INDEX

Explanations

celebratory phrases or expressions of enthusiasm

oai_token-act-pair · gpt-3.5-turbo

phrases involving mystical or legendary elements

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

ufact

-0.61

 mosqu

-0.59

ividual

-0.58

chnology

-0.55

 phosphate

-0.53

 neighb

-0.52

Downloadha

-0.51

 incent

-0.49

icken

-0.49

ierrez

-0.49

POSITIVE LOGITS

âĶĢâĶĢâĶĢâĶĢ

0.78

âĶĢâĶĢâĶĢâĶĢâĶĢâĶĢâĶĢâĶĢ

0.74

laughs

0.70

----------------------------------------------------------------

0.68

Laughs

0.62

İ

0.60

laughter

0.57

________________________________________________________________

0.56

QUI

0.56

âĺħ

0.55

Activations Density 0.666%

celebratory phrases or expressions of enthusiasm

phrases involving mystical or legendary elements

No Comments

No Known Activations