INDEX

Explanations

mentions of individuals being awarded or recognized with a title or position

oai_token-act-pair · gpt-3.5-turbo

instances of the word "named" in various contexts

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

Beg

-0.83

EMS

-0.81

Cu

-0.66

jj

-0.66

romy

-0.66

Storm

-0.65

Yo

-0.65

Const

-0.65

Anim

-0.64

cession

-0.63

POSITIVE LOGITS

paces

1.08

 naming

1.00

 named

0.96

 names

0.91

named

0.89

 aliases

0.88

 surn

0.85

comings

0.77

essee

0.71

gling

0.71

Activations Density 0.016%

mentions of individuals being awarded or recognized with a title or position

instances of the word "named" in various contexts

No Comments

No Known Activations