INDEX

Explanations

email addresses

oai_token-act-pair · gpt-3.5-turbo

mentions of social media handles and email addresses

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

 Elephant

-0.76

Sod

-0.66

 embold

-0.66

 retali

-0.66

cru

-0.65

 Pumpkin

-0.65

 catapult

-0.65

 subjug

-0.65

 coerc

-0.64

 rehabilit

-0.63

POSITIVE LOGITS

gmail

1.32

#$

1.18

yahoo

1.15

gs

0.97

mic

0.94

lists

0.90

tm

0.89

gt

0.89

debian

0.89

home

0.87

Activations Density 0.015%

email addresses

mentions of social media handles and email addresses

No Comments

No Known Activations