activation

The output value a neuron in a neural network produces in response to its input, representing the model's internal state.

7 articlesLast mentioned

An activation is the output value a neuron in a neural network produces by applying an activation function to the weighted sum of its inputs. The set of activations across a layer represents how that layer encodes the input and serves as the input to the next layer.

Activation functions such as ReLU, sigmoid, and GELU apply a nonlinear transformation that allows neural networks to approximate complex functions. In interpretability research, the magnitude and pattern of a neuron's activations on particular inputs are analyzed to infer which features the neuron detects.

This entry is based on AIPOST articles and widely known facts. If something is wrong, please send us a correction request.

Articles covering this entry

  • Capy AI Captains Run a Game Release Flow on 3% of a Weekly Claude Ultra LimitAI in Practice ·

    …large cloud instance, ran the project initializers and npm installs, executed build validation scripts, and captured an activation snapshot in about 11 seconds.

    Read article
  • Claude Code Outreach Agent Emails Trial Dropouts and Checks Its Own RepliesAI in Practice ·

    …points are onboarding drop-off (lowest reply rate, highest volume, reveals activation barriers), a mid-trial check-in, trial end without conversion, users who e…

    Read article
  • Pre-Training, Fine-Tuning and RL: Where Frontier LLMs Get Their QualityAI Performance ·

    Adding hidden layers and non-linear activation functions fixes this.

    Read article
  • Temperature, Attention and Context Rot: Model Mechanics Behind Agent DesignAI in Practice ·

    Intermediate activations are discarded, but keys and values are kept as the KV cache, which grows with every token.

    Read article
  • Neural Networks Are Already Clean: Study Claims 90% of Neuron Poles Are LegibleAI Security ·

    As networks grew, nobody could say what any individual activation meant, and the "black box" reputation set in.

    Read article
  • Sam Altman at Oracle AI World: GPT-6, AI-Written Code and a Custom Chip on OCIAI News ·

    …e between ideas without stalling on knowledge gaps or waiting, and tasks whose activation energy was once too high to justify starting are now cheap to explore.

    Read article
  • Attention Budgets and a Chief of Staff Agent: Four Rules Against AI BusyworkAI in Practice ·

    Lower activation energy for starting things means you can accumulate more ongoing obligations in a month than you can carry.

    Read article

Found a mistake or missing an entry?

Wiki entries are based on AIPOST articles and widely known facts. If an entry is wrong, send us a correction request and we will check and fix it. Names from new articles are added to the wiki after the article is published.