INDEX

Explanations

mentions of individuals associated with terrorism or political conflict

oai_token-act-pair · gpt-3.5-turbo

references to specific individuals associated with jihadist groups

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

ãĥķãĤ©

-0.74

 Townsend

-0.71

 Gemini

-0.71

xual

-0.69

 enclosed

-0.69

cision

-0.67

 Kubrick

-0.66

 MÃ¼

-0.64

 Muller

-0.64

 Hitler

-0.64

POSITIVE LOGITS

akening

1.30

ashington

1.05

aii

1.04

yers

1.02

akens

1.01

abit

1.01

adesh

1.00

kward

0.97

Aw

0.96

oln

0.94

Activations Density 0.006%

mentions of individuals associated with terrorism or political conflict

references to specific individuals associated with jihadist groups

No Comments

No Known Activations