INDEX

Explanations

words related to feelings of embarrassment or guilt

oai_token-act-pair · gpt-3.5-turbo

instances and discussions of shame

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

ancial

-0.84

enhagen

-0.75

aeda

-0.73

ersive

-0.72

ichick

-0.71

ITNESS

-0.71

natureconservancy

-0.71

thus

-0.71

intendo

-0.70

opez

-0.70

POSITIVE LOGITS

faced

1.33

fully

1.27

 Shame

1.04

 shame

1.00

face

0.90

ful

0.88

ously

0.87

 ashamed

0.85

imaru

0.85

fulness

0.82

Activations Density 0.012%

words related to feelings of embarrassment or guilt

instances and discussions of shame

No Comments

No Known Activations