INDEX

Explanations

terms related to demagoguery and agitation

oai_token-act-pair · gpt-3.5-turbo

terms related to manipulation and potential deceit in media coverage

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

staking

-0.83

backer

-0.78

 tracing

-0.74

ball

-0.69

 Toll

-0.67

rem

-0.67

pora

-0.66

cos

-0.66

 modelling

-0.64

 traced

-0.64

POSITIVE LOGITS

glers

1.38

agog

1.16

agogue

1.10

ically

1.09

atory

1.05

acles

1.00

uments

0.98

ues

0.96

ibles

0.91

acle

0.90

Activations Density 0.016%

terms related to demagoguery and agitation

terms related to manipulation and potential deceit in media coverage

No Comments

No Known Activations