INDEX

Explanations

words related to degradation or decline

oai_token-act-pair · gpt-3.5-turbo

references to political scrutiny and accountability

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 unlaw

-0.78

Sty

-0.75

usa

-0.74

andro

-0.73

irez

-0.72

 hous

-0.71

ster

-0.71

ju

-0.71

wik

-0.68

mar

-0.65

POSITIVE LOGITS

ithering

2.25

avering

2.06

akable

1.41

onew

1.27

merga

1.23

graduate

1.22

 admissions

1.20

arranted

1.13

romising

1.09

 tortured

1.08

Activations Density 0.063%

words related to degradation or decline

references to political scrutiny and accountability

No Comments

No Known Activations