INDEX

Explanations

specific Chinese words and phrases indicating a form or identifier in a text

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

obtained described mentioned relevant

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-2B @ 16-gemmascope-res-65k

Configuration

google/gemma-scope-2b-pt-res/layer_16/width_65k/average_l0_128

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

65,536

Data Type

float32

Hook Name

blocks.16.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

of

-0.48

 mukana

-0.47

 sklo

-0.45

-0.43

 while

-0.42

 before

-0.41

then

-0.40

名を

-0.40

 nodig

-0.40

 NgModule

-0.40

POSITIVE LOGITS

 otomatig

0.93

 tartalomajánló

0.91

LookAnd

0.85

UVWXYZ

0.82

 ſeveral

0.78

ſelf

0.77

ſelves

0.75

Hentet

0.74

'}>

0.73

indeki

0.73

Activations Density 0.010%

specific Chinese words and phrases indicating a form or identifier in a text

obtained described mentioned relevant

No Comments

No Known Activations