Up nextNext guide
Stochastic Gradient Descent with Momentum
Fundamentals
Fundamentals GUIDE
Gradient Descent
Plain gradient descent updates parameters by subtracting a learning rate times the objective's gradient.
On this page3 min read
Overview
It aims to reduce the objective locally, while step size and loss shape determine whether a particular update succeeds.
Deep Dive
For a differentiable scalar objective, the gradient identifies the direction of steepest local increase under the usual Euclidean norm. Plain gradient descent moves in the opposite direction. Its update is new weight equals old weight minus learning rate times gradient. In a scalar illustration with old weight two, gradient three, and learning rate one-tenth, the new weight is one-point-seven. That formula describes the proposed update, not a guarantee that every finite step reduces loss. A learning rate that is too large can overshoot or lead to divergence, and a very small rate can make progress slow. Full-batch gradient descent uses the selected training objective over the whole dataset. Stochastic and mini-batch methods instead use gradients from sampled examples or batches. The batch gradient is exact for that batch calculation but estimates the full-data gradient; the resulting variability can help or hurt depending on the problem. Momentum combines information from earlier gradients to influence the current update. Adam maintains per-parameter moment estimates that adapt updates, so it should not be confused with plain unmodified gradient descent. Nonconvex loss surfaces also contain stationary points. A zero gradient is not proof of a globally lowest objective: local minima, maxima, and saddle points can have zero gradient. Training therefore needs monitoring of the objective and relevant evaluation metrics rather than assuming that a named optimizer or a stationary point establishes success.
Strategic Impact
Clearer decisions
It helps you separate clear technical claims from marketing language.
Cost and budget
You can ask better implementation questions before spending money or time.
Team and workflow
Teams with shared understanding make better product, policy, and learning decisions.
The Future of Gradient Descent
Optimization research studies schedules, moment estimates, curvature, and distributed implementations under specific objectives and workloads. A method that performs well in one experiment may require different settings elsewhere. Report learning rates, sampling, state rules, and evaluation conditions when comparing training runs. Verify that updates change parameters as intended and track whether useful metrics improve; a direction motivated by local calculus is not a universal guarantee of convergence or application quality.
Real-World Implementation
Lowering a language model's prediction error across billions of training tokens using mini-batch updates
Tuning the learning rate so an image model converges quickly without the loss exploding
Using momentum to speed up training of a speech recognition network stuck in a long, narrow loss valley
Applying Adam to fine-tune a model on a small dataset where per-parameter learning rates help stability
Risks & Guardrails
Different teams may use the same term differently, so define scope early.
Benchmarks can look strong while real-world performance is uneven.
Ignoring data quality and evaluation plans often creates fragile outcomes.
Implementation Roadmap
Start with a plain-language definition of the outcome you need.
Pick one success metric and one failure condition before testing.
Run a small pilot with representative data, not a polished demo set.
Document where Gradient Descent helps and where simpler methods are better.
Keep Exploring
Free newsletter
Get the daily AI briefing
Three verified AI stories every weekday morning, written in plain English. Free forever, no ads.
One email each weekday. Unsubscribe in one click. We never sell or share your address.
Test yourself
Take the Gradient Descent quiz
Instant feedback on every answer, and a shareable certificate with a verifiable ID once you pass a course.
Support free AI education. AI Understanding is a 501(c)(3) nonprofit — no ads, no paywall, ever. Make a donation
Frequently asked questions
What is Gradient Descent?
Plain gradient descent updates parameters by subtracting a learning rate times the objective's gradient. It aims to reduce the objective locally, while step size and loss shape determine whether a particular update succeeds.
In plain gradient descent, which direction is used relative to the objective's gradient?
Plain gradient descent subtracts a scaled gradient, with improvement dependent on step size and objective shape.
What does the learning rate control?
The learning rate scales how far the weights move on each update; too large overshoots, too small makes training painfully slow.
How does a mini-batch gradient relate to the full selected training-data objective?
The computed batch derivative is exact for that batch, while sampling makes it an estimate of the full-data derivative.
What problem can a learning rate that is too large cause?
Large steps can jump past the minimum and bounce around or blow up, causing the loss to increase rather than settle.
What information does momentum carry into later gradient-descent updates?
Momentum combines gradient history, which can affect progress and oscillation; its effect still depends on settings.
Keep learning
Related guides
More guides picked for this topic