INDEX

Explanations

high-profile names, likely associated with news or media

oai_token-act-pair · gpt-3.5-turbo

references to specific individuals, particularly those in media or controversial contexts

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

requires

-0.76

IRE

-0.75

aic

-0.71

 Uriel

-0.68

 Archdemon

-0.66

committee

-0.65

 captcha

-0.64

ension

-0.62

 Warrant

-0.62

JV

-0.61

POSITIVE LOGITS

gyn

1.14

zona

0.92

 Kelly

0.84

amera

0.84

iliate

0.83

uala

0.79

witz

0.75

oche

0.74

 Huckabee

0.74

ately

0.74

Activations Density 0.022%

high-profile names, likely associated with news or media

references to specific individuals, particularly those in media or controversial contexts

No Comments

No Known Activations