INDEX

Explanations

names related to religious and political leaders, specifically Ayatollahs

oai_token-act-pair · gpt-3.5-turbo

references to the Iranian Supreme Leader and related political themes

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

nexus

-0.79

bands

-0.78

kt

-0.74

photos

-0.71

idious

-0.71

insula

-0.70

Ec

-0.68

wire

-0.67

hound

-0.67

fox

-0.66

POSITIVE LOGITS

eenth

0.93

sburgh

0.92

ain

0.87

rition

0.81

iful

0.80

ï¿½ï¿½ï¿½ï¿½

0.78

ifully

0.78

abulary

0.75

liga

0.75

tesy

0.73

Activations Density 0.048%

names related to religious and political leaders, specifically Ayatollahs

references to the Iranian Supreme Leader and related political themes

No Comments

No Known Activations