INDEX

Explanations

the word 'Jihad' at different levels of activation

oai_token-act-pair · gpt-3.5-turbo

references to jihad

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 2-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.2.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.2.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

 births

-0.72

å£«

-0.71

ãģ¦

-0.67

 Presbyter

-0.67

 tumors

-0.66

çĦ

-0.66

Cth

-0.65

 anomalies

-0.65

ACA

-0.65

Examples

-0.64

POSITIVE LOGITS

ihad

1.47

edIn

0.95

ipedia

0.94

 Airways

0.94

idis

0.93

istani

0.93

assadors

0.92

renheit

0.88

aganda

0.85

uates

0.84

Activations Density 0.007%

the word 'Jihad' at different levels of activation

references to jihad

No Comments

No Known Activations