INDEX

Explanations

phrases indicating an author's name and affiliation

oai_token-act-pair · gpt-3.5-turbo

references to news articles and authorship

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

 calendars

-0.73

 cabinets

-0.68

 queues

-0.63

 fridge

-0.61

 refunds

-0.61

aults

-0.61

 silence

-0.60

 purs

-0.60

 retaliate

-0.60

uments

-0.60

POSITIVE LOGITS

Edited

0.93

Narr

0.86

Editor

0.80

BBC

0.79

EVA

0.78

roit

0.78

WASHINGTON

0.78

cerpt

0.75

SAN

0.74

Ù

0.74

Activations Density 0.106%

phrases indicating an author's name and affiliation

references to news articles and authorship

No Comments

No Known Activations