INDEX

Explanations

words related to persuasion or manipulation

oai_token-act-pair · gpt-3.5-turbo

actions and tactics related to manipulation and coercion

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

chwitz

-0.77

blance

-0.75

ynski

-0.72

elt

-0.68

Ü

-0.68

arna

-0.65

Found

-0.64

gal

-0.64

bard

-0.63

ophen

-0.61

POSITIVE LOGITS

ingly

0.91

 gull

0.87

 unsuspecting

0.85

 electors

0.83

 INTO

0.82

 into

0.79

 unwitting

0.79

gery

0.76

 voters

0.76

eer

0.71

Activations Density 0.159%

words related to persuasion or manipulation

actions and tactics related to manipulation and coercion

No Comments

No Known Activations