ManySpace
English Giriş yap Kayıt ol

Yapay ZekâUzmanlıkAIML-INT-005

Mechanistic Interpretability: Circuits

Reverse-engineering the algorithms inside networks: features, circuits, and causal interventions such as activation patching.

Bu konunun henüz Türkçe çevirisi yok, bu yüzden İngilizce gösteriliyor.

Zorluk
5 / 5
Önem
3 / 5
Çalışma süresi
1–2 sa
Çalışma türü
Karma

Neden önemli

The learner can use activation patching to locate which component carries information for a behaviour and explain why patching is causal where probing is not.

Bu konuya giden rota

Rota AIML-INT-005
What Is a Model? Tasks: Regression, Classification, and Beyond Features and Targets Classification Framing Numeric and Categorical Data Linear Regression I: Prediction and Residuals Embeddings Scores and Logits Encoding Categories Loss and Fitting: Squared Error Similarity in Vector Space Sigmoid and Probabilities Multiclass Scores and Softmax The Artificial Neuron Distributed Representations Query, Key, Value Logistic Regression Activation Functions Word Embeddings Similarity Scores The Multilayer Perceptron Contextual Representations Weighted Aggregation Probing Representations Self-Attention The Attention Block The Causal Transformer Reading a Transformer Mechanistic Interpretability: Circuits