INDEX

Explanations

mentions of spending time or making calculations

oai_token-act-pair · gpt-3.5-turbo

quantitative or numerical information

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

 Mankind

-0.51

 raping

-0.49

 bathing

-0.48

 virginity

-0.47

 Temper

-0.47

 ãĤµãĥ¼ãĥĨãĤ£ãĥ¯ãĥ³

-0.46

 homosexuality

-0.45

 highs

-0.45

 foreigners

-0.45

 husbands

-0.44

POSITIVE LOGITS

ascript

0.66

endum

0.54

endix

0.48

arching

0.48

EEK

0.46

disciplinary

0.46

 methodological

0.45

emonium

0.44

rences

0.43

ciplinary

0.43

Activations Density 2.720%

mentions of spending time or making calculations

quantitative or numerical information

No Comments

No Known Activations