INDEX

Explanations

references to warnings or disclaimers about sensitive content

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

domada

-0.63

 Reſ

-0.63

 Conſ

-0.61

 Houſe

-0.61

ſelves

-0.60

DoubleQuotes

-0.59

 kasarigan

-0.58

 faſt

-0.56

ſelf

-0.55

 OFDb

-0.55

POSITIVE LOGITS

Generally

0.38

 verder

0.35

 here

0.35

 geluk

0.35

 generally

0.35

ndose

0.34

äß

0.33

 Hinsicht

0.33

 Generally

0.33

 there

0.32

Activations Density 0.049%

references to warnings or disclaimers about sensitive content

No Comments

No Known Activations