INDEX

Explanations

references to administrative positions or roles

oai_token-act-pair · gpt-3.5-turbo

occurrences of the word "administrator."

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 2-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.2.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.2.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

çīĪ

-0.78

roads

-0.73

{"

-0.73

PO

-0.72

eele

-0.71

False

-0.71

True

-0.70

DonaldTrump

-0.69

tra

-0.69

doors

-0.66

POSITIVE LOGITS

 administrator

1.12

rators

0.91

 admin

0.89

 administ

0.88

intendent

0.85

stration

0.84

 administrators

0.81

uthor

0.78

ially

0.76

iation

0.76

Activations Density 0.012%

references to administrative positions or roles

occurrences of the word "administrator."

No Comments

No Known Activations