INDEX

Explanations

references to the National Public Radio (NPR) broadcaster

oai_token-act-pair · gpt-3.5-turbo

references to NPR and PBS

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

aviour

-0.78

minster

-0.69

 bricks

-0.67

 colours

-0.65

wil

-0.65

 Cout

-0.64

 favour

-0.62

 manoeuv

-0.61

 tyre

-0.61

 defences

-0.60

POSITIVE LOGITS

NPR

3.95

NPR

3.22

PBS

1.72

 MSNBC

1.50

toggle

1.46

NBC

1.31

Enlarge

1.30

NYT

1.29

CBC

1.24

CNN

1.22

Activations Density 0.008%

references to the National Public Radio (NPR) broadcaster

references to NPR and PBS

No Comments

No Known Activations