INDEX

Explanations

phrases related to subscribing to newsletters or reading stories online

oai_token-act-pair · gpt-3.5-turbo

mentions of newsletters

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

 unintended

-0.72

DC

-0.69

iera

-0.68

OTT

-0.66

WC

-0.66

less

-0.65

Cost

-0.65

ORN

-0.64

Chain

-0.63

same

-0.61

POSITIVE LOGITS

Newsletter

1.23

gdala

0.95

ertodd

0.93

 Sign

0.89

 Emails

0.88

 subscriptions

0.87

ptoms

0.86

ometimes

0.82

 Newsletter

0.81

agonists

0.81

Activations Density 0.009%

phrases related to subscribing to newsletters or reading stories online

mentions of newsletters

No Comments

No Known Activations