INDEX

Explanations

references to Warner company, possibly Warner Bros

oai_token-act-pair · gpt-3.5-turbo

mentions of the company Warner Bros

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

ebin

-0.79

idences

-0.79

oral

-0.78

xual

-0.74

NetMessage

-0.72

omorphic

-0.72

bered

-0.71

ombo

-0.71

EStream

-0.70

olulu

-0.70

POSITIVE LOGITS

 Warner

1.24

 Bros

1.19

 Brothers

0.96

 Cable

0.72

Thor

0.72

Film

0.71

ewater

0.68

frey

0.67

Girls

0.66

 Tower

0.65

Activations Density 0.011%

references to Warner company, possibly Warner Bros

mentions of the company Warner Bros

No Comments

No Known Activations