INDEX

Explanations

a specific pattern of repeated letters and codes, likely relevant to structured data or sequences

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

Hexadecimal color codes

np_acts-logits-general · gemini-2.0-flash

hexadecimal codes and indices

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-2B @ 18-gemmascope-res-16k

Configuration

google/gemma-scope-2b-pt-res/layer_18/width_16k/average_l0_74

Prompts (Dashboard)

36,864 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

16,384

Data Type

float32

Hook Name

blocks.18.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

 autorytatywna

-0.82

,:);

-0.77

Autoritní

-0.77

,:),

-0.76

SourceChecksum

-0.75

RenderAtEndOf

-0.75

__':

-0.75

']):

-0.74

 >=",

-0.70

}}_{\

-0.69

POSITIVE LOGITS

 femininas

0.59

 novamente

0.55

 again

0.55

 himſelf

0.54

 Himself

0.52

 opět

0.51

 themſelves

0.51

 nocturna

0.50

 marquées

0.50

warnai

0.49

Activations Density 0.621%

a specific pattern of repeated letters and codes, likely relevant to structured data or sequences

Hexadecimal color codes

hexadecimal codes and indices

No Comments

No Known Activations