INDEX

Explanations

phrases related to media attention and manipulation

oai_token-act-pair · gpt-3.5-turbo

instances of manipulation or exploitation in political contexts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

ãĤ´ãĥ³

-0.70

 Compass

-0.67

dayName

-0.66

 Explore

-0.65

ķ

-0.65

SourceFile

-0.65

fleet

-0.65

 Neurolog

-0.64

riad

-0.63

Sphere

-0.63

POSITIVE LOGITS

 publicity

1.34

 sympathy

1.24

 attention

1.09

 discredit

1.06

 clicks

1.06

 ridicule

1.04

 viewers

1.03

 votes

1.01

 gull

1.01

 unsuspecting

0.99

Activations Density 0.548%

phrases related to media attention and manipulation

instances of manipulation or exploitation in political contexts

No Comments

No Known Activations