INDEX

Explanations

mentions of hidden or secret things

oai_token-act-pair · gpt-3.5-turbo

references to cameo appearances in media

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

Dispatch

-0.74

velt

-0.70

 Depression

-0.69

Express

-0.68

zos

-0.67

 depression

-0.64

 "$:/

-0.63

 backlog

-0.63

tru

-0.63

 addict

-0.62

POSITIVE LOGITS

 cameo

2.36

 undisclosed

1.45

cellaneous

1.20

 plum

1.09

cheon

1.09

reenshots

1.08

 shout

1.07

 ceremonial

0.99

 nominal

0.97

 apiece

0.94

Activations Density 0.066%

mentions of hidden or secret things

references to cameo appearances in media

No Comments

No Known Activations