INDEX

Explanations

mentions of elevators or elevator shafts

oai_token-act-pair · gpt-3.5-turbo

references to elevators

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

fulness

-0.83

nesses

-0.81

lust

-0.80

ned

-0.75

ness

-0.74

arah

-0.72

words

-0.71

nar

-0.70

nia

-0.69

wcsstore

-0.68

POSITIVE LOGITS

 shaft

1.10

inson

0.94

 elevator

0.93

 doors

0.84

 elev

0.83

 entrances

0.83

 pitch

0.82

 stairs

0.78

ysis

0.77

idge

0.76

Activations Density 0.025%

mentions of elevators or elevator shafts

references to elevators

No Comments

No Known Activations