INDEX

Explanations

mentions of the term "jihadi" and related words or mentions of locations associated with the term

oai_token-act-pair · gpt-3.5-turbo

references to extremist groups and ideologies

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

 disinfect

-0.70

 correcting

-0.64

 present

-0.62

bl

-0.62

 duck

-0.62

 woods

-0.62

 bind

-0.61

 Appalach

-0.61

 rods

-0.59

 appra

-0.59

POSITIVE LOGITS

ihad

5.12

ihadi

1.94

azeera

1.17

imet

1.05

soever

1.04

ourney

0.99

iannopoulos

0.97

iott

0.96

abet

0.95

igi

0.94

Activations Density 0.010%

mentions of the term "jihadi" and related words or mentions of locations associated with the term

references to extremist groups and ideologies

No Comments

No Known Activations