INDEX

Explanations

references to ancient religious practices, rituals, and beliefs

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

New Auto-Interp

Configuration

jbloom/Gemma-2b-IT-Residual-Stream-SAEs/gemma_2b_it_blocks.12.hook_resid_post_16384

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

HuggingFaceFW/fineweb

Features

16,384

Data Type

float32

Hook Name

blocks.12.hook_resid_post

Hook Layer

Architecture

standard

Context Size

1,024

Dataset

Skylion007/openwebtext

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 akut

-0.79

 notor

-0.78

 demokra

-0.76

 panik

-0.75

 alkoh

-0.73

 mikrofon

-0.73

 kriminal

-0.69

 praktik

-0.69

 foton

-0.68

 ekos

-0.68

POSITIVE LOGITS

 disagre

1.30

 unspeak

1.27

 shewn

1.24

 withal

1.22

 encomp

1.22

 unwarran

1.21

 indescri

1.20

 tolerably

1.20

 apprehen

1.20

 gaily

1.18

Activations Density 0.105%

references to ancient religious practices, rituals, and beliefs

No Comments

No Known Activations