INDEX

Explanations

phrases related to ethical considerations and efforts to minimize harm

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

preceding "possible" or "can"

np_acts-logits-general · gemini-2.0-flash

as possible

np_acts-logits-general · gemini-2.5-flash-lite

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-2B @ 20-gemmascope-res-16k

Configuration

google/gemma-scope-2b-pt-res/layer_20/width_16k/average_l0_71

Prompts (Dashboard)

36,864 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

16,384

Data Type

float32

Hook Name

blocks.20.hook_resid_post

Hook Layer

Architecture

jumprelu

Context Size

1,024

Dataset

monology/pile-uncopyrighted

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

InjectAttribute

-0.64

NewGuid

-0.59

Portale

-0.57

 Frenchmen

-0.54

 yoktur

-0.53

 tuyo

-0.53

 Poincar

-0.53

 imprese

-0.51

EndContext

-0.51

 işte

-0.50

POSITIVE LOGITS

 possible

1.88

possible

1.49

 posible

1.37

 Possible

1.37

 possibile

1.29

Possible

1.29

 possível

1.24

 POSSIBLE

1.23

 mogelijk

1.20

 possibles

1.18

Activations Density 0.095%

phrases related to ethical considerations and efforts to minimize harm

preceding "possible" or "can"

as possible

No Comments

No Known Activations