INDEX

Explanations

proper nouns and specific names in different contexts

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

zetten

-0.33

alı

-0.32

Fun

-0.32

mic

-0.31

금

-0.31

 Maier

-0.31

ura

-0.31

con

-0.30

คา

-0.30

 Rezepte

-0.30

POSITIVE LOGITS

 ſtate

0.66

 greateſt

0.65

 pleaſure

0.64

 indígen

0.64

MessageTagHelper

0.62

 autorytatywna

0.60

MigrationBuilder

0.60

AccessorTable

0.59

 poffe

0.58

 leaſt

0.58

Activations Density 0.346%

proper nouns and specific names in different contexts

No Comments

No Known Activations