INDEX

Explanations

expressions conveying strong negative emotions, especially regarding regret or disappointment

oai_token-act-pair · gpt-3.5-turbo

expressions of shame

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

ancial

-0.86

enhagen

-0.71

thus

-0.70

ITNESS

-0.69

ichick

-0.69

aeda

-0.68

 accur

-0.67

ernand

-0.66

ullivan

-0.66

ersive

-0.66

POSITIVE LOGITS

fully

1.31

faced

1.23

 shame

1.08

 Shame

1.06

imaru

0.95

face

0.90

fulness

0.88

ously

0.86

 ashamed

0.84

ful

0.81

Activations Density 0.010%

expressions conveying strong negative emotions, especially regarding regret or disappointment

expressions of shame

No Comments

No Known Activations