INDEX

Explanations

phrases expressing strong emotions, particularly negative ones like shame, disgust, and anger

oai_token-act-pair · gpt-3.5-turbo

words related to condemnation or criticism

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 7-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.7.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.7.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

 lumber

-0.72

 brass

-0.69

 Tanz

-0.69

 complement

-0.68

 fuller

-0.66

 stocked

-0.64

 cart

-0.64

 stocking

-0.62

 vantage

-0.62

 complementary

-0.62

POSITIVE LOGITS

ful

1.27

fully

1.26

fulness

1.18

FUL

1.17

ingly

1.05

ously

1.00

ous

0.99

bringer

0.96

mong

0.94

lessly

0.93

Activations Density 0.223%

phrases expressing strong emotions, particularly negative ones like shame, disgust, and anger

words related to condemnation or criticism

No Comments

No Known Activations