INDEX

Explanations

references to apartments or apartment complexes

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

>>;

-0.51

causes

-0.50

lead

-0.49

 Głów

-0.48

Causes

-0.48

Lead

-0.47

ensure

-0.46

 gdyż

-0.45

getCause

-0.45

saw

-0.45

POSITIVE LOGITS

 apartment

1.06

 Apartment

1.05

 apartments

1.04

Apartment

0.97

 Apartments

0.96

artments

0.86

apartment

0.84

Apartments

0.77

 apartamento

0.75

artment

0.74

Activations Density 0.008%

references to apartments or apartment complexes

No Comments

No Known Activations