INDEX

Explanations

words related to extreme or controversial opinions and stances

oai_token-act-pair · gpt-3.5-turbo Triggered by @bot

references to irrational behavior or extreme views

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 12-res-jb

Configuration

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

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_post

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

 numbered

-0.71

 Hath

-0.69

 viol

-0.65

 Panther

-0.62

 backlog

-0.62

 whistleblower

-0.61

 layoffs

-0.61

 Woodward

-0.60

CLASSIFIED

-0.60

 reluctance

-0.60

POSITIVE LOGITS

acy

2.36

acies

2.12

ates

1.73

acists

1.71

ately

1.66

atic

1.62

atics

1.60

atism

1.50

acist

1.49

ACY

1.39

Activations Density 0.081%

words related to extreme or controversial opinions and stances

references to irrational behavior or extreme views

No Comments

No Known Activations