INDEX

Explanations

phrases indicating inclusion or exclusion of specific entities

oai_token-act-pair · gpt-3.5-turbo

phrases that express inclusion or emphasize exceptions

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

estyles

-0.78

ahime

-0.73

ulations

-0.68

ript

-0.67

oren

-0.66

format

-0.65

bis

-0.65

iken

-0.65

ntax

-0.64

brace

-0.64

POSITIVE LOGITS

 myself

1.38

 ourselves

1.34

 yourselves

1.27

 oneself

1.26

 yourself

1.22

 themselves

1.17

 politicians

1.16

 strangers

1.14

 bureaucrats

1.12

 outsiders

1.12

Activations Density 0.303%

phrases indicating inclusion or exclusion of specific entities

phrases that express inclusion or emphasize exceptions

No Comments

No Known Activations