INDEX

Explanations

references to religious or spiritual entities and places

oai_token-act-pair · gpt-3.5-turbo

references to the term "Holy."

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.hook_resid_pre

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

Data Type

torch.float32

Hook Point

blocks.4.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

PsyNetMessage

-0.95

atown

-0.91

phrine

-0.85

-0.79

ĸļ

-0.78

ļéĨĴ

-0.78

ickr

-0.77

xual

-0.77

apses

-0.74

yrinth

-0.74

POSITIVE LOGITS

 Grail

0.94

 Scripture

0.81

 Trinity

0.77

father

0.74

 Father

0.73

 canonical

0.72

 canon

0.71

field

0.71

 communion

0.70

church

0.69

Activations Density 0.012%

references to religious or spiritual entities and places

references to the term "Holy."

No Comments

No Known Activations