INDEX

Explanations

words related to home or domestic settings

oai_token-act-pair · gpt-3.5-turbo

references to "hom" related terms, particularly those associated with "homosexual" topics

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 9-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.9.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.9.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

 sender

-0.71

 canopy

-0.70

 shroud

-0.67

 indirect

-0.66

 bananas

-0.64

UGE

-0.64

 negatives

-0.62

 hint

-0.59

 spiral

-0.59

 confir

-0.59

POSITIVE LOGITS

estead

1.65

emade

1.41

ogeneous

1.32

ework

1.30

ogenous

1.29

ogeneity

1.27

osexual

1.27

icide

1.26

eless

1.16

icidal

1.14

Activations Density 0.025%

words related to home or domestic settings

references to "hom" related terms, particularly those associated with "homosexual" topics

No Comments

No Known Activations