INDEX

Explanations

references to screens or screen-related activities

oai_token-act-pair · gpt-3.5-turbo

references to "screen" or related concepts in media and technology

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

ETA

-0.79

 extrem

-0.74

 corpor

-0.70

 extreme

-0.62

IGH

-0.62

 rush

-0.62

FIN

-0.61

 frig

-0.60

 punishing

-0.60

 consistent

-0.60

POSITIVE LOGITS

screen

1.31

Screen

0.98

reens

0.98

 Screen

0.91

TVs

0.88

wip

0.85

 screenings

0.79

acters

0.78

iew

0.77

ographer

0.77

Activations Density 0.014%

references to screens or screen-related activities

references to "screen" or related concepts in media and technology

No Comments

No Known Activations