INDEX

Explanations

the word "drama" occurring in various contexts

oai_token-act-pair · gpt-3.5-turbo

references to drama, particularly in television and film 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

ovo

-0.87

anza

-0.77

achev

-0.75

ossier

-0.73

ees

-0.72

 redistributed

-0.71

aton

-0.71

itude

-0.71

laws

-0.71

enhagen

-0.70

POSITIVE LOGITS

 drama

0.94

 queens

0.93

 dramas

0.91

 unfold

0.89

 queen

0.79

 Drama

0.78

 involving

0.78

 revolving

0.78

 unfolds

0.77

 unfolding

0.74

Activations Density 0.018%

the word "drama" occurring in various contexts

references to drama, particularly in television and film contexts

No Comments

No Known Activations