INDEX

Explanations

phrases related to political figures and government offices

oai_token-act-pair · gpt-3.5-turbo

references to the White House

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

anwhile

-0.85

ngth

-0.85

raints

-0.80

yrinth

-0.79

olls

-0.79

odcast

-0.78

ITAL

-0.78

trak

-0.77

rawdownloadcloneembedreportprint

-0.76

Downloadha

-0.73

POSITIVE LOGITS

house

1.05

hall

1.02

Sox

0.99

hurst

0.96

caps

0.96

berry

0.95

 supremacist

0.94

 House

0.92

 supremacists

0.87

ewater

0.86

Activations Density 0.021%

phrases related to political figures and government offices

references to the White House

No Comments

No Known Activations