INDEX

Explanations

words related to attitudes and opinions

oai_token-act-pair · gpt-3.5-turbo

references to social attitudes and perceptions

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 2-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.2.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.2.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

ded

-0.79

gran

-0.74

amen

-0.73

avez

-0.71

addafi

-0.70

aman

-0.67

cuts

-0.66

Jub

-0.64

Delivery

-0.64

MER

-0.63

POSITIVE LOGITS

 attitudes

1.07

 guiActiveUn

0.89

pring

0.86

ocial

0.84

terness

0.82

 toward

0.79

insula

0.76

 towards

0.76

hovah

0.76

yip

0.75

Activations Density 0.013%

words related to attitudes and opinions

references to social attitudes and perceptions

No Comments

No Known Activations