INDEX

Explanations

references to Roman themes or figures

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-2B @ 20-gemmascope-res-16k

Configuration

google/gemma-scope-2b-pt-res/layer_20/width_16k/average_l0_71

Prompts (Dashboard)

36,864 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

16,384

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

 Monfieur

-0.84

 propOrder

-0.82

 itſelf

-0.80

 myſelf

-0.79

Datuak

-0.76

 reaſon

-0.74

 ModelExpression

-0.72

 Jefus

-0.72

setPreferred

-0.71

 becauſe

-0.70

POSITIVE LOGITS

Rom

0.68

 Roman

0.64

Rom

0.59

rom

0.59

Roman

0.58

ROM

0.57

 Numer

0.55

 Abra

0.55

 ROMAN

0.54

 Reigns

0.53

Activations Density 0.075%

references to Roman themes or figures

No Comments

No Known Activations