparison set, DAID_{AI} for the AI-generated set, rθr_\theta for the reward model, x for a prompt, and ywy_w, yly_l for the winning and losing response in a pair, the objective is [displayed…

Read this part in the article →

Learn the underlying idea

Addition combines quantities; subtraction measures the signed difference between them. Parentheses show what is combined before the rest of the expression is evaluated.

Open the illustrated addition and subtraction in an equation guide →

Sources cited in the article section

These citations provide research context; check each source for the exact claim it supports.