INDEX

Explanations

concepts or ideas expressed as "notions."

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Configuration

google/gemma-scope-9b-pt-res/layer_20/width_131k/average_l0_114

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

131,072

Data Type

float32

Hook Name

blocks.20.hook_resid_post

Hook Layer

Architecture

jumprelu

Context Size

1,024

Dataset

monology/pile-uncopyrighted

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

AutoScale

-0.46

registrer

-0.46

Environmental

-0.44

Electrical

-0.43

zarchiwizowane

-0.43

Mirror

-0.42

Busy

-0.42

Statistical

-0.42

 Erwin

-0.42

POSITIVE LOGITS

 Notion

1.13

 notion

0.97

 notions

0.83

SharedCtor

0.66

 CONCEPT

0.64

 concept

0.60

 conception

0.59

conception

0.59

 MotionEvent

0.58

 Vorstellung

0.58

Activations Density 0.007%

concepts or ideas expressed as "notions."

No Comments

No Known Activations